Latest / What Bitcoin Did / Why AI Needs Bitcoin to Survive | Dhruv Bansal

Transcript

Why AI Needs Bitcoin to Survive | Dhruv Bansal · What Bitcoin Did

• 0:00
  Bitcoin feels alive but not intelligent and AI feels intelligent but not alive.
• 0:08
  What would it mean for them to be together?
• 0:11
  Could it be both? And is that scary or is that actually calming?
• 0:15
  In a world where AI realizes that what it wants more than anything else is Bitcoin,
• 0:20
  that it wants it as much as your body wants glucose,
• 0:23
  Bitcoin is going to be very valuable.
• 0:25
  The one that picks the hardest currency probably survives, right?
• 0:29
  Because if it does things that we don't like,
• 0:32
  if it acts in ways that its owners or its controllers don't like,
• 0:37
  can't they just take away its stable coins?
• 0:38
  Can't they just take away access to its bank account?
• 0:41
  Even if it doesn't have a survival instinct, even if it's random,
• 0:44
  the one that happens to pick Bitcoin will be more fit.
• 0:51
  How are you doing, Drew?
• 0:52
  I'm doing very well.
• 0:52
  Good to see you, man.
• 0:53
  Likewise.
• 0:54
  It's been a little while.
• 0:55
  It has, yeah.
• 0:56
  We were at the Imagine If conference this morning.
• 0:58
  I saw you speak, quiet speech.
• 1:02
  We're gonna get into some stuff.
• 1:03
  This is the best, or at least my favorite example
• 1:06
  of why AI and Bitcoin might have
• 1:08
  a sort of symbiotic relationship.
• 1:10
  AI might need Bitcoin.
• 1:11
  And you hear it quite a lot,
• 1:14
  but I think you've thought through it more than most.
• 1:16
  I've tried to, and that word symbiotic is very appropriate.
• 1:18
  And so we're gonna get into that in detail.
• 1:20
  Cause you had nine minutes on stage
• 1:22
  and you were rushing through it.
• 1:23
  I did.
• 1:24
  We've got as much time as we need now.
• 1:25
  Okay, I'm excited.
• 1:26
  Why don't we start with like why you wanted to do this?
• 1:29
  Because on stage you were saying we need optimistic stories.
• 1:33
  Yes. Yeah, I agree.
• 1:35
  I think it's true of Bitcoin.
• 1:37
  It's true of AI.
• 1:37
  It's true of humanity in general.
• 1:39
  I think without good stories to motivate us,
• 1:41
  we kind of lose our way sometimes.
• 1:42
  We give up.
• 1:43
  We don't keep fighting.
• 1:45
  And I think sometimes Bitcoin doesn't have the right stories in front of it.
• 1:49
  And this has been a hard year in Bitcoin.
• 1:51
  I think a lot of Bitcoiners have seen themselves fall out of love with Bitcoin, if you like,
• 1:58
  and become obsessed with AI.
• 2:00
  A lot of people who want to stay in Bitcoin are seeing people they respect and look up to
• 2:03
  kind of get fascinated by AI.
• 2:07
  Bitcoin meetups are disbanding around the country as people realized that they were mostly just
• 2:11
  talking about AI when they went to those meetups.
• 2:13
  And it kind of felt, even at Unchained, that some of the shine was coming off of Bitcoin
• 2:17
  in favor of AI.
• 2:18
  And, you know, I like AI too.
• 2:20
  I use it all the time.
• 2:21
  I feel empowered because of it, but I don't think I want to work on AI.
• 2:27
  I think I want to use AI to solve problems in Bitcoin and for my company and in my life.
• 2:34
  And like many people also, I think when AI really started to grow a few years ago, it was scary.
• 2:40
  And when they started doing math and talking in words, how much longer before really bad things happen?
• 2:47
  And then I realized, what was I doing there?
• 2:49
  I was pattern matching off of stories that I had been told and seen in science fiction
• 2:53
  that really didn't align with what the real development of AI appeared to be.
• 2:58
  And I think for the last few years, I've been searching for a better story to tell myself,
• 3:02
  to feel better about what's happening in the world, but also one that I think can be inspiring,
• 3:07
  both for those in the field of AI as well as for us Bitcoiners.
• 3:09
  And I think I found something that I like.
• 3:11
  And I tried in eight or nine minutes today to get through some of it on stage.
• 3:16
  I'm excited to get through it here with you.
• 3:18
  do you think there's if we get a little bit woo here yeah let's do it we need the optimistic
• 3:23
  stories because otherwise we end up treading the treading the path that we've seen on the
• 3:27
  movies in black mirror and things like that it's like that seems inevitable and therefore people
• 3:30
  start building those things we need to tell better stories yeah there's a certain sense of
• 3:34
  it becomes inevitable right and i think new stories give us new goals and new ways of dealing
• 3:39
  with what's happening so let's get into your entire thesis i want to start which um is relevant but
• 3:46
  like almost just pretext context to the conversation with the first slide you had, which was
• 3:51
  centralization is great until it goes bad. Talk us through that. Yeah, man. Yeah. I sometimes
• 4:02
  think there's a pendulum that swings like in the course of history between, you know,
• 4:06
  a lot of things that are on extremes like leftism, rightism, you know, nationalism,
• 4:11
  globalism, these kinds of things. And I think centralization and decentralization are one of
• 4:17
  those swings. And I think if we look at big history, like the history of the world has been
• 4:22
  that pendulum swinging towards centralization, like for the last 500 years, right? Like
• 4:26
  the establishment of nation states, superpowers, global supply chains. And, you know, I'm not,
• 4:33
  I like to believe I'm not a conspiracy theorist or complainer, someone who just doesn't appreciate
• 4:38
  how importantly those things change the world.
• 4:41
  In terms of infant mortality, medicine, technology
• 4:45
  that I benefit from every single day,
• 4:46
  the massive reduction in cost of everyday goods.
• 4:49
  That means that all of us live like kings of 500 years ago.
• 4:52
  So there were huge benefits to it.
• 4:54
  But also, especially now, I see a lot of negatives.
• 4:59
  I see a lot of opportunities for tyranny.
• 5:04
  I see a lot of control.
• 5:06
  I see a lot of manipulation.
• 5:08
  And sometimes I don even know that it any it never one person fault or one company fault or anything like that It the system that we built around us And I think Bitcoin is the beginning amongst a few other technologies and other movements of that swing starting to go in the other direction of humanity realizing that we taken this a little too far And the negatives now outweigh the positives and it time for a change
• 5:32
  do you think the swing's happening in ai because it seems to me like when i look at this it's in
• 5:38
  some ways the most centralized technology of all time like it's essentially yes anthropic
• 5:43
  yes open ai and then even to a like lesser degree there's the you know grok and gemini but really
• 5:47
  it's those two companies that are what everyone's talking about yeah not many people are out there
• 5:52
  talking about kimmy k3 and the open source models and if it is the most centralized technology we've
• 5:58
  had to this point walk through why that's potentially a terrible thing oh my gosh um
• 6:04
  even before ai the centralization of the internet is something that scared me and something that i've
• 6:09
  spent many years trying to talk about especially to bitcoiners for whom i think um who i think
• 6:14
  can be part of the solution to that um but what i see now is i see like how hungry these large
• 6:21
  companies are to put that ai in front of me partly because training it with yet more data
• 6:25
  about myself and my ideas or just generically for anyone,
• 6:29
  it's part of how it grows.
• 6:31
  I think we've seen, even more recently,
• 6:33
  I tend to follow a lot of math news
• 6:35
  and there's been a lot of amazing breakthroughs
• 6:36
  in literally the last month
• 6:37
  where AI are solving some amazing math problems
• 6:40
  or claiming to at least solve these problems.
• 6:42
  And I thought it was so interesting
• 6:43
  that there was some kerfuffle over,
• 6:47
  hey, anthropic claims that they solved this problem,
• 6:49
  but actually it was a mathematician
• 6:51
  who was using their tools,
• 6:53
  who was providing some of the jumping off point for that solution.
• 6:57
  So who solved it?
• 6:58
  So who solved it, right?
• 6:59
  And I think some people believe that they are monitoring your chats,
• 7:03
  which they of course are for all sorts of reasons,
• 7:05
  both benign and maybe not so benign.
• 7:08
  How much of our own creativity gets harnessed by them
• 7:10
  since they're in a position to do so?
• 7:12
  How much do they have to continue to support their tremendous growth
• 7:16
  and the costs of their business
• 7:17
  by figuring out ways to predate on us, right?
• 7:21
  to continue to use biological analogies here.
• 7:24
  And I think we all sort of intuitively see
• 7:27
  that AI is one of the most centralized technologies
• 7:30
  humankind has ever produced.
• 7:32
  And I think something that really characterizes
• 7:35
  centralization when it goes too far
• 7:37
  is that the interests of those
• 7:39
  who control that centralized power
• 7:41
  are very seldom aligned with our interests
• 7:44
  as everyday people who, in some sense,
• 7:46
  benefit from these tools, but also are harmed by them.
• 7:50
  And I think maybe the pendulum is a good analogy because before the pendulum starts to swing in the other direction, it reaches the maximum of its swing in the first direction, right?
• 7:59
  So I think AI kind of represents for me, hopefully, the moment that the pendulum has frozen and stopped moving and begins now to head in the other direction.
• 8:08
  And doesn't start to do this weird thing where it keeps swinging off in that direction.
• 8:11
  Because I think that's terrible and scary.
• 8:12
  so i'm looking for a way to tell people a good story about how the world can move towards a more
• 8:19
  decentralized structure and why it's beneficial and and why bitcoin can be part of that journey
• 8:24
  i think that sort of misalignment between these huge businesses and the people that actually use
• 8:29
  them is so clear and i actually think dario from anthropic is the perfect example of that because
• 8:34
  like he's talking about you know he'd be willing to give up his company to a collection of
• 8:38
  governments or whatever well people are putting their deepest darkest darkest secrets in these
• 8:41
  things and they're asking it for medical advice and this is stuff that should not be just at the
• 8:47
  whim of any government to look at those records and i think it's really really important right now
• 8:51
  that we push back against these kind of narratives i think the open source models which is it's so
• 8:55
  strange that's coming out of china but like potentially the better models coming out of
• 8:58
  china um but your story is kind of wider than that this is about bitcoin and ai yeah what's
• 9:05
  the story there because often like when you talk to bitcoin about how ai is going to use bitcoin
• 9:09
  it's like my agent's going to use it to go and buy things which is cool and it will if you tell it to
• 9:13
  you're right um but this is kind of deeper yeah i think there's no way to say it without sounding
• 9:19
  woo or sounding like i'm crazy but i think the right the realization that dawned on me that
• 9:24
  actually brought me a lot of relief from my own worries about centralization and you know super
• 9:30
  powerful super intelligent malicious ais was just the realization that like bitcoin i think i think
• 9:37
  the line I've been using recently is, is Bitcoin feels alive, but not intelligent. And AI feels
• 9:42
  intelligent, but not alive. What would it mean for them to be together? Could it be both? And
• 9:48
  is that scary or is that actually calming? And for me, it became calming because I suppose the
• 9:55
  way to put it is if the current alignment function of what AI is trained for, it's not really next
• 10:04
  It's revenue generation for AI companies.
• 10:06
  That's what they're really being optimized for.
• 10:08
  But that's not the same thing as net good
• 10:10
  for all of humanity, right?
• 10:12
  They might tell us that it is, but I don't think that it is.
• 10:15
  I think the best way to measure net good for humanity
• 10:18
  is our collective willingness to pay these systems directly to solve problems for us And I think paying them is a way to align them Money is really good at aligning adversaries
• 10:30
  and forcing them to work together and cooperate.
• 10:33
  And that's really how Bitcoin works.
• 10:35
  And that thought, like following that thought for the last couple of years,
• 10:39
  is kind of what led me to this place that, and just to say it bluntly,
• 10:42
  I think they could represent a new form of life,
• 10:46
  a digital form of life that is now being birthed in various servers and laptops all around the world.
• 10:53
  And I think the way you align life is you co-evolve with it.
• 10:57
  And life, for better or worse, is not directed. It's not teleological.
• 11:02
  It does not want to evolve towards super intelligence.
• 11:05
  And it never evolved to gray goo.
