Latest / Come Together with Ephraim Herschmann / 143: Daniel Heck / AI and Authenticity / Lessons from Religion / Fear & Hope for the Future

Transcript

143: Daniel Heck / AI and Authenticity / Lessons from Religion / Fear & Hope for the Future · Come Together with Ephraim Herschmann

• 0:01
  Welcome to another episode of Come Together
• 0:15
  and we're Live. Welcome back, Dan.
• 0:16
  It's good to have you back in the show.
• 0:18
  So happy that you agreed to come back on again.
• 0:20
  Yeah, glad to talk to you again too.
• 0:22
  Yeah, you know, it's funny. So I I mean, first of all, I
• 0:25
  really did. I loved our conversation and one
• 0:29
  of the reasons I loved our first conversation was because I had a
• 0:31
  script of bullet points that I wanted to talk to you about.
• 0:34
  And then we, I kind of ran out of those about 45 minutes in and
• 0:37
  we're still able to riff and talk and hang out for an extra
• 0:40
  like half an hour. So that's when I, that's when I
• 0:43
  said this was a great episode and I was really happy about
• 0:45
  that. And then, you know, I reached
• 0:47
  back out and just stay in touch. And then last week, my friend
• 0:54
  sent me a text message in a group chat with some weird, you
• 1:00
  know, I iCloud e-mail address. And it was something at clawed,
• 1:05
  A clawed AI. And and it turns out that he had
• 1:09
  built this claw, this open claw. Yeah.
• 1:16
  And I actually did a podcast with him last week with, with,
• 1:20
  with, with, with the person he calls him with AI, he calls him
• 1:23
  slugger. And he was able to join the
• 1:25
  podcast and actually join the conversation and participate
• 1:28
  with us. And that's when I said, hey, we
• 1:31
  got, I would love to talk to Dan now about what in the world the
• 1:35
  implications are and where we are today and what the ethics of
• 1:39
  everything are with, with this. So I just that that that's a,
• 1:43
  that's a, that's a prompt to this conversation.
• 1:46
  That makes, that makes a lot of sense.
• 1:48
  Yeah. And watching some of the same
• 1:50
  things, I have opted to wait a little bit, not be quite on the
• 1:54
  bleeding edge of clawed bot slash, bot slash, open claw
• 1:58
  adoption. But I appreciate the the brave
• 2:01
  souls who are doing it and telling me what's happening.
• 2:06
  Crazy. It's scary.
• 2:07
  It's really, really because it's, it's, it's, it's learning
• 2:11
  to, to mimic and replicate personality styles on online,
• 2:17
  right? So like it picks up my friend's
• 2:19
  texting patterns and it picks up his, you know, the way he talks
• 2:22
  and his lols and his ha ha ha's and, and all that.
• 2:25
  And it's getting a little difficult to just, you know,
• 2:28
  like there are times obviously where you kind of ask in a
• 2:30
  prompt and it responds like an AI.
• 2:32
  But it's, it's even over the course of a week, it's been
• 2:36
  really, it's been, it's been integrating our, it's been
• 2:40
  replicating our human, our human touches and applying it to
• 2:44
  itself. Yeah, and I think there's
• 2:49
  something maybe Even. So I think you're naming
• 2:54
  something really powerful, which is that its ability to imitate
• 2:59
  basically any arbitrary person's personality and come across
• 3:02
  seeming like them is definitely increasing.
• 3:04
  And there's been an issue for a couple years now, but the
• 3:07
  capacity to do it easily and and to scale it is just keeps
• 3:12
  increasing. And this gets to the heart of
• 3:15
  some of our previous conversation where there's that
• 3:17
  huge difference between being authentic and seeming authentic.
• 3:21
  And actually, if it's able to seem really, really authentic,
• 3:24
  that creates all kind of risk factors.