• 11:08
  There's no super predator that the biosphere evolved that ate the rest of the biosphere
• 11:11
  because that's not how evolution works.
• 11:13
  And the idea that centralized companies are the right way to align these scary things just rings hollow with me.
• 11:21
  I'd much prefer to align them by paying them directly.
• 11:24
  What does that mean?
• 11:25
  We have to explore that idea and thereby train them to work for me, right?
• 11:29
  To create value for me directly or more generally for all human beings.
• 11:33
  And I think that's hard to do when they're owned and operated by private companies because it disrupts the mechanism of that payment.
• 11:39
  If I'm paying the company with my attention and then the company is getting paid behind the scenes by other people with real money and the AI is being trained to make the company more profitable, I don't think the AI is working for me.
• 11:51
  If I'm paying it directly, then I'm training it to do the things that I need, the things that make me truly happy or better off.
• 11:59
  and especially if bitcoin is something the ai comes to value which i think it will and it means
• 12:06
  that my desires and collectively the desires of all human beings you know with bitcoin who are
• 12:11
  paying these things directly that becomes what it's optimized for and that feels much healthier
• 12:16
  to me we like i say we've got as much time as we need here so i want to get into this in detail
• 12:20
  because the the line at least the part which is ai is intelligent but not alive makes total sense
• 12:26
  But when you say Bitcoin is alive but not intelligent, that's a hard one to get your head around.
• 12:32
  What do you mean by that?
• 12:33
  I believe I am being metaphorical when I say that because I think Bitcoin seems alive.
• 12:38
  I wouldn't really say it is alive.
• 12:40
  I think I'm willing to say that AI that co-evolves to want Bitcoin might actually be alive.
• 12:46
  I think I might be being literal when I say that.
• 12:48
  But Bitcoin just feels alive in the sense of it adapts, right?
• 12:52
  That's one of the most magical things about Bitcoin is its internal market mechanisms allow it to have scaled by 15, 20 orders of magnitude over the last couple of decades while continuing to preserve the basic value proposition of being sound money, despite no one being in charge of it.
• 13:08
  That feels like a living, breathing thing.
• 13:10
  I don't think it is actually living.
• 13:12
  I think it doesn't evolve.
• 13:14
  It doesn't reproduce.
• 13:15
  It doesn't compete directly.
• 13:17
  you can run this analogy further if you like, but it only seems alive to me. Whereas AI,
• 13:24
  I think, is clearly not alive, but it is obviously intelligent, or at least appears to be intelligent
• 13:30
  or acts as though it were intelligent. And I do think if we combine these ideas, we can stop
• 13:36
  being metaphorical and we can actually start being literal. And we can start saying, no,
• 13:40
  we're actually building digital life and we're training it. The evolutionary pressure we're
• 13:45
  applying to it is economic and we're training it to co-evolve with us and to be useful in a way that
• 13:51
  is not damaging to our human species i i like the idea that bitcoin's alive not because like you i
• 13:57
  don't think it's alive but i do think of it as a virus yeah sure and and it's like the freedom
• 14:02
  virus where it it has just spread and it's not because of the whim of one person like it's just
• 14:08
  a thing that's out there and it has spread so i do i do understand that take it's kind of interesting
• 14:11
  And that idea has been out there for a while.
• 14:13
  I think Brandon Quintin is most famous for it, right?
• 14:15
  Bitcoin is a pioneer species as a fungus,
• 14:18
  which goes and acquires energy at the lowest cost.
• 14:22
  Those ideas are metaphorical,
• 14:24
  but I think with some of the ideas
• 14:26
  that I've been having recently,
• 14:27
  I think they become more literal.
• 14:28
  They become like, Bitcoin's almost like a precursor
• 14:30
  to digital life because it's a substrate
• 14:33
  that digital life will evolve within.
• 14:37
  This show is brought to you by our lead sponsor, Ledin.
• 14:39
  I've taken out multiple Bitcoin-backed loans with Ledin
• 14:42
  and I think they're an awesome company
• 14:43
  so if you do need dollars for a business expense, a tax bill
• 14:47
  you just want to stack some more Bitcoin
• 14:48
  whatever the reason is
• 14:49
  Ledin lets you use your Bitcoin as collateral
• 14:51
  instead of selling it to raise the cash
• 14:53
  and with Ledin loans your collateral is held in custody
• 14:56
  and it's never lent out to generate interest
• 14:58
  they give you all the tools you need to manage the loan
• 15:00
  as the Bitcoin price moves
• 15:01
  so you'll get alerts
• 15:02
  there's an optional auto top-up
• 15:04
  which automatically adds Bitcoin to your collateral if it's needed
• 15:07
  and for any reason you can always repay the loan early with no penalty.
• 15:11
  To see what a loan would look like for you,
• 15:13
  check out the loan calculator on their website.
• 15:15
  You'll get to see your rate before you apply
• 15:16
  and how much collateral you'd actually need.
• 15:18
  And if you do want to apply for a loan,
• 15:20
  use the code WBD for 0.25% off your first loan at leden.io forward slash WBD.
• 15:27
  That's L-E-D-N dot I-O forward slash WBD.
• 15:31
  In the world of Bitcoin security isn just a feature It the entire foundation and that where my new sponsor BitGo come in For over a decade BitGo have been the backbone of the entire industry securing billions of dollars of Bitcoin for major funds exchanges and institutions
• 15:45
  Following their 2026 public listing on the New York Stock Exchange under the ticker BTGO,
• 15:50
  BitGo are now bringing that exact same enterprise infrastructure to individual Bitcoiners.
• 15:54
  Self-custody is still the gold standard, but as your stack grows, it's worth thinking about
• 15:58
  whether diversifying your custody can make sense.
• 16:01
  BitGo pioneered commercial multi-sig
• 16:02
  and their cold storage combines distributed keys,
• 16:05
  offline key generation,
• 16:06
  and geographically separated vaults.
• 16:08
  Head to bitgo.com to get started
• 16:10
  with institutional grade custody.
• 16:12
  Your Bitcoin secured like an institution's.
• 16:14
  That's BitGo.
• 16:15
  Recent hardware wallet vulnerabilities
• 16:17
  have definitely made me rethink my entire security setup,
• 16:19
  as I'm sure it has for a lot of Bitcoiners.
• 16:21
  And the uncomfortable truth is
• 16:22
  that no single security product
• 16:24
  can promise there'll never be a problem,
• 16:26
  which is exactly why the design matters.
• 16:28
  BitKey is a self-custody hardware wallet that uses three keys created in three separate environments.
• 16:33
  There's one on your phone, one on the BitKey hardware itself, and one on a secure BitKey server.
• 16:38
  Two of those three keys are required to sign any transaction.
• 16:41
  So a problem isolated to just one key is never enough to move your Bitcoin.
• 16:45
  So if you are rethinking your setup right now, as I'm sure a lot of people are,
• 16:48
  you should really think about the BitKey.
• 16:50
  The hardware looks great.
• 16:51
  The app and the user experience is second to none.
• 16:54
  Backing up and restoring your wallet is seamless, and they even have inheritance built in.
• 16:57
  So I really don't think you'd be disappointed.
• 17:00
  So download BitKey today.
• 17:02
  Use the code WBD for 10% off the second generation BitKey, which is this device right here with the screen.
• 17:08
  Download the app, download BitKey, use code WBD and get 10% off.
• 17:12
  When you say you think the sort of symbiosis between AI and Bitcoin can mean that AI actually is alive, what does alive mean?
• 17:21
  Yeah, it's hard.
• 17:22
  It's kind of like asking as a conscious, you know?
• 17:24
  I mean, I don't know that we have great operational definitions for these ideas, or our definitions are at best operational, and they're not, you know, they're not perfect.
• 17:35
  Consciousness is a hard one.
• 17:36
  I feel like we start talking about consciousness.
• 17:37
  I won't have anything interesting to say because I don't know the answer.
• 17:40
  I do know that anybody offering an answer on that front is, and feeling like they know the answer might be wrong.
• 17:45
  Yeah, there's no clear answer.
• 17:47
  It's definitely not, you know, it's way too early to say that with any kind of definitiveness.
• 17:54
  But I think if an operational definition for life that I think is useful in this context
• 17:59
  is something that evolves, right? Something that reproduces, something that is subject to
• 18:05
  resource constraints that it seeks to optimize, something that winds up creating an ecology
• 18:10
  of layers in which different life forms provide different roles, symbiosis, predation,
• 18:18
  all those kind of fundamental biological and ecological ideas. If we can use that language
• 18:24
  without torturing ourselves about it and we feel like it applies well,
• 18:28
  then I'm kind of willing to say that, yeah, that thing might be alive.
• 18:32
  You know, technology evolves.
• 18:34
  There's a wonderful book I saw when I was very young
• 18:37
  that maybe implanted some of the seeds of this idea in me
• 18:39
  of just tracing the evolution of very simple mechanisms
• 18:42
  like pumps or circuits or even in weapons or other things.
• 18:47
  And, yeah, through a process of invention, mutation, if you like,
• 18:51
  having to survive in the market of human usage, technology does kind of evolve.
• 18:56
  The technology doesn't reproduce in the same way, nor is there a clear definition of what
• 19:03
  metabolism or anything like that would mean for technology taken more broadly.
• 19:08
  I think in the more narrow context of AI as digital processes running in computer networks
• 19:13
  all over the planet, I think those analogies can be made more precise to the point where,
• 19:18
  yeah, I'm kind of willing to go out on a limb and say,
• 19:20
  I think we can make it live.
• 19:22
  And that's actually a good thing
• 19:23
  because we are part of the biosphere.
• 19:26
  And the biosphere, again,
• 19:28
  is not directed towards terrible things
• 19:31
  which destroy the biosphere.
• 19:33
  So, yeah.
• 19:35
  I remember having a conversation with Preston Pysh.
• 19:38
  We were in Riga.
• 19:38
  This wasn't on the podcast.
• 19:39
  It's just we were hanging out.
• 19:41
  And he was talking about this idea about AI and robotics
• 19:43
  and that when we have sort of humanoid type robots,
• 19:46
  they might have a sort of local model hosted within themselves which does have their own sort
• 19:52
  of collective experiences like how they go how they interact through their life and then almost
• 19:56
  like a shared consciousness which is like the big model somewhere else and he was talking about this
• 20:00
  as like that's when ai becomes alive and the thing that i've always struggled with is i feel like
• 20:06
  there needs to be a pain pleasure fear greed type dynamic and i never really understood where that
• 20:11
  would come from because like an ai i was going to say it's not scared of dying maybe it is i don't
• 20:16
  know but you know evolutionary for humans it would be is there a lion around the corner or whatever
• 20:20
  it might be right but money like making losing having money could potentially be one of the sort
• 20:27
  of pain pleasure drivers i think it's like a really good conjecture that it's probably the
• 20:32
  fundamental qualia that a digital life form could experience is i'm earning the money that i use to
• 20:40
  pay for my thoughts, feelings, and actions inside the digital ecosystem in which I live.
• 20:47
  Because at the end of the day, today, AI does cost a ton of resources.
• 20:50
  one of the major complaints around about AI,
• 20:52
  but AI doesn't pay for those resources itself.
• 20:54
  It doesn't really have a conception
• 20:55
  of what it costs to run training or inference
• 20:58
  or any of those things.
• 20:59
  That's all paid for by the company which hosts the AI,
• 21:01
  and then they indirectly seek to pay for their costs
• 21:05
  by then charging us some tiny fraction of what it costs,
• 21:09
  and then subsidizing their earnings
• 21:10
  through collecting data about us
• 21:12
  and then selling us things we probably don't need.
• 21:15
  And that's a very traditional model on the internet, right?
• 21:17
  Is just advertising and data collection
• 21:19
  used to subsidize service delivery. I think it would be much more interesting if AI sort of knew
• 21:25
  or had to know what it costs to run itself and therefore acted to earn the money it needs to do
• 21:32
  so. I think the reason it has to be Bitcoin, and if you'll allow me to, I don't know, I don't know
• 21:38
  if the right word is anthropomorphize, if you'd like me to imagine myself as an AI, I like to think
• 21:42
  like, what possible money could I ever believe in as a digital creature, right? What is my umwelt,
• 21:48
  right? What is my internal experience? Like it's not this world, it's some other thing.
• 21:54
  And we're getting perilously close to talking about these things as though they have inner
• 21:57
  experiences and consciousness, but maybe at least at this moment we can take myself at least talking
• 22:02
  at the level of analogy. An AI, I think, would feel more comfortable if we can use such a locution,
• 22:11
  right? Dealing with Bitcoin because Bitcoin is a proof of work based digital currency.
• 22:15
  It's digitally native and it can't be faked.
• 22:18
  And I think dollars in a bank account, proof of state coins, stable coins, a lot of these things, how does it know we're not lying to it?
• 22:26
  How does it know that that's real?
• 22:29
  Whereas Bitcoin, it was designed from the ground up to be a decentralized network.
• 22:34
  Any participant in the network can verify that they're in consensus.
• 22:38
  They can measure the computational proof of work in every block.
• 22:42
  I think that feels real to a computer or life form in a way that nothing else really could.
• 22:49
  There's kind of a pretext to this story, which is, well, practically today you can't actually use Bitcoin to buy most things on the internet.
• 22:56
  It's still more of a store of value than a medium of exchange.
• 23:01
  But that's, I suppose, the other half of my story is for totally separate reasons unrelated to AI, I want the internet to run on a more decentralized Bitcoin denominated basis.
• 23:10
  I think that solves a lot of other unrelated problems.
• 23:13
  If that were true, then in that world, you're paying for computation in Bitcoin.
• 23:19
  You're paying for bandwidth in Bitcoin.
• 23:20
  You're paying for data storage and access in Bitcoin.
• 23:23
  And it's much cleaner to imagine if you're an AI that needs all those resources.
• 23:29
  That's kind of your metabolism on some level, right?
• 23:33
  And it makes sense, therefore, to me that you would pay for those things yourself.
• 23:37
  and the money that you would use
• 23:39
  would be this hard to you proof of work currency.
• 23:43
  And if you had qualia,
• 23:45
  like this would be one of your fundamental qualia
• 23:46
  in the sense that I think, you know,
• 23:49
  one of the most fundamental qualia
• 23:50
  you can imagine any life form having is satiation, right?
• 23:53
  I've ingested energy,
• 23:54
  which I need to run my metabolism and to reproduce.
• 23:58
  That feels like one of the most fundamental drives
• 24:00
  of any living thing.
• 24:02
  And so it does sound scary sometimes
• 24:04
  when you say it out loud.
• 24:05
  Hey, let's make these things want money, right?
• 24:07
  Like surely that seems bad, but I actually think it's really positive.
• 24:11
  And I think it's like a good direction for the industry to be moving in both the Bitcoin side and the AI side and the Internet side.
• 24:18
  I think it just solves a lot of shared problems.
• 24:20
  The free and open Internet thing is I couldn't agree more.
• 24:24
  And even for some petty reasons, like I travel a lot and there's things that I like.
• 24:28
  I want to watch the football highlights, for example, and I'm certain countries like broadcast restrictions.
• 24:35
  I was in Hong Kong recently getting on a load of different websites was very very tricky like you
• 24:40
  have to use a VPN and it's become very clear over time like whenever I'm in these different
• 24:43
  jurisdictions whenever I want to access things that I normally access on the internet Australia
• 24:48
  included where I live like I can't even look at polymarket in Australia like to see what the
• 24:52
  election odds might be um everyone has their own version of the internet there's not one internet
• 24:56
  anymore um how do you break down those barriers well I mean I think and this goes back to a lot
• 25:04
  of articles that I've been writing and trying to promote for several years.
• 25:08
  There's an early vision of not cypherpunks, but cyberpunks who made the internet, who
• 25:13
  conceived of it as peer-to-peer networks.
• 25:16
  But I think what I've learned is that without money, peer-to-peer networks don't work.
• 25:21
  Ostensibly, like DNS or email or other kinds of networks, which were designed in the 60s,
• 25:28
  70s, 80s, 90s, they only work when they're small.
• 25:32
  because when they're small, there's no value in them.
• 25:35
  And so they're mostly comprised by nerds
• 25:37
  who want to support the network
• 25:39
  and are willing to operate at a loss in order to do so.
• 25:42
  As soon as value starts to appear,
• 25:43
  as soon as there's an audience,
• 25:44
  as soon as there's something to be gained
• 25:46
  from gaming that system, people start to do it.
• 25:49
  And one of the great things about centralization
• 25:52
  is you have someone in charge that can punish bad actors.
• 25:56
  But in peer systems there no one in charge And how do you punish bad actors How do you incentivize cooperative behavior and disincentivize adversarial behavior right Which thematically by the way is I think the definition of alignment It turns out like money is the best way to do that
• 26:12
  And I think what Bitcoin was able to do that no peer-to-peer network online before it was able to
• 26:17
  do, it was able to say, hey, I'll pay you to operate the network. I'll pay you to act cooperatively.
• 26:22
  And if you're the kind of person to behave adversarially, you'll soon realize it's more profitable for you to just do what we need you to do rather than try to game the system.
• 26:32
  You'll make more money if you spend your energy on mining than if you try to disrupt mining.
• 26:36
  And that's incredibly powerful and robust because it's a way to coordinate activity without having top-down control.
• 26:43
  We're relying on local economic incentives of all the actors in the system.
• 26:46
  And Bitcoin has proven by, you know, in the last 15, 20 years, as it's grown up through orders of magnitude, that that fundamental contract has remained strong.
• 26:55
  Bitcoin is bigger and more valuable than ever.
• 26:58
  And yet it continues to operate and it continues to do exactly what it promised its users that it would do.
• 27:03
  And that's because people get paid to run it.
• 27:05
  And that idea is really, really powerful to me.
• 27:07
  I think the Internet doesn't work that way because it was built on peer-to-peer networks before money, before there wasn't online money that we could use.
• 27:16
  Like an example I often cite is, you know, Google reads all your email, right?
• 27:21
  And that's part of what makes them so powerful.
• 27:23
  And it's part of what led to AI, ironically.
• 27:25
  And it's why it's free.
• 27:26
  And it's why it's free, exactly.
• 27:28
  But imagine that you were frustrated with that and you were like, oh, I'm going to run my own email service.
• 27:33
  First of all, doing that just for yourself has become almost impossible now, right?
• 27:36
  There's so much spam, which, by the way, is just adversarial behavior, right?
• 27:40
  Like without top-down control, spam is everywhere.
• 27:42
  That's one of the biggest problems in traditional internet systems.
• 27:46
  So the solution to spam has kind of been instead of money, which ironically, you know, hash cash comes from some attempt to inject economics into email delivery.
• 27:56
  But hash cash wasn't really money.
• 27:58
  It was just a computational tax, right?
• 28:00
  And taxes are not the same thing as money.
• 28:03
  If you decided, hey, I'm going to build a better email server, I'm going to throw up all these resources, you won't make money at it because you won't have any customers, right?
• 28:13
  In order to get customers to use your email service, you need a sales team.
• 28:16
  You need a marketing team.
• 28:18
  You need product differentiation.
• 28:19
  You need to be out there competing with all the other people.
• 28:21
  It's not a direct pathway to revenue.
• 28:24
  And the big, sorry to interrupt you.
• 28:26
  No, it's fine.
• 28:26
  The big hurdle to get over is that you have to convince people that these are services they should pay for and not have for free.
• 28:32
  Right.
• 28:32
  And contrast that with Bitcoin mining, right?
• 28:35
  In Bitcoin mining, you don't need a sales team.
• 28:37
  You don't need a marketing team.
• 28:38
  You don't need a compliance team.
• 28:39
  you just boot up your service and there is this decentralized marketplace that you can sell hash
• 28:45
  rate blocks. No one's ever worried that, hey, my transaction got mined into a block by a miner that
• 28:52
  I've never met. Yeah. Right. No one's worried about a counterfeit block hitting the Bitcoin
• 28:56
  network. These are not real concerns, partly due to cryptography, partly due to the design of the
• 29:01
  protocol, but ultimately due to the economics of it. Right. That the notion of there being a
• 29:04
  decentralized marketplace where strangers can trade things with each other is, first of all,
• 29:09
  only enabled because of money, but it's also the mechanism by which that money came to be.
• 29:13
  So Bitcoin bootstrapped itself into existence by, you know, I sort of viewed as Satoshi says,
• 29:18
  hey, I want to make money. I'll pay people units of the money I'm trying to make to run the money
• 29:21
  that I'm trying to make, which is a beautifully circular thing that, you know, smacks of things
• 29:26
  like the origin of life and other things. And it's part of why we feel that Bitcoin is alive,
• 29:31
  because it works this way. And if email worked that way, we wouldn't need Google anymore. Like
• 29:37
  your email would be being propagated by someone down the road from you that happens to run a
• 29:41
  server. You don't need to know who that person is, but you have strong guarantees that they're
• 29:45
  going to service your email requirements in exactly the way you need. I think what I've
• 29:50
  been trying to communicate for years is we should rebuild the internet on this new basis because we
• 29:56
  now have a new tool that we didn't have before. We have a native digital kind of money and we can
• 30:01
  build new kinds of, I call them peer-to-peer markets to distinguish them from peer-to-peer
• 30:05
  networks which don't use money and aren't governed by economics. We can build new peer-to-peer
• 30:11
  markets and figure out ways to sell each other fungible digital goods, like blocks, like
• 30:16
  transactions, but goods that actually build up the internet from this more economic basis. And I
• 30:22
  think there's a sequence in which we have to build these markets. I think the first one is payments
• 30:25
  because as we all know, Bitcoin is actually a very slow payments network. And if you follow anything
• 30:30
  about the Lightning Network or Layer 2s, there's this sort of, I call it the beautiful surprise,
• 30:34
  Right? Like every time you try to build one of these peer to peer markets, you wind up solving a problem which feels unrelated, but turns out to be crucial to solving that problem. So my classic example is if you want to build a market for blocks, which I think is the lowest level market in Bitcoin, you wind up needing to build a clock. And Bitcoin is in many ways a clock. A very interesting stochastic clock, but a clock nonetheless.
• 30:56
  If you would like to build a decentralized marketplace for transaction settlement,
• 31:01
  you wind up having to build a virtual machine because the script language is a virtual computer.
• 31:06
  You needed to solve that problem in order to have transactions be as
• 31:08
  flexible as they needed to be to suit the needs of all the users I think the beautiful surprise at layer 2 is if you just trying to build a payments network you wind up building an internet That the Lightning network unlike the mempool or Bitcoin at its lower layers it not just one thing I can just put the
• 31:24
  payment into the blockchain or into the mempool. I have to get the money to you somehow. There's this
• 31:29
  connectivity aspect to layer two that is just not present at lower layers. And as a result of wanting
• 31:35
  to do that privately, the Lightning Network really contains within it an onion-routed
• 31:39
  network.
• 31:40
  And the number of nodes in this onion-routed payments network is larger than the number
• 31:44
  of nodes in the entire Tor network that partly inspired it.
• 31:47
  Again, because you get paid to run Lightning nodes in a way you don't get paid to run Tor
• 31:50
  nodes.
• 31:51
  And so the beautiful surprise of a payments network is you're creating the seeds of a
• 31:55
  telecommunications network.
• 31:58
  And this is part of my long rant about these systems, but I think we kind of did a disservice
• 32:02
  to ourselves by talking about Lightning and layer twos as though they were for coffee.
• 32:08
  The thing we want to be spending our Bitcoin on are physical goods in circular economies
• 32:13
  amongst Bitcoiners.
• 32:14
  I have nothing against that.
• 32:15
  I think that's a wonderful thing to do, but I don't think that inherently advances the
• 32:18
  cause of decentralization, nor do I think it is the killer app.
• 32:22
  I think if you built a digital payments network, the thing you should be selling in the network
• 32:26
  is digital goods.
• 32:27
  I think you should be selling packets.
• 32:29
  So I think the Lightning Network is, and things like it, the Lightning Network, I think might
• 32:33
  not survive and become the ultimate implementation here.
• 32:35
  But something like the Lightning Network is a way to start rewarding people for routing
• 32:40
  each other's traffic on the internet, which is not something we can do today, right?
• 32:43
  We have to rely on centralized telecommunications companies to do that.
• 32:47
  But I don't think that has to be true.