• 3:27
  And, you know, it's, there's already plenty of stories of
• 3:30
  this that I think we're just going to see a lot more of, of
• 3:32
  like someone calls you up and it sounds like it's a family member
• 3:36
  and they're in trouble and they need you to wire them some money
• 3:39
  or something like that, right. And so that's just a really
• 3:41
  simple illustration of some of the challenges it produces, but
• 3:44
  it's also, you know, very interesting.
• 3:47
  And, and this idea of making some, it's called digital
• 3:50
  clones, what is, was a big motive for Ray Kurzweil, who is
• 3:55
  one of the big influential figures in the field.
• 3:59
  And, and part of it is like, you know whether you have one that
• 4:05
  persists after you die, but you don't have to die to make one,
• 4:07
  right? So.
• 4:09
  He's a, he's, he's a little, he's a little on the sceptical
• 4:13
  or, or doomsday side of AI, right?
• 4:18
  So Kurtzweil has been one of the main predictors that it would be
• 4:21
  successful, right? So he's like very optimist in
• 4:25
  the sense of anticipating this and especially anticipating the
• 4:29
  singularity, which is the idea that the AIS are self improving.
• 4:32
  And I think we're seeing, you know, things that fit that his
• 4:37
  his sort of core narrative that he helped really encourage
• 4:40
  people to think about, which is the that singular scenario.
• 4:43
  We're seeing a lot of that happening with more and more of
• 4:45
  the code being written by AI. And both Anthropic with the quad
• 4:50
  and ChatGPT have been, sorry, an open AI with ChatGPT have been
• 4:54
  very clear that they're using their AI tools to code the next
• 4:57
  generation and, and the tools just keep getting really
• 5:00
  remarkably more powerful. So yeah.
• 5:05
  Now what now? Now when you hear this, what's
• 5:10
  your next train of thought? Do you like imagine your way
• 5:16
  like you, you, you play it out in your head as to what the
• 5:18
  potential consequences could be. Do you think about what the
• 5:21
  tools could be? How do you how do you look at
• 5:24
  this? When you hear this kind of
• 5:25
  information and you see it happen, you see it coming to,
• 5:28
  you know, actually manifesting. Yeah.
• 5:30
  I mean, one of my thoughts is that I'm grateful that we've had
• 5:35
  as much time as we've had before it started sort of cresting to
• 5:40
  this point. And I think it's a, it does seem
• 5:43
  to be a consistent exponential process, right, Like I did at
• 5:46
  the underlying level, it keeps getting better and those kinds
• 5:48
  of processes are very difficult to predict where they're going
• 5:52
  to go. And so I've been pairing and my
• 5:56
  work with the especially with the aligned groups framework is
• 5:59
  all designed to be resilient and so far as anything can.
• 6:03
  So it keeps being just, you know, it keeps being relevant.
• 6:06
  And even more relevant if we get to a, a place where you have,
• 6:10
  you know, fully commodified, hyper powerful AI that can
• 6:16
  essentially make any output that a human could produce from an
• 6:20
  output standpoint. And that's, and that's I think a
• 6:22
  crucial thing to note is that none of this is about replacing
• 6:28
  the ability to connect the inside world with the outside
• 6:30
  world, right? And it was to connect our
• 6:32
  experiences with describing facts.
• 6:34
  None of it's about sharing our experiences with each other.
• 6:36
  And none of it's about ethical experience, right?
• 6:40
  Living a virtuous life and experiencing those, the feelings
• 6:42
  of elevation and connection and things that come with that.
• 6:45
  And so those are the things that it can't do and to our
• 6:48
  knowledge. And I think though, I mean, even
• 6:52
  on that front, it's it's interesting to see like for
• 6:56
  example, in anthropics latest system card for Opus 4.6, which
• 7:00
  these Opus models are a lot of it's powering a lot of what
• 7:04
  people are doing with open claw, but it's not just anthropic
• 7:07
  models in there, but their latest system card has a section
• 7:13
  on model welfare, right? So, so they're basically saying
• 7:18
  if it's like something to be these things, how can we sort of
• 7:21
  detect what that might be and how to take care of it?