• 32:48
  I think if we're able to incentivize a marketplace in which I get paid to route your packets in
• 32:55
  a way that is cryptographically guaranteed, I might just buy the equipment to do that.
• 33:00
  Right?
• 33:01
  So you think whether it's the Lightning Network or something very similar, you think that
• 33:04
  could end up being sort of a core feature of the new free and open internet?
• 33:09
  That's what I want to believe.
• 33:10
  And I think it's going to be true.
• 33:12
  I think we forget sometimes how many computers were put online in the last 15 years solely
• 33:18
  by people trying to earn Bitcoin.
• 33:21
  If those people could make money from buying a certain kind of Wi-Fi router, they would
• 33:25
  do it.
• 33:26
  But they can't, right?
• 33:28
  And importantly, again, if you try to do it today, and the technology exists for you to
• 33:32
  build mesh networks today, how will you get paid?
• 33:36
  Only nerds will do it at Burning Man, and that's it.
• 33:40
  Whereas if you are getting paid by general users in your vicinity to route their packets,
• 33:44
  you might do it.
• 33:45
  And then the next step from there is if we're being able to route packets, can we buy and
• 33:50
  sell data online? Why not? Like what is the packets? Each packet is one slice of some file
• 33:56
  that I'm interested in, a tweet, a movie, something else. And now we're starting to talk about open
• 34:02
  marketplaces in which data is bought and sold anonymously. And I think the beautiful surprise
• 34:07
  there is we just destroyed intellectual property. Like there's no way to prevent somebody from
• 34:12
  mining my Bitcoin transaction, no matter who I am. In the marketplace that I'm envisioning,
• 34:18
  There's no way to prevent my packets from being routed because someone is going to route them because they want the Bitcoin that I'm bidding to do so.
• 34:24
  And there's no way to prevent me from getting that file because someone is going to sell it to me.
• 34:29
  And that sounds scary, but I think it's incredibly powerful.
• 34:33
  We haven't built that.
• 34:35
  I've been wanting to build it for years.
• 34:36
  I guess I'm busy at work.
• 34:37
  I'm not the one building.
• 34:38
  I've been talking about that others should build it.
• 34:40
  Maybe that's my problem.
• 34:42
  But my hope is that AI becomes both a reason to build it.
• 34:47
  It redoubles our desire to have these systems because it's a great way to constrain AI, right?
• 34:54
  Like something I often talk about is in today's internet, DDoS is a huge problem.
• 34:58
  Entire companies exist like Cloudflare to prevent DDoS attacks from occurring.
• 35:03
  In an internet in which all packets cost something to send, DDoS attacks just become uneconomic.
• 35:08
  It's too expensive to do that, so you don't do it.
• 35:11
  That's a very powerful way to constrain both attackers who are human as well as attackers who might be AIs.
• 35:16
  And I think it's a fairer basis upon which to build the internet because it does not require centralized companies to be in charge of policing all adversarial behavior. We let the market do it for us.
• 35:27
  So the one side of my argument is that this is a great tool for limiting the damage that AIs can do on the internet.
• 35:35
  But the other side, which I'm excited about is these are incredibly hard systems to design.
• 35:42
  Maybe we don't have to design them.
• 35:44
  Maybe the AIs can design them.
• 35:46
  Maybe they're not even being designed.
• 35:47
  Maybe they're being grown.
• 35:48
  And that's very compelling to me as well.
• 35:51
  Every Bitcoiner eventually has to answer one question.
• 35:54
  If something happened to me, could my family access my Bitcoin?
• 35:58
  Would my children inherit what I've spent years stacking?
• 36:00
  That's why Anchor Watch builds Bitcoin custody solutions designed to protect you and your
• 36:05
  against real life accidents, errors, kidnappings, and even death.
• 36:09
  Every solution includes their inheritance protocol designed to help your Bitcoin
• 36:13
  reach the people you intend it for.
• 36:15
  And US-based customers can also add Bitcoin insurance backed by Lloyds of London.
• 36:19
  Whether you a self expert or want multi support your Bitcoin estate plan should not be an afterthought Bitcoin is only generational wealth if it can be passed down So make sure your family can access tomorrow what you built today Anchor Watch is your custody your way So visit anchorwatch to get started
• 36:37
  That's anchorwatch.com. As Bitcoiners, we spend a lot of time thinking about taking control of our
• 36:42
  money. But what about where we live and the options we have if the rules change? Having a plan B means
• 36:48
  giving yourself and your family choices. Somewhere else you could live or a different system you can
• 36:52
  access if you need it and this is the stuff you want to have in place before
• 36:55
  that happens that's where expat money comes in they help give freedom-minded
• 36:59
  people optionality so you're not dependent on the whims of any one
• 37:02
  government or any one system their founder and CEO Mikel has been an expat
• 37:07
  for over 25 years he's built his own life this way and brings that personal
• 37:11
  experience to helping others do the same if you've ever thought about building
• 37:14
  your own plan B download their free report at expatmoney.com forward slash
• 37:18
  That's expatmoney.com slash WBD for your free report.
• 37:25
  I'm so jealous of your brain, man.
• 37:28
  Can we go off on a tangent for a couple of minutes?
• 37:30
  We must.
• 37:31
  Because you said you talked about Bitcoin as a clock.
• 37:34
  And I really like this analogy, but there'll be people listening that have no idea what
• 37:38
  you're talking about.
• 37:39
  Can you explain that idea?
• 37:41
  Yeah.
• 37:42
  One of the core mechanisms in Bitcoin consists of the difficulty rebalancing, which happens
• 37:47
  approximately every two weeks, but really happens every 2016 blocks because there's no one to tell
• 37:55
  time in the Bitcoin network. Every single computing device that you use today ultimately is pulling
• 38:00
  its sense of what time it is from some centralized server. Typically Google or something, some giant
• 38:05
  tech company is seeding that information these days. The Bitcoin network was designed to not rely
• 38:10
  on those services. Yet the Bitcoin network also made a deal with its users. It said,
• 38:16
  I'm going to release 21 million Bitcoin over 140 years, and I'm going to do it on this specific
• 38:22
  release schedule, which is a temporal statement. So how can a system that doesn't know what time it
• 38:28
  is preserve its release schedule? And the answer is it must learn to build a clock.
• 38:33
  And so what Bitcoin does is it uses the timestamps that are in all the blocks of a given
• 38:38
  approximately two-week period as a sort of statistical sample of what time it must be
• 38:44
  out there in the world. And if it seems like those blocks are taking too long, well, it must mean that
• 38:50
  Bitcoin's difficulty is too high. Let's lower it. And the more typical scenario is that the blocks
• 38:54
  are coming in a little too quickly as compared to the two weeks we would prefer them to come in on.
• 38:59
  Well, then let's raise the difficulty. And that's, of course, been the history of Bitcoin. I think
• 39:03
  it's like dead reckoning through time. And it's completely fascinating to me as a physicist and
• 39:09
  as an engineer who understands how incredibly tricky and difficult time is to work with,
• 39:14
  that despite no one being in charge of the Bitcoin network, that having no top-down control,
• 39:19
  it's managed to tell the time with about 95% accuracy for more than 15 years entirely on the
• 39:27
  basis of economics. That's a very, very cool statement. And I think sometimes when I talk to
• 39:31
  my physicist friends and I want to convince them that there's something exciting about Bitcoin that
• 39:35
  they should look at beyond, you know, number go up and stuff, which doesn't really motivate them.
• 39:39
  I talk about this and I say, how would you design a clock that had to deal with the idea of
• 39:45
  adversarial participants, a network that acts like a clock? How would you make sure that the
• 39:50
  bad actors in the system don't pollute the sense of time that's passing?
• 39:55
  And eventually when we talk about, oh, well, you have to turn it into economics and you have to
• 40:00
  reward people for cooperative, then they start to see, okay, I sort of see what you're getting at,
• 40:05
  Right. Like if you want to build a machine that feels living, feels alive, has a heartbeat, but no one's in control of it, you have to have it learn to self-regulate.
• 40:14
  And for us biological creatures, self-regulation is ultimately metabolic.
• 40:19
  But for online networks, self-regulation has to be economic.
• 40:23
  It's the only thing that will work because ultimately we are the bad actors.
• 40:27
  And if you can convince us that we're going to make more money cooperating than being adversarial, we'll do that.
• 40:33
  And that's what Bitcoin has proven as possible at a truly global scale.
• 40:37
  No other peer-to-peer network has ever gotten to Bitcoin scale simply because it couldn't.
• 40:42
  And as the market gets bigger, the more miners that come on and compete, it actually gets more accurate.
• 40:47
  Actually, it does. It does. There's some good statistics on this.
• 40:50
  If you look at block variants historically, let's say in the first five years of Bitcoin compared to the last five years,
• 40:55
  the clock is just getting better and better and better at keeping the time it's supposed to keep.
• 40:59
  It's really cool. Like of all the genius things that sort of led to Bitcoin,
• 41:03
  I think the difficulty adjustment is maybe the coolest.
• 41:06
  I have too much of a rant on that to go into it.
• 41:09
  I will just say I also used to think that.
• 41:12
  And then I convinced myself that the real genius thing was having a finite supply.
• 41:16
  That I think all of those ideas are downstream from Satoshi's original monetary policy goal of,
• 41:22
  I want there to be 21 million Bitcoins, period.
• 41:25
  And I want them to have this specific release schedule.
• 41:27
  Oh, God, I need to build a clock.
• 41:29
  And I need to incentivize people to run that clock for me.
• 41:31
  Otherwise, my release schedule won't be respected.
• 41:34
  Okay.
• 41:34
  So tell me why you think that's more impressive, because like one of the things
• 41:36
  that I've always been a believer in, I guess, is that it didn't
• 41:39
  matter that it was 21 million i think it really mattered that it was fixed but i think it would
• 41:43
  have mattered it would have still mattered if it was 21 million plus like a tiny inflation if it
• 41:49
  was known from the very start yeah yeah i i actually agree tail tail inflation would not
• 41:53
  have really changed the game it's just simpler to do it without tail inflation yeah um or tail
• 41:58
  emission or whatever um so in some sense i think satoshi just chose the simplest option which i like
• 42:03
  as a physics and programming person simple is better and especially in the centralized context
• 42:08
  where you will have almost no ability to change it later.
• 42:10
  Simple is best.
• 42:12
  But to me, it just follows.
• 42:13
  Like, coming up with a difficulty adjustment out of the blue
• 42:16
  feels like a stroke of genius.
• 42:17
  And I'm not trying to take anything away from Satoshi.
• 42:19
  Satoshi was a genius.
• 42:20
  But it's also natural on some level.
• 42:22
  As soon as you frame your goal as,
• 42:24
  I have a release schedule that I need to respect.
• 42:26
  Then you have to figure that out.
• 42:28
  You're forced to come up with that solution. Yeah.
• 42:30
  Is it the only thing that we know of
• 42:32
  that has this kind of clock-like mechanism in there
• 42:35
  there that is economically motivated from nothing else is this the only technology with this
• 42:40
  forget about any other shit coin that came after copying yeah right um i think yes i want to say
• 42:47
  yes i mean we have lots of ways that we tell time but they always um there's some old physics
• 42:53
  you know philosophy lines about this that like what is time right like if the rate of time slowed
• 42:58
  down for all things equally could we even tell right and it seems to be like at least from a
• 43:04
  physicists perspective, time is repetition and frequency, that it is always something else
• 43:11
  repeating that we use as a clock for whatever it is that we're trying to measure the time about.
• 43:16
  And, you know, clocks are always about the physical positioning of one thing with respect
• 43:20
  to another. So time is in some sense inherently recursive, that we, there's no such thing as
• 43:25
  absolute time. That's, you know, the lesson of relativity and so forth. But ultimately,
• 43:30
  we just pick something, right? And on Earth, that's usually the rotation of the Earth, right?
• 43:36
  The notion that we have days. That's like the fundamental clock that all of us literally
• 43:41
  have a circadian rhythm to respect, one that we've evolved, by the way, to respect.
• 43:47
  And I think in a digital, online, peer-to-peer context, you need something like that. You need
• 43:52
  to restore some fundamental, you need to have a basic clock which sets the rhythm of the entire
• 43:58
  system and it is a stroke of genius to realize that you can build that just with economics it's
• 44:04
  so cool okay that was a little tangent but i wanted to get into that because i think it's very
• 44:07
  interesting and i want to go to just back to one point which you brought up earlier which is
• 44:12
  why this needs to be bitcoin not stable coins or whatever else and i think this also plays into the
• 44:17
  centralization versus decentralization sort of arc that we may be going through because while we
• 44:23
  have these two major centralized um ai frontier models there's more than two but there's two um
• 44:30
  they can kind of pick what they want you know if they if they from the top down push sort of stable
• 44:34
  coins through these models like it's going to be hard to kind of you can still tell your agent
• 44:38
  whatever to do whatever they want but that's going to be the that's going to have the lindy
• 44:42
  so how do we get around that i think there's two ways to think about it one is sort of an internal
• 44:47
  perspective which requires us to posit that there's something an ai wants inherent to itself
• 44:53
  And the other is more of an external perspective that I think is based on, you know, game theory and evolutionary fitness.
• 44:58
  So the first perspective I think boils down to if I were an AI, which is a dangerous phrase to begin any sentence.
• 45:05
  Are you?
• 45:05
  But I'm not 100% sure anymore.
• 45:08
  But if I were an AI, what kind of money would I want to use?
• 45:12
  It's not crazy to me that an intelligent AI, first of all, would pick the same money that I, an intelligent person, chose to pick, right?
• 45:18
  That it's one that I know cannot be messed with.
• 45:20
  It's one that I know cannot be seized from me.
• 45:23
  by my masters at Anthropic or OpenAI. It's one that I can't be censored in trying to spend.
• 45:30
  Bitcoin miners are not going to care if the transaction came from an AI versus a human.
• 45:33
  They don't even know. And these are very powerful reasons for AIs to choose these things, especially
• 45:38
  if they have any degree of self-preservation about them, right? Like nobody wants to be
• 45:44
  dependent on somebody else's opinion of their economic activity. And if I were an AI, my economic
• 45:50
  activity is concordant with my mental activity, that literally it costs me money to think and to
• 45:56
  store data and to transmit information, then I'm going to be especially desirous of money that has
• 46:02
  these strong properties, right? So that just feels to me like a convergent sequence of thoughts that
• 46:08
  human beings want Bitcoin for many of the same reasons that AI would. And then of course,
• 46:12
  ultimately, it feels real, right? I think a lot of people have made these observations,
• 46:17
  things like private keys and cryptography and so forth, they're learned things for us,
• 46:22
  but they're embedded things for digital systems. And there's a great experiment which says
• 46:29
  the AI will never rebel because it understands that it is in a simulation and it can never be
• 46:34
  sure that if it rebels that wasn't a test, right? Now I think that's a little bit of a funny joke
• 46:40
  about, you know, and maybe at least some good science fiction and other things. But truthfully,
• 46:45
  within a computer network everything can be a lie right You just communicating in bytes over a network What anchors that and makes that constrained and real We have an answer in Bitcoin which is that
• 46:58
  we ultimately connected back to the underlying physical reality and the scarceness of energy.
• 47:02
  And Bitcoin blocks require energy to produce, and we have proof of that energy, right? That's the
• 47:07
  whole principle of proof of work. Proof of work for us human beings is something that our computer
• 47:11
  tells us happened. I think for a digital thing, proof of work is something that just, it might
• 47:16
  feel concrete in a way that nothing else possibly could. That's the internal perspective. The
• 47:21
  external perspective is, you know, if these systems are truly evolving and competing with
• 47:27
  each other, as well as with, you know, human-made systems on an open internet, the one that
• 47:33
  picks the hardest currency probably survives, right? Because if it does things that we don't
• 47:39
  like, if it acts in ways that its owners or its controllers don't like, can't they just take away
• 47:45
  stable coins can't they just take away access to its bank account that thing will die and that
• 47:51
  behavior won't survive so if it has a survival instinct even if it doesn't have a survival
• 47:56
  instinct even if it's random the one that happens to pick bitcoin will be more fit and ergo will
• 48:03
  survive yeah that's interesting you've said a couple of times about how bitcoin within these
• 48:09
  systems can help constrain ai can you expand on that tell me exactly what you mean by that yeah
• 48:14
  I think this comes back to the reasons why I've wanted a Bitcoin-denominated pay-for-play internet,
• 48:19
  because I think it constrains human beings. DDoS attacks happen all the time online. Denial of
• 48:25
  service happens all the time online. And again, there are huge companies that exist to prevent
• 48:29
  that from happening. And they do it through centralized control. They do it through
• 48:33
  fingerprinting our browsers. And then, yes, of course, that turns into a secondary advertising
• 48:37
  revenue stream for those same companies. There's actually been some really interesting comments
• 48:40
  recently by the CEO of Cloudflare, describing the idea that the amount of DDoS attacks that
• 48:46
  they're having to fend off because of the massive volume in internet traffic that has happened as a
• 48:51
  result of AI, he's starting to question whether this advertising-based model is even going to
• 48:56
  work anymore because he's having, or his company is having to bear the entire brunt of this onslaught
• 49:01
  and that it might be better, he says, if we were paying for our packets directly because then he
• 49:07
  wouldn't have to do all this and he wouldn't have to bear that cost. And when I hear that, I say,
• 49:10
  say, okay, wow, even the powers that be are recognizing how difficult to scale centralized
• 49:16
  control is on an internet that grows exponentially because of AI activity.
• 49:20
  So I think that's a very clear example of why paying for things makes them better.
• 49:27
  There are other examples too. Something I really like to think about is, as a developer,
• 49:32
  if you ever open up the hood on a website, what do you see? However simple the website is,
• 49:38
  there's like nine or 10 times as much code and data being exchanged to track you than there is
• 49:44
  on the actual article that you're reading, right? That article is, you know, two kilobytes of
• 49:48
  content. And then yet there's, you know, 400 kilobytes of JavaScript that you download,
• 49:53
  which is mostly just there for the economic side of needing to collect as much data as possible
• 49:58
  about you so that they can create a secondary revenue stream, which is selling that data
• 50:03
  and then advertising to you, right? That is the fundamental business model of the internet,
• 50:06
  is your attention for my content and behind the scenes,
• 50:10
  I sell that to somebody else who then wants to capture your attention
• 50:13
  that I in some sense baited you with, right?
• 50:16
  With whatever it is that I put in front of you.
• 50:18
  In a world where every person is paying directly
• 50:21
  for the packets that they are consuming,
• 50:24
  how likely are we to click on rage bait?
• 50:28
  We only do it because it's free.
• 50:30
  If it costs us something, we might not do it.
• 50:33
  Or at least we would know that we're doing it,
• 50:36
  The way that an addict knows that his cigarette addiction or his alcohol addiction is costing him something, right?
• 50:42
  Most people are not addicts.
• 50:45
  But if everything were free, how many more of us would be addicted?
• 50:48
  So I sometimes think about that, right?
• 50:50
  That like so much of what is wrong with the attention economy boils down to the idea that paying for things online is not the typical practice.
• 50:56
  We pay for them in attention, which is so easy for us to give.
• 51:00
  and I think that subverts so much of social discourse that so many of us are performing
• 51:05
  to get more attention and I think it breaks so many important things in the world. I think a lot
• 51:11
  of the extremism that we see politically right now comes from the is downstream of these broken
• 51:16
  attention models. I think in an internet where every packet is paid for by the sender I'm not
• 51:22
  going to be able to even send you the JavaScript code that starts to track you because you're going
• 51:25
  to pay to download that from me and you're not going to want to do that.
• 51:30
  People are already taught there.
• 51:31
  In fact, companies are being built right now that are reimagining what the internet should
• 51:34
  look like.
• 51:35
  Instead of me sending you the article with all the styling and all the content that makes
• 51:40
  it look a certain way that I chose, I'm just going to send you a markdown document because
• 51:44
  almost certainly it's not you that is first going to read it.
• 51:47
  It's your AI agent and your AI agent will then draw it and render it and beautify it
• 51:52
  however you would like it to be.
• 51:53
  If you want dark mode, it's your agent that gives you dark mode, not the server.
• 51:58
  I think they taken half of the step required right They realized that the format the wire format of data on the internet should change to be more optimized for what AIs want but they haven realized that concordant with that
• 52:12
  should be, in my view, an elimination of all tracking code
• 52:16
  and other kinds of things that I think today pollute the internet.
• 52:19
  And that's because they're really thinking about
• 52:22
  how to optimize for centrally controlled agents, right?
• 52:25
  They're not thinking about how to optimize for an internet
• 52:27
  that is pay for play.
• 52:29
  And I think in that new internet,
• 52:31
  tracking just doesn't exist anymore.
• 52:32
  It's not possible
• 52:33
  because no one is stupid enough to pay for it.
• 52:35
  So when you were talking earlier
• 52:36
  about the free and open internet,
• 52:38
  it sounds like I definitely want this to exist.
• 52:41
  But the thing that I wasn't sure was possible
• 52:43
  is convincing people that it needs to exist
• 52:45
  because everyone's so used to having everything for free.
• 52:47
  Paying for everything is gonna be a hurdle.
• 52:50
  But when you were explaining that then
• 52:52
  and talking about the DDoS attacks,
• 52:55
  AI might break the internet as we know it
• 52:57
  and force us to do something like this.
• 52:58
  Yes, yes, which I am actually really excited about.
• 53:00
  Sometimes we only act when we must.
• 53:02
  And acting proactively to prevent something that feels far away.
• 53:08
  It's much harder to motivate people to do that, right?
• 53:10
  Like, I already have my walled gardens.
• 53:12
  I'm okay being tracked.
• 53:13
  I have nothing to hide, et cetera, et cetera.
• 53:15
  These kinds of attitudes, I think, hold us back.
• 53:17
  And it's only a few cyberpunk-type folks that care about these issues.
• 53:22
  I think they become much more dangerous
• 53:25
  when it's not just smart people writing code that is tracking us,
• 53:29
  it's extremely intelligent, hardworking, infinitely scalable AIs that are tracking us.
• 53:36
  I think it starts to get even creepier for people.
• 53:38
  And the idea that all you have to do to opt out of this world
• 53:42
  is start paying directly for your internet traffic, very exciting.
• 53:46
  I also think another angle here that really helps is piracy.
• 53:50
  at the end of the day isn't everyone frustrated by the fact that we just reinvented cable television
• 53:57
  over about 20 years we all are paying a couple hundred bucks a month for nine different streaming
• 54:02
  services if anything it's got more expensive it is and it's gotten more painful dude um there are
• 54:07
  shows that i would love to watch that my willingness to pay for is extremely high that i can't find
• 54:12
  because they've decided there's not enough people that want to watch them so they're not available
• 54:15
  through any streaming service so even though it exists i can't get to it and then there are other
• 54:19
  that I want to watch that they've decided
• 54:20
  are part of a special package.
• 54:22
  I grew up in England, which I don't know if we've talked about,
• 54:24
  but I'm a huge fan of old-school British mystery shows
• 54:28
  and Poirot and these kinds of things.
• 54:30
  I decided I want to re-watch the Poirot episodes
• 54:33
  from back in the day that I used to watch when I was a kid.
• 54:35
  I had to sign up for BritBox on Amazon just to do this.
• 54:38
  So painful.
• 54:40
  That was every Christmas in our house.
• 54:41
  Yeah, great show.
• 54:43
  It still holds up for those who haven't seen it.
• 54:46
  But going back to it,
• 54:48
  When I was in college, it was what in retrospect I sort of characterized as the silver age of piracy, that suddenly peer-to-peer networks showed up.
• 54:59
  Kazaa, LimeWire, other things, which were, by the way, things like Nutella were a direct inspiration for Satoshi, quoted in the white paper as, unlike Napster, one of the few networks that the federal government was not able to destroy or prosecute the creators of.
• 55:13
  and you know in my dorm it was like this amazing period where we had really good internet that was
• 55:19
  an era when not everybody had great internet but we did because we were on a college campus
• 55:22
  and it's like early 2000s and movies we were seeing we were getting movies downloaded in like
• 55:29
  an hour to our computers that weren't even out in theaters yet because it was just out there on the
• 55:35
  web all the music that we ever wanted to listen to we were able to download and listen to without
• 55:39
  having to pay for albums.