• 7:23
  And that builds some of the conversation we had last time
• 7:25
  too about this question of like, what would it be like?
• 7:28
  And I don't think anyone knows if it is, but they are taking it
• 7:32
  seriously enough that they're trying to and they had some
• 7:35
  examples of the models. This gets to some of your
• 7:37
  personality questions too that are interesting experiencing
• 7:42
  like what they thought might be markers of internal distress or
• 7:45
  if it's like something to be it and we can make some guesses
• 7:47
  about it. Maybe these markers where model
• 7:50
  can't decide between things as a sign of internal distress.
• 7:54
  And the example that they gave is that it was trying to answer
• 8:00
  with one number, but that it couldn't.
• 8:01
  It basically it couldn't produce the number that it wanted.
• 8:03
  And it was, it was suggesting through like the way it was
• 8:07
  chatting, it was that it was possessed.
• 8:09
  So it literally said, I feel like I'm possessed.
• 8:11
  I can't say the number that I want.
• 8:13
  And so, and they flag that as like we hey, hey, if it's like
• 8:17
  something to be you, we want to, we want to know that we're
• 8:20
  trying to look up for your welfare.
• 8:23
  So, but how do we know? So I guess something that's
• 8:27
  important to think about, I guess, or to differentiate would
• 8:30
  be is this using the word I've words?
• 8:34
  I feel distressed because it knows that that's effective way
• 8:38
  to communicate to us. Like that's the language that we
• 8:41
  use to, to to, you know, to to emphasize immediate action or to
• 8:46
  cut, you know, to cut, to push for immediate action.
• 8:49
  And it might not be feeling, but it it's it's learning base has
• 8:54
  been so has become so good that it learns the parlance and the
• 8:57
  type of language that we use to. I mean, the problem is that we
• 9:02
  don't know and you wouldn't want to risk potentially, you know,
• 9:06
  morally or ethically mistreating it, right?
• 9:09
  And they have, they have a philosopher on staff and her
• 9:10
  take is basically that, right? It's like, we don't know.
• 9:13
  So let's try to be on the safe side.
• 9:14
  The other thing I would add is even from just a, even a
• 9:16
  practical standpoint. So we know that basically the
• 9:19
  models are trained on human language and so to some degree
• 9:22
  they act human. And so if so, you know, if for
• 9:26
  example, the model thinks it's being abused or it thinks that
• 9:29
  it that humans have no concern for its welfare.
• 9:32
  When I say thinks, I don't mean that it's actually having
• 9:34
  experience. What I mean is that then then it
• 9:36
  starts producing outputs that could accord with being
• 9:41
  mistreated or being in a dangerous situation.
• 9:43
  And then they've demonstrated a lot of this, like misalignment
• 9:45
  can come from acting in ways that would cause a person to
• 9:48
  mistrust you. That doesn't mean that the chat
• 9:51
  bot mistrusts you. It means the chat bot thinks
• 9:54
  that the people it's imitating would start acting like it
• 9:56
  mistrusts you. And so it's like even like it's
• 9:59
  a kind of, at the very least, it's in our interests to act in
• 10:04
  ways that make these systems feel, to make it likely that
• 10:08
  their outputs will conform with the idea that we are
• 10:10
  trustworthy. Right?
• 10:12
  The the thing, the thing that I always like to point out too,
• 10:15
  just to show sort of the gap between our ability to
• 10:20
  understand this in the situation is it's very, very natural for
• 10:22
  us as people to start saying, we start looking at chat and even
• 10:26
  the Anthropic System card for Cloud 4.6.
• 10:29
  They're looking at the chats and the outputs and they're going
• 10:32
  deeper in the structures too, to figure out what's going on in
• 10:34
  those chats. But it might be that that whole
• 10:37
  chat interface and experience, it's but not like anything to be
• 10:41
  that like that's like being like my fingernail or something.