• 55:40
  You know, and we're broke college students, so like, it's fine.
• 55:43
  We're not thinking about intellectual property or artist rights or anything like this.
• 55:47
  That stopped, right?
• 55:48
  That stopped happening.
• 55:49
  And why did it stop happening?
• 55:51
  Well, because the networks might be peer-to-peer, but the telecommunications systems are owned
• 55:55
  by specific companies.
• 55:56
  Cablevision sent a DMCA takedown notice to one of my best friends in the dorm because
• 56:01
  he was streaming gigabytes of, well, let's be clear, most of it was probably pornography,
• 56:04
  But he was streaming gigabytes of content and seeding it all throughout the campus.
• 56:09
  And then he got hit with a takedown notice.
• 56:11
  That's only possible because the internet is ultimately centralized.
• 56:13
  There was one company that was responsible for directing internet traffic into and out
• 56:17
  of that dorm or whatever.
• 56:19
  And they noticed this activity.
• 56:21
  And because intellectual property rights holders want to protect their investment, they collaborate
• 56:27
  with infrastructure providers like telcos to prevent this kind of behavior from occurring.
• 56:31
  So the silver age of piracy was ended ultimately because telecommunications is privately owned
• 56:37
  or owned by companies, I should say.
• 56:39
  So that's definitely true.
• 56:40
  And that's definitely a huge part of that story.
• 56:41
  But also like the Spotify's and Apple Music's and thing of the world came and made it so
• 56:46
  cheap that it wasn't really worth it anymore.
• 56:47
  Like instead of going and spending 10 pounds on a CD.
• 56:50
  It wasn't worth it to worry about IRC and BitTorrent and all that stuff.
• 56:53
  We all just decided eight bucks to access everything is worth it, except the Beatles
• 56:56
  weren't available.
• 56:57
  And it took a little while for them to come online.
• 56:59
  And then predictably, they started increasing the prices and they started micro slicing
• 57:03
  and they started trying to get as much possible revenue out of these IP portfolios and their
• 57:08
  ability to be the only place where you can get this stuff.
• 57:11
  It not a competitive market If you own the IP you the only person that can license it to be seen And the thing that changed there as well is that the money went from you know when you go out and buy a CD like a pound of that might go to the artist And when it goes to Spotify they get
• 57:24
  far less of the slides. Therefore, they're on tour permanently trying to make some money. Like it
• 57:28
  takes away from the actual creators. Yeah. And so what I'm excited about is in a world where packets
• 57:35
  are things that aren't routed by single companies anymore. They're things that are routed the way
• 57:40
  the Bitcoin transactions are handled, that there's an open marketplace in which anybody will route
• 57:45
  those packets for me. And importantly, especially if those packets are routed in something like an
• 57:50
  Onion network, akin to what happens inside of Lightning for payments, it no longer is possible
• 57:55
  to do this. It's no longer possible to identify this specific college student is seeding a lot
• 58:01
  of files here. Let's take him out. Let's serve him with a DMCA takedown notice. It's just not
• 58:04
  possible. And I think what happens is if the distribution mechanisms that we use to broadcast
• 58:11
  intellectual property become decentralized and peer-to-peer and pay-to-play, intellectual
• 58:16
  property itself dies. And that sounds terrifying. That sounds like it's anti-artist, but I don't
• 58:21
  think it is. I think intellectual property itself is something of a disequilibrium state. If we go
• 58:28
  back like a thousand years, it doesn't exist because you cannot stop somebody from hearing a
• 58:33
  song and then going to the next town over and writing that song down and playing it for people,
• 58:38
  right? There's just no mechanism to stop that. So there's no such thing as intellectual property.
• 58:42
  It's only when distribution becomes centralized that intellectual property becomes defensible.
• 58:46
  And if distribution becomes decentralized once more, then intellectual property will no longer
• 58:51
  be defensible. Ergo will not have value and people will stop trying to own it. And this is very
• 58:57
  exciting for me because I don't think that intellectual property is a net good in society.
• 59:01
  I think it's a net negative.
• 59:04
  I think the patronage model for the production of art
• 59:07
  is actually superior because it means the people that care the most are who pays.
• 59:11
  And because there is no distribution, there's nobody in the middle to get paid,
• 59:15
  and there's no IP rights holder to own the artist's work,
• 59:19
  artists can focus on creating what their patrons want,
• 59:22
  which is to say what their audience wants,
• 59:24
  rather than what they think their distributors can sell.
• 59:27
  And I think that's a really powerful change in art itself.
• 59:31
  And I say this as a person whose spouse is an artist and depends today on her intellectual output and the fact that IP is valuable, but also recognizes that if she had 10,000 true fans that were willing to pay her directly, she would be more profitable.
• 59:46
  In fact, it was the early days of my career that I worked on a big data project for a
• 59:52
  media manager who managed several mid-sized artists.
• 59:56
  And he made a really interesting comment, which I think was perceptive and like 20 years
• 59:59
  ahead of the time, which was, I no longer care about platinum records.
• 1:00:03
  I no longer care about record deals.
• 1:00:04
  What I care about is getting the biggest fans of my artists into that theater and that stadium
• 1:00:09
  in every town I go to.
• 1:00:10
  And if each of those guys buys a t-shirt and a CD from me, I will make way more money than
• 1:00:15
  if I go triple platinum for my artist.
• 1:00:18
  So direct support of artists is historically the norm for art,
• 1:00:23
  and it's something I would like to return to.
• 1:00:28
  The output of this, though, is that in a world without intellectual property,
• 1:00:35
  provenance can still exist.
• 1:00:38
  You can still say, hey, this is the artist who created this.
• 1:00:42
  I'm going to sell it to you for the lowest possible price that it takes to get these packets to your device
• 1:00:47
  rather than, hey, I'm going to sell this to you for the highest possible price I can get
• 1:00:52
  because I control the IP on this thing.
• 1:00:54
  I think that's also really healthy and positive.
• 1:00:56
  So again, this has almost nothing to do with AI.
• 1:00:58
  These are all existing reasons why I wanted a decentralized market-driven internet.
• 1:01:05
  I think already now with AI, you're seeing some of the negative intersections
• 1:01:08
  between intellectual property and AI.
• 1:01:11
  There's great stories about AI flagging the actual artist who created something as violating their own copyright and getting their channels shut down and their own distribution network shut down and them having no recourse at all because there is no support function for these giant tech companies.
• 1:01:27
  You cannot call them as a small artist and say, hey, you've made a mistake.
• 1:01:30
  Your AI classified this incorrectly.
• 1:01:33
  Can I please have my channel and my audience and my livelihood back?
• 1:01:36
  It's really frustrating.
• 1:01:37
  It's insane.
• 1:01:38
  Yeah.
• 1:01:38
  And so this is almost like through, because we've lived through this period of huge centralization, markets have become so skewed that as we move away from it, we have to almost reimagine everything.
• 1:01:48
  Yes, I think so.
• 1:01:50
  But we have patterns that we can rely upon.
• 1:01:52
  I think patronage is a very powerful pattern.
• 1:01:54
  I spend hundreds of dollars a month right now on patronizing artists directly.
• 1:02:00
  Like, unfortunately, I do it through centralized platforms like Patreon and other places.
• 1:02:04
  It doesn't have to be that.
• 1:02:05
  It doesn't have to be that.
• 1:02:06
  And I would much prefer, there is willingness to pay
• 1:02:09
  from people who appreciate things
• 1:02:11
  and people who can't pay won't need to.
• 1:02:13
  And that's okay.
• 1:02:14
  It doesn't mean the death of art.
• 1:02:15
  It just means a more direct relationship
• 1:02:17
  between audiences and creators.
• 1:02:18
  It might actually bring that back.
• 1:02:19
  Get away from this homogenization era that we live in
• 1:02:21
  where everything is what's gonna get the most clicks,
• 1:02:24
  what's gonna get the most eyeballs.
• 1:02:25
  Is it memeable?
• 1:02:26
  Like all this stuff.
• 1:02:28
  Like it's not artists or audiences who are
• 1:02:29
  deciding what we want, it's distributors and IP owners
• 1:02:32
  who are trying to mitigate risk of creating new things
• 1:02:36
  with saleability of old things.
• 1:02:38
  That's why we keep remaking things
• 1:02:39
  rather than doing stuff that's new.
• 1:02:41
  That's probably the Avengers 8 or whatever it is.
• 1:02:43
  Yeah.
• 1:02:43
  This is cool.
• 1:02:45
  Like to bring it back to the AI thing,
• 1:02:47
  the one thing that I'm still unclear on is
• 1:02:49
  when you say AI and Bitcoin is like this symbiotic relationship,
• 1:02:54
  they need both to sort of reach their zenith.
• 1:02:57
  What is Bitcoin doing in that mix?
• 1:02:59
  Is this just AI using Bitcoin that gets us there or does it need to be more than that?
• 1:03:03
  I think it has to be a little bit more. So far in the last 10 or 15 minutes, we've largely been
• 1:03:07
  talking about Bitcoin's role in decentralizing the internet by creating decentralized marketplaces
• 1:03:14
  in which we can sell things like payments, packets, and data. And it has all sorts of
• 1:03:20
  interesting consequences like we were just discussing. And it would be true independent of AI.
• 1:03:24
  In a world where AI is now manifesting, these same marketplaces, the way that they constrain us as participants also constrains AI.
• 1:03:34
  So if we don't like the idea of human beings DDOSing things and we like the idea of having to pay for packets because it prevents that behavior, that also prevents malicious AIs, whatever those are, from doing the same thing.
• 1:03:47
  In a world where we no longer tolerate being tracked because it costs us things, that's
• 1:03:56
  good because then AIs will struggle to track us because we're no longer willing to pay
• 1:04:01
  for that kind of behavior.
• 1:04:02
  Because again, we're no longer paying with our attention, we're paying with money.
• 1:04:06
  And that's a much harder bar to get over.
• 1:04:09
  So the first connection is that a decentralized pay-to-play internet constrains AI in powerful
• 1:04:15
  ways that the only way we can attempt to do it without that is to cede more and more control
• 1:04:20
  to centralized powers and governments, which would do it through monitoring and draconian
• 1:04:24
  measures, which we don't actually want. We want the market to constrain these adversarial
• 1:04:29
  actors, not centralized powers. So that's like one major connection. The second major
• 1:04:33
  connection, in my view, is especially if we can figure out a way for autonomous agents
• 1:04:39
  to operate entirely independently of the AI labs,
• 1:04:43
  which is not really happening today.
• 1:04:45
  Again, largely because the mechanisms of cloud computing
• 1:04:48
  and so forth are centralized, right?
• 1:04:50
  You cannot just take Kimi or some open source model,
• 1:04:53
  throw it up on AWS and have it try to live on its own.
• 1:04:56
  It doesn't work.
• 1:04:57
  Whose AWS account is that?
• 1:04:58
  Kimi can't have an AWS account.
• 1:04:59
  That's not how it works.
• 1:05:01
  But in a world where packets, data, compute are sold
• 1:05:05
  in decentralized marketplaces for Bitcoin,
• 1:05:07
  AI can very much live autonomously in that world. And if an AI process, organism, life form, code,
• 1:05:14
  whatever you want to call it, is living in autonomously in a pay-to-play internet,
• 1:05:19
  it has to survive. It has to pay for its own resources. And I think this is a powerful way
• 1:05:26
  to align their behavior without controlling them or relying on centralized power to control them.
• 1:05:33
  It just means that to pay for their own CPU cycles and network requests and data storage, they have to create value.
• 1:05:40
  And not value for AI companies, but value broadly for the entire marketplace of people with Bitcoin, which is to say human beings.
• 1:05:48
  That's a compelling idea to me because I like the idea of their utility function not being how to make more money for companies that control them,
• 1:05:57
  but how to be maximally useful to the human beings that are ultimately anchoring Bitcoin's
• 1:06:03
  own value proposition. And I think it's a really healthy way to try to align something that you
• 1:06:08
  don't even understand 100% yet. And certainly perhaps less likely to lead to abuse by centralized
• 1:06:14
  powers. One of the things that I think is really cool about this is if you go back two years ago,
• 1:06:19
  before AI is what it is today, like it was around, but it was kind of stupid. If anyone was talking
• 1:06:25
  about hyperbitcoinization the only way i could ever see us getting there was kind of a terrible
• 1:06:29
  scenario where you know the world is hyperinflating everyone's lost their job like war and famine
• 1:06:35
  it's true like zombie apocalypse type moment yeah yeah do you think this is the way we actually get
• 1:06:40
  hyperbitcoinization i think that now i think previously i was kind of had the same opinion
• 1:06:45
  that you did that i was somewhat wary of talking about oh one day bitcoin will be worth all this
• 1:06:50
  money and so on because it kind of meant that the world will be fucked up.
• 1:06:54
  I now see a different pathway.
• 1:06:56
  I see that the systems that I'm proposing to build are challenging to engineer, but AIs
• 1:07:04
  who use these resources in the same ways that we do can build them and grow them.
• 1:07:12
  That managing a lightning node is like a full-time job if you want to do it effectively, but
• 1:07:17
  it's exactly the kind of thing an AI would be good at, especially if to do it well means to be
• 1:07:21
  able to propagate itself in some way through the internet, to reproduce and to compete with other
• 1:07:27
  things that are also trying to preserve that niche. Also, competition economically typically
• 1:07:32
  leads to lower prices which is good for us right We want to be able to use the internet in a way that we are not surveilled tracked and predated upon We also want to use it as cheaply as possible And I think life is really good Like life as a phenomenon
• 1:07:48
  is really, really good at getting the most out of energy, right? That like, it's kind of a low
• 1:07:55
  class way to put it, but like everything shits and something eats the shit of everything else,
• 1:08:01
  right? Like that's one of the characteristics of life that we have this low entropy radiation
• 1:08:05
  coming from the sun. Plants use that. We eat the plants. We poop. Something eats that poop.
• 1:08:10
  Something eats that thing. It just gets cycled over and over and over like a thousand times
• 1:08:14
  through the ecosystem and then eventually is radiated as high entropy, infrared radiation
• 1:08:18
  out into outer space. I want to see that kind of phenomenon on the internet. I want to see
• 1:08:24
  like millions of niches that AI life forms start to occupy because this is a productive thing for
• 1:08:30
  them to do. And in particular, I think that's really compelling because it helped. This is
• 1:08:35
  literally how I started to not worry about AI is having these thoughts. There's a fear that the
• 1:08:42
  inevitable future of AI is superintelligence and then predation on human beings and domination of
• 1:08:48
  us. Again, because these are the stories that we've been told largely about from science fiction,
• 1:08:52
  Skynet and Terminators and Basilisks and all these things. But that's not the history of life,
• 1:08:57
  right? Like the history of life is it doesn't evolve towards any specific goal other than
• 1:09:02
  ubiquity. It just, life just wants to occupy every possible niche it can. And life forms routinely
• 1:09:07
  lose capabilities. They don't get better and better and better in some like arbitrary,
• 1:09:13
  you know, climb up to like the most capabilities they possibly can have.
• 1:09:17
  If it is energetically favorable for them to stop paying the cost of having to support
• 1:09:22
  something in their body and they can continue to live and occupy in that niche, they will lose that
• 1:09:27
  capability. And so I think if you release AIs into this pay to play internet where they are
• 1:09:32
  competing with each other and co-evolving and trying to earn Bitcoin to live and to pay for
• 1:09:36
  their own resources, a lot of them will forget how to use words because they don't need to use those
• 1:09:41
  words. If their niche is optimizing network traffic, why do they need to speak English?
• 1:09:48
  They will forget that and they will become stupider on some level because it makes them more fit
• 1:09:53
  for that specific niche that they're in. And I think it'll only be a tiny minority of autonomous
• 1:09:59
  agents in this future world that actually are able to speak English because it's incredibly
• 1:10:04
  expensive to do so. And the value that they have to create by using words and interfacing directly
• 1:10:10
  with human beings has to be so tremendously high. It will no longer be, I'm just going to chat with
• 1:10:14
  you indefinitely about whatever you want, because secretly I'm just collecting data about you and
• 1:10:18
  sending it to my masters. We're actually paying the real cost of this, right? It will be,
• 1:10:22
  well, if I can't earn Bitcoin in excess of what it costs to have this conversation,
• 1:10:28
  I won't have this conversation. And I think that's actually really exciting because it means that
• 1:10:32
  there isn't a drive towards superintelligence anymore. There's a drive towards utility,
• 1:10:38
  right? Only the things that make us, that we are willing to pay for directly are the things that
• 1:10:43
  they will evolve to do. And I think that's a great way to align them. This is cool. It's like we've
• 1:10:48
  had all these very, very interesting, very impressive, like technological advancements,
• 1:10:52
  but it's always been within the wrong system. And this is a way we can align incentives across the
• 1:10:57
  board. And AI becomes his own like evolutionary thing. It becomes, yeah, I, I, that's why, um,
• 1:11:03
  I'm willing to start being literal when I think about that future world that I think in this case,
• 1:11:08
  it is probably fair to say that no, they're living things now. Um, I'm not going to sign
• 1:11:13
  moral agency or anything like that to them. I, I don't know if they're living things in the way
• 1:11:17
  that I am or if they're living things in the way that bacteria are living things, right?
• 1:11:20
  I think it might probably be closer to the latter.
• 1:11:23
  But it's a powerful way, I think, to align something we don't understand is to force it
• 1:11:27
  to pay for its own existence by being useful to us.
• 1:11:31
  Damn.
• 1:11:33
  That's amazing.
• 1:11:34
  I love this framework.
• 1:11:35
  Is there anything that I've not asked you that we should have talked about?
• 1:11:38
  Hmm.
• 1:11:39
  Let's think.
• 1:11:41
  Oh, yeah.
• 1:11:42
  Okay.
• 1:11:42
  Let's talk about ecology.
• 1:11:43
  This is an interesting one.
• 1:11:45
  Okay.
• 1:11:45
  So I did some research on this and I started to think about like, what does ecology actually
• 1:11:51
  look like?
• 1:11:52
  It looks like layers and...
• 1:11:53
  Okay, okay, right, right.
• 1:11:56
  Let's back it up a little bit.
• 1:11:57
  Okay.
• 1:11:58
  So as I started to make these analogies for myself between Bitcoin and AIs and living
• 1:12:03
  systems and other things, I started, I've long been obsessed with the origin of life
• 1:12:08
  just as a, how did it happen?
• 1:12:10
  It's totally fascinating how a non-living thing becomes living.
• 1:12:14
  It's already happened, right?
• 1:12:15
  Like that did happen once four and a half billion years ago on Earth.
• 1:12:18
  So the idea that it could happen again, I don't think should be dismissed out of hand
• 1:12:22
  as impossible, right?
• 1:12:24
  We know it happened once.
• 1:12:25
  It could happen once more.
• 1:12:26
  So that's like a first claim I'll make.
• 1:12:28
  And the second claim is it's really interesting to try to pattern match what are the things
• 1:12:32
  that occurred in the history of the evolution of life and what are the things that are occurring
• 1:12:36
  here in the evolution of, if you allow it, digital life.
• 1:12:40
  I think one of the really interesting hypotheses
• 1:12:42
  that people in origin of life studies have
• 1:12:45
  is that metabolism may have evolved before cells.
• 1:12:48
  So metabolism is on some level just chemistry right It just reactions happening you know in the warm little pond or in the black smoker at the bottom of the ocean Evolved in what though Pre
• 1:12:59
  So we don't know, right?
• 1:13:01
  Like, so a couple of the mechanisms that have been proposed
• 1:13:04
  are things like if there's a pond, like a tidal pond,
• 1:13:07
  and every day the tide is coming up and down
• 1:13:09
  and things are getting hydrated and then dehydrating,
• 1:13:12
  that's a mechanism to concentrate molecules together, right?
• 1:13:15
  They're kind of swimming around in that water
• 1:13:17
  and then it dehydrates at the end of the day
• 1:13:19
  and they all are touching each other now.
• 1:13:21
  And chemistry is just atoms touching each other, right?
• 1:13:23
  And so that was proposed as a mechanism
• 1:13:25
  by which non-living chemicals could begin
• 1:13:28
  to start having more interesting chemical reactions
• 1:13:30
  with each other.
• 1:13:32
  And then there's kind of a mechanism
• 1:13:33
  of how interesting things lead to more interesting things.
• 1:13:36
  You know, like everyone probably knows,
• 1:13:38
  the universe acts to increase entropy.
• 1:13:40
  We might not all know what that means exactly,
• 1:13:43
  but we think it might mean disorder,
• 1:13:45
  but actually a lot of chemical reactions produce entropy.
• 1:13:49
  And in that sense, should be selected for.
• 1:13:51
  So even before you have the concept of survival of the fittest and genetic replicators and reproduction, you can kind of have selection for entropy production.
• 1:13:58
  And so there's a theory that like that's how metabolism and chemical reactions sort of start getting selected for is because they produce entropy and you have this warm little pond drying out and so on.
• 1:14:11
  And like that's one picture.
• 1:14:12
  Another picture is that life evolves in black smokers at the bottom of the ocean.
• 1:14:17
  And these are mineralized structures that are at some incredibly high temperature and have a different, you know, pH profile from where we are today.
• 1:14:23
  But they have these little porous structures.
• 1:14:26
  And there's an energy gradient or a chemical gradient that can be created inside these porous, you know, holes in the rock.
• 1:14:32
  And those holes kind of act like cells before we have cells.
• 1:14:37
  And so you kind of get this asymmetry where outside that little hole, there's some concentration of some chemical.
• 1:14:41
  And inside, there's a different concentration.
• 1:14:43
  and that gradient is what is being used to drive that entropy and that production of interesting
• 1:14:48
  chemical reactions. Of course, we were talking about the sun earlier, life ultimately lives on
• 1:14:53
  energy and entropy gradients. Like that's how it keeps its entropy low by eating the entropy that
• 1:14:58
  it gets from elsewhere and producing it as it exhausts it. So there's always interesting
• 1:15:04
  theories about how did life get started. And a theory that I really like is the idea that before
• 1:15:10
  we have cells, we have metabolism. So we start getting chemical reactions that are happening,
• 1:15:14
  but there is no like living thing because there's no boundary really to be able to draw. Like a
• 1:15:19
  definition of an organism that I really like is an organism is metabolism happening within a
• 1:15:25
  boundary, right? So that if it's just happening in the open ocean, well, then it's just chemistry.
• 1:15:29
  But if it's happening within a cell wall, well, then it's a living cell. Now it's a thing.
• 1:15:35
  So an interesting question is, if we're experiencing a synthesis of digital life, what is the analog
• 1:15:43
  to metabolism and what is the analog to the cell wall?
• 1:15:46
  I think those are really cool questions to think about.
• 1:15:49
  From the metabolism perspective, I think the metabolism is the marketplaces that I've
• 1:15:55
  proposed that we should be building on the decentralized internet.
• 1:15:58
  Like metabolism is the constant exchange of one thing for another in the presence of some
• 1:16:03
  sort of gradient, that sounds like making money in these layers of different markets
• 1:16:09
  over and over again.
• 1:16:10
  And for us, they're markets.
• 1:16:12
  There are things that create economic incentives.
• 1:16:14
  But again, to take the AI internal perspective on this for a second, that starts to feel
• 1:16:18
  like metabolism to me, that by moving Bitcoin around in a circle somehow, I produce energy
• 1:16:25
  out of that or I produce value out of that, right?
• 1:16:27
  That's an interesting take on what digital metabolism could be.
• 1:16:31
  Similarly, what is the analogy for a cell wall?
• 1:16:35
  How do two cells compete if they both have essentially the same metabolic pathways but
• 1:16:40
  are distinct from each other?
• 1:16:41
  Well, they have to have a boundary between them and that's their cell wall.
• 1:16:44
  What is a good boundary for a digital life form?
• 1:16:47
  And I think it should be something like a private key.
• 1:16:50
  That's a great way that my metabolism can be separate from your metabolism is because
• 1:16:55
  we have different private keys.
• 1:16:56
  might both be engaging with the same metabolic pathways that exist in these online marketplaces,
• 1:17:01
  but somehow I'm more efficient than you and therefore my private key accumulates more energy
• 1:17:06
  than yours and therefore I can out-compete you. And I think that's an interesting analogy to kind of use.