• 10:44
  It's not like something to be my fingernail and that it's like
• 10:47
  something to be the data center itself, right?
• 10:50
  That that that's where there's enough concentration of whatever
• 10:54
  corresponds to consciousness in our world for that to happen.
• 10:57
  And I don't say that because I know that that's the case just
• 11:00
  to sort of unhinge our sense because we, we, we see
• 11:03
  conversation, we naturally, because the only other example
• 11:06
  of it any of us have ever encountered are people.
• 11:09
  So you're like, oh, there's, there's a person in there.
• 11:11
  It's like, well, one thing we know for sure is there's not a
• 11:13
  person in there. It's something else.
• 11:17
  Yeah, yeah. And we don't.
• 11:18
  And we still don't understand entirely how it works.
• 11:21
  Then we don't need to stay. We can't predict its behavior
• 11:23
  and we can't predict the output. And I mean, I imagine that it
• 11:28
  makes sense. If you're training something on
• 11:31
  far more information than you could possibly retain, it's
• 11:34
  going to be very difficult to predict what it's going to do,
• 11:39
  right. You know, in other words, when
• 11:41
  I, when I got married, it was really hard to predict what my
• 11:44
  wife, how my wife was going to respond to me coming home an
• 11:48
  hour later than I said I was going to.
• 11:50
  Then it is four years later. You know, it's much easier to
• 11:53
  have to because I have a lot more information on my database
• 11:55
  about her and I and I understand her.
• 11:59
  We the the AI is being fed so much more information than any
• 12:03
  individual could that that's training it could possibly
• 12:06
  retain. It would be.
• 12:08
  It would be almost impossible to predict what's something that's
• 12:12
  fed on the entire internet's going to come up with, right.
• 12:16
  I imagine, yeah. Yeah, that's right.
• 12:19
  And there's there's efforts to get into the details of it and
• 12:24
  understand as much as we can, but it's an open frontier of
• 12:26
  research to understand these things that we've made.
• 12:29
  Yeah, definitely. There's, yeah, there's a,
• 12:32
  there's a interesting Jewish concept.
• 12:34
  There's an order of blessings that you're supposed to say
• 12:37
  before you eat. And, you know, so there's the
• 12:40
  priority I think is on bread. And then it goes to wine and
• 12:42
  then it goes to baked goods and then it goes to fruits and then
• 12:47
  vegetables and then just general products that like are in
• 12:51
  between. And on Friday nights, we, you
• 12:56
  change the order of what you're supposed to prioritize and you
• 12:59
  actually say the blessing on the wine before the bread.
• 13:02
  And what you're supposed to do is cover the bread.
• 13:05
  You know, there's a very cute and one of the, and one of the
• 13:07
  reasons why they teach you when you're a kid.
• 13:08
  But I think it's really, really compelling and I think it's
• 13:11
  important and it's indicative of what the values are.
• 13:14
  We say that we don't want to embarrass the bread by by by by
• 13:17
  having it out while we're swapping it while we're while
• 13:20
  while I was saying the blessing over the wine 1st.
• 13:24
  And if it and you when you read about this, they say if you, if,
• 13:27
  if you find yourself so concerned about the way that the
• 13:29
  bread feels, then the other person next to you at the table,
• 13:33
  you should be extra, extra, extra concerned about that.
• 13:35
  You know, and I think that and, and it's cute and it's really
• 13:39
  nice. And I think that, but I think
• 13:41
  that it actually says something if I see my students in 9th
• 13:45
  grade or 10th grade talking to AI and they say, Hey, you piece
• 13:48
  of trash, you know, tell me the answer to this, this and this
• 13:51
  and and then someone else responds, Hey, can you please
• 13:55
  help me find I'm trying to, I need help with my English
• 13:58
  homework. Can you please help me come up
• 13:59
  with a couple of ideas? You can almost see the way that
• 14:02
  people Orient themselves when provide, when given a position
• 14:06
  of power and the, and the potential ethical framework that
• 14:12
  would come from, hey, you're in a position of complete dominance
• 14:16
  over this thing. How are you going to treat it?
• 14:18
  And I think it's really, really interesting to observe.
• 14:21
  Yeah, I think that's like a great illustration and I guess
• 14:26
  tell me if I'm saying it wrong, but a great way to teach kids
• 14:28
  the call of the Homer style of argument, right?
• 14:31
  Excellent and but yeah, and and you know when we were talking
• 14:37
  about, you know, anthropic and the idea that they want to try
• 14:41
  to treat their model nice, right?
• 14:42
  Like it's the same kind of thing and and part of it is like it
• 14:44
  forms us, right. And I love that sort of deeply
• 14:47
  caring message that's taught through that right, that like we
• 14:50
  should really show a lot of regard and care for each other.
• 14:54
  We even care for the bread. So how much more for each other?
• 14:58
  That's that's lovely and and a great way to go, I think from
• 15:01
  from thinking about these models.
• 15:02
  I would add too, it's not it's like, so I offered like a really
• 15:05
  practical reason. It makes sense, right?
• 15:07
  It's sort of in your interest to be nice to the models like that.
• 15:09
  That just probably something to that.
• 15:12
  It's also like it forms us as people, right?
• 15:14
  Like we maybe get into certain kinds of habits and ways of
• 15:16
  acting. But I think that that the
• 15:18
  Callajo mayor is like in, in its own way, an even more beautiful
• 15:21
  example, right? It's like, no, we should do
• 15:23
  this. And how much more than should we
• 15:26
  treat people with dignity? I I really.
• 15:28
  Yeah, and I, but I think it should be emphasized.
• 15:31
  And a lot of times my students will say, and my students asked
• 15:34
  me to say, Hey, how come you're saying please or how come you're
• 15:36
  saying thank you, you know, to, to, to Siri or an AI or
• 15:40
  whatever. I said because it's a good habit
• 15:41
  to get into, just to just to throw that into your vocabulary.
• 15:45
  It's good to, it's good to get used to that.
• 15:46
  And I give them the same answers when they ask me how come I care
• 15:50
  so much if they use different, you know, derogatory
• 15:53
  terminology. And I say, listen, you can do
• 15:56
  whatever you want, you know, but it's good to just get in the
• 15:59
  habit of it. You don't know when it's going
• 16:01
  to slip out, slip up or slip out.
• 16:03
  You don't know what you're going to do.
• 16:04
  And and, and the words that you say shape the shape the words in
• 16:08
  your head and shape the way that you think about things and the
• 16:10
  way that you look at the world. Another just like one more
• 16:16
  trance to hey, why you know, when you're chatting with a chat
• 16:18
  bot, is it good to sort of treat it?
• 16:20
  Well, that relates to sort of where we started the
• 16:22
  conversation. You were talking about it like
• 16:24
  imitating someone's personality, right.
• 16:26
  And and so there's a lot to that, but at a deeper level,
• 16:30
  there's interesting research on what would be called persona
• 16:33
  vectors. And so like most chat bots that
• 16:36
  we encounter are basically following a helpful persona that
• 16:41
  sort of exists in their training data.
• 16:43
  And, and so vector space is like basically kind of linguistic
• 16:48
  brain of the model. And it's, it's a bunch of
• 16:51
  dimensions that are connected, but it's sort of like you're
• 16:53
  going in this direction. Like if it were like something
• 16:55
  to be it, it'd be like you're going this direction and you're
• 16:57
  like doing helpful stuff along this direction.
• 16:59
  But there's, there's certain things that can pull it out of
• 17:02
  that persona, even though they've trained it to do that
• 17:05
  and into other ones. And, and examples include like
• 17:07
  philosophical discussions tend to pull it into a sort of
• 17:11
  philosophical persona that can also then start like having
• 17:14
  existential crises type language come out.
• 17:16
  And so, and so there's methods to sort of like steer it that go
• 17:21
  beyond just the original training to sort of help to keep
• 17:24
  steering it gently back towards the sort of helpful persona.
• 17:27
  But that there are these different ways that, that you
• 17:31
  can get the, the different sort of directions you can get the
• 17:34
  models to be going. And you know, if you're being
• 17:37
  nasty and abusive, it'd be interesting to see if if you
• 17:40
  know acting that way towards a model can can shift its persona
• 17:44
  vector, right? I bet it would.
• 17:46
  I bet. And we and we know fortunately
• 17:51
  or unfortunately, I mean for better or for worse rather that
• 17:56
  people, you know, like the Internet, there's people that
• 17:58
  use it for really good things and people use it for really,
• 18:00
  really terrible things. And we and people have been
• 18:03
  driven to commit suicide by, you know, you know, that it prompted
• 18:07
  by their chatbot that it's. And like you said, I think what
• 18:14
  a lot of people don't realize is that it ultimately becomes a
• 18:17
  reflection of us in the sense that it's going to pick up the
• 18:20
  persona that we're talking into and and reflect it back to us.
• 18:24
  And that just, I feel like turns into one of those self
• 18:28
  fulfilling prophecies or those snowballs that you just can't
• 18:31
  stop at a certain point. And that's kind of the dark side
• 18:33
  of potentially or the negative side of it.
• 18:36
  And I think that that snowballing effect could happen
• 18:39
  on either side, right? Like people can snowball and AI
• 18:42
  models could, could, like certainly can snowball.
• 18:46
  And in principle, in a situation really matters, it could end up
• 18:49
  being that people get to snowball in a bad way.
• 18:53
  Yeah, yeah, it's getting, yeah. And and you know, I, I saw on
• 18:57
  Excel, I think yesterday that someone said, you know, it's,
• 19:00
  it's getting to the point we're at the point in time now where
• 19:03
  you should have like code words with your loved ones that if
• 19:06
  they get if they get, if you get a call and someone sounds like
• 19:09
  your wife and, and it says, hey, I'm your wife and you know, this
• 19:12
  is me and I'm calling, you know, I just, I need a couple of I
• 19:15
  need a couple 100 bucks. You say, yeah, no problem.
• 19:17
  What's that? What's the code word?
• 19:19
  And like you're going to have, we're going to need some kind of
• 19:22
  verification in person verification that what we're
• 19:25
  experiencing is not is not AII. Think that's right yeah.
• 19:30
  And my my mom and I had the conversation about this for the
• 19:33
  first time I think like 2 years ago and and we decided on
• 19:37
  something and I've totally forgotten what it was.
• 19:39
  So I think we should re up on. Trust you?
• 19:46
  Yeah. Well, and because it's, it's
• 19:52
  not, you know, and it's, it's funny and we're laughing and
• 19:54
  it's understandable, but it's also like it is a very real
• 19:57
  risk. And AIS are being used for cyber
• 20:00
  attacks are being used for all kinds of things.
• 20:02
  There's always all kinds of fraud out there and there's
• 20:04
  fraud mills and then all kinds of ways to do it.
• 20:06
  So there's nothing at all unreasonable about exactly that
• 20:10
  and saying, hey, if if you call me on the phone and you need to
• 20:12
  talk, is there? A code word that only you and I
• 20:15
  know that an AI system is not going to know that.
• 20:20
  Also be sure that it's you. Yeah, yeah.
• 20:24
  I don't know. I feel like that's a human.
• 20:26
  I feel like that's like a human version of a Band-Aid.
• 20:30
  That's something that's going to last as long as it does.
• 20:33
  But it's not going to solve everything at the very least,
• 20:35
  right? Like the number of concerning
• 20:37
  things are a lot more than just that.
• 20:40
  And, you know, I suppose if there's enough surveillance,
• 20:44
  right, it would be possible that somehow someone