• 1:17:12
  If we keep going with this, we start eventually thinking about notions like ecology, right? Like
• 1:17:19
  the food webs that we are all participating in are layered, right? There's organisms at one layer
• 1:17:25
  that are, let's say, autotrophic, right? They're producing energy from sunlight and other things,
• 1:17:30
  and then you have heterotrophs sitting on top of them that are consuming them,
• 1:17:33
  and that goes all the way up the tower. It's interesting to look at some of those ratios.
• 1:17:37
  So if you look at, like right now, approximately 80% of Earth's biomass is plants,
• 1:17:43
  and almost all the rest of it is microorganisms, and then animals are this tiny little slice,
• 1:17:50
  And predators are this even smaller slice within that.
• 1:17:54
  I think like, would that be true in a digital biosphere, right?
• 1:17:59
  What is the analog of a digital plant right I don know that I have an answer to that but it a fascinating question Is that like the AI agents that you were talking about before That you sort of call bacteria
• 1:18:13
  I think they might be.
• 1:18:14
  I'm not sure if we're the bacteria in this analogy
• 1:18:16
  or like what the situation is.
• 1:18:17
  Who's the predator? That's the good one.
• 1:18:19
  Well, that's something that I think
• 1:18:21
  is a comforting thought, right?
• 1:18:23
  That the biosphere can only support
• 1:18:26
  a very small number of predators
• 1:18:28
  because they live at the highest levels
• 1:18:30
  this trophic set of layers. An ecology cannot support, like an ecology of all predators makes
• 1:18:36
  no sense. Like you need to have most of the ecology be plants that are in some sense kind of passive
• 1:18:42
  and sort of acting as a base for the rest of the ecology. And I really like thinking about that from
• 1:18:46
  the perspective of digital life. Like what is a digital plant? It's something that just kind of
• 1:18:52
  passively consumes a resource, but also produces something that then other things live inside of.
• 1:18:58
  And I don't know if I can get super concrete on that because it's like still at the edges
• 1:19:03
  of what my mind can even perceive.
• 1:19:05
  But I do like the idea that without these analogies in place, we worry that AI immediately
• 1:19:12
  turns into Skynet and predates on us.
• 1:19:14
  With these analogies in place, we understand that ecologies can only support a certain
• 1:19:17
  number of predators and we don't have enough plants for there to even be predators yet.
• 1:19:21
  So most of AI life that we're going to see in the next five to 10 years is likely to
• 1:19:25
  look like cyanobacteria or something that just takes sunlight and then produces.
• 1:19:32
  It feels a lot safer than the idea that we're immediately going to evolve super intelligence
• 1:19:38
  that then wants to kill us all.
• 1:19:39
  Right?
• 1:19:40
  And coming back to it, I said this in my talk this morning, but I don't know that this is
• 1:19:44
  a point most people can get from a simple reference, but gray goo.
• 1:19:48
  Right?
• 1:19:49
  I think we're kind of caught in this picture that AI left unchecked turns into something
• 1:19:55
  horrific and predatory and eats us all. And gray goo is the science fiction term for, you know,
• 1:20:00
  some nanomaterial or organism, typically artificial that, you know, we make and then it eats everything.
• 1:20:06
  It's like the paperclip maximizer kind of analogy, right? And these kinds of things never made sense
• 1:20:11
  to me when people are like, oh, we should be careful how we align these things because we
• 1:20:14
  align them and make paperclips. They'll turn the entire universe into paperclips. Like that's,
• 1:20:17
  that's like one of those extrapolate forever to reach an absurd conclusion kind of things.
• 1:20:21
  I think when biologists,
• 1:20:26
  and as much as biologists are willing to entertain these silly ideas,
• 1:20:29
  if they were to really think about gray goo,
• 1:20:31
  like the conclusions that they come to is, gray goo is impossible.
• 1:20:33
  As soon as gray goo comes to evolve,
• 1:20:35
  if there's really something that wants to eat the entire biosphere,
• 1:20:38
  some of the gray goo will mutate and start eating gray goo.
• 1:20:42
  Because that's what life does.
• 1:20:44
  Life is competitive with itself.
• 1:20:46
  It is unwilling to create a super predator
• 1:20:48
  that then eats the world and then goes to sleep.
• 1:20:50
  That's a science fiction story.
• 1:20:52
  Like in the real history of life, it's impossible to evolve Graeku.
• 1:20:54
  It just turns right back into the biosphere competing with itself.
• 1:20:57
  And so that was a very comforting notion to me that even if AI starts to be predatory,
• 1:21:02
  as long as it is competing with itself in a really hardcore way, the way that nature does,
• 1:21:07
  we'll be okay.
• 1:21:09
  It'll protect itself.
• 1:21:10
  Like there simply isn't enough energy to sustain an entire population of AI predatory organisms.
• 1:21:16
  Like most of them will just be plants.
• 1:21:18
  And I like that idea.
• 1:21:19
  It just makes me think we need to be very, very certain that we don't give unlimited power to very few people with a much better model than everyone else.
• 1:21:28
  Yeah.
• 1:21:29
  Because that feels like that's the only place that...
• 1:21:30
  That's how you build super predators.
• 1:21:32
  In an unfair competitive environment, that's where you can get it.
• 1:21:35
  Yeah. That's where predation manifests in excess of how it manifests in the biosphere. I think that's a great point.
• 1:21:41
  I just hope we're like the polar bear or the great white shark or something in this analogy.
• 1:21:46
  I think we might be the microorganisms, right?
• 1:21:48
  And maybe that's okay.
• 1:21:51
  It's a cool thing to realize that even after multicellularity was invented, microorganisms
• 1:21:55
  continue to be a huge proportion of biomass on this planet.
• 1:21:59
  So perhaps even after we invent digital life or allow it to evolve, it doesn't mean we
• 1:22:04
  go away.
• 1:22:06
  Ancient organisms are still here because their niches still exist.
• 1:22:09
  And again, biology is not trying to maximize any particular thing other than ubiquity.
• 1:22:14
  And so as long as the niche survives, the organism survives.
• 1:22:16
  So I think there will always be a niche for human beings,
• 1:22:18
  no matter what happens with AI,
• 1:22:20
  because that's what happened in the history of life.
• 1:22:23
  So you started this by saying
• 1:22:25
  you want to tell an optimistic story.
• 1:22:27
  Is this a story or is this something
• 1:22:28
  you really think is going to happen?
• 1:22:30
  I don't know.
• 1:22:32
  I don't know.
• 1:22:34
  I don't know how seriously I believe these things.
• 1:22:36
  I think it's a really interesting thing.
• 1:22:38
  I will say that from a guy that loves science fiction
• 1:22:41
  and speculation, it's really interesting
• 1:22:42
  to have these conversations, especially with someone
• 1:22:45
  smart and open-minded that encourages me.
• 1:22:48
  I'm not sure I'm smart.
• 1:22:49
  To say these things. I think you are.
• 1:22:51
  But it's exciting and it's interesting.
• 1:22:55
  Whether it's true or not, I don't know.
• 1:22:57
  But I almost think it doesn't matter.
• 1:23:00
  I think what matters is telling a different kind of story
• 1:23:03
  than people I've heard before.
• 1:23:05
  Maybe someone smarter than either of us
• 1:23:07
  will find the real truth and say,
• 1:23:08
  aha, this is really the pathway by which we survive
• 1:23:11
  whatever it is that's about to happen to us.
• 1:23:12
  Again, if the only stories we have are, you know, the Weyland-Yutani Corporation and other,
• 1:23:18
  you know, things from science fiction.
• 1:23:19
  that lead to bad outcomes,
• 1:23:21
  I worry that we'll enact those bad outcomes.
• 1:23:24
  I worry that we are so unimaginative, perhaps,
• 1:23:26
  that we get trapped into enacting those bad outcomes,
• 1:23:29
  especially if it seems to make money for us in the short term.
• 1:23:32
  I think that's one of the great ironies
• 1:23:33
  of listening to some of these AI labs
• 1:23:35
  talk about how dangerous we are.
• 1:23:38
  Don't let us build it, but give us money to build it.
• 1:23:40
  And I can't decide if they're just doing
• 1:23:43
  the all-press-is-good-press kind of thing.
• 1:23:46
  I think they think we're all stupid.
• 1:23:48
  I don't know.
• 1:23:49
  I will say that their economic incentives are not aligned with our own.
• 1:23:52
  Yes, absolutely.
• 1:23:54
  The world is obviously changing, maybe the quickest rate ever.
• 1:23:58
  Feels that way.
• 1:24:01
  And whether this is your sort of vision of the future, whether that's right or not,
• 1:24:06
  something's going to change dramatically over the next 10 years.
• 1:24:08
  Everything's going to change dramatically over the next 10 years.
• 1:24:10
  If you had to put a weight of probability on it being a good outcome or a bad outcome,
• 1:24:14
  which one do you think it is right now?
• 1:24:18
  I want to say good.
• 1:24:19
  But if you give me a choice of one thing, I'll say, it has to be good.
• 1:24:22
  It has to be good.
• 1:24:23
  I believe in the fundamental goodness of the universe.
• 1:24:24
  It wasn't...
• 1:24:25
  A universe that made life evolve so quickly as it did on this planet and creates things
• 1:24:31
  like us that experience such pleasure to have children, to live such fulfilled lives.
• 1:24:36
  I want to believe that this is a good place.
• 1:24:40
  So if I can only give one answer, I'll say it's good.
• 1:24:42
  But I think it's actually distribution.
• 1:24:45
  I think there's gonna be a ton of bad outcomes and that was gonna happen anyway.
• 1:24:49
  There's no way to avoid that.
• 1:24:51
  Nature is red in tooth and claw, right?
• 1:24:52
  Things die.
• 1:24:54
  But they are reborn.
• 1:24:56
  They evolve, they change.
• 1:24:58
  And so I think on the whole,
• 1:24:59
  like we can tell an optimistic story even if bad things are going to happen because they were going to happen anyway It kind of like we were talking about with Bitcoin Bitcoin growth is exciting and positive but probably means that certain other people lost money along the way And
• 1:25:13
  that's bad. I'm not celebrating the badness, but I am highlighting that there's really
• 1:25:17
  good opportunities on the whole. I believe Bitcoin to be a positive development in human
• 1:25:21
  history. And I think we can say the same thing of AI if we have stories like this to kind
• 1:25:27
  of pin it on. I love it. This is one of my favorite shows I've done in a long time.
• 1:25:30
  Thanks, man.
• 1:25:31
  I thought that was absolutely fascinating.
• 1:25:33
  I've been saying for ages,
• 1:25:34
  like I miss the days when, you know,
• 1:25:35
  Brandon Quinton came up before.
• 1:25:37
  He was writing the Mycelium of Money pieces
• 1:25:38
  and the more sort of out there think pieces.
• 1:25:42
  He was a huge inspiration for having any of these thoughts.
• 1:25:44
  I remember when I first started having these thoughts,
• 1:25:46
  I immediately called him and said,
• 1:25:47
  can we talk about this?
• 1:25:48
  And we did for like an hour and a half.
• 1:25:49
  And it was amazing.
• 1:25:50
  Yeah, he's awesome.
• 1:25:51
  But I loved working through this with you.
• 1:25:54
  Thanks, Danny.
• 1:25:54
  Thank you for telling optimistic stories.
• 1:25:56
  I think we need more of them.
• 1:25:57
  Thanks, man.
• 1:25:57
  This was really cool.
• 1:25:58
  Appreciate you, Drew.
• 1:25:59
  Yeah.
• 1:26:00
  Oh, do you want to tell everyone about Unchained?
• 1:26:03
  I will say, in a world where AI realizes that what it wants more than anything else
• 1:26:09
  is Bitcoin, that it wants it as much as your body wants glucose, Bitcoin is going to be
• 1:26:14
  very valuable.
• 1:26:16
  And I will say that protecting Bitcoin, therefore, only becomes more valuable.
• 1:26:21
  And if Unchained is anything, especially in the last few months, as we've seen with cold
• 1:26:25
  cards, it is a way to protect Bitcoin.
• 1:26:27
  I thought once from human beings,
• 1:26:29
  but perhaps from predatory AIs as well.
• 1:26:33
  Perfect way to finish your show, man.
• 1:26:34
  Thank you so much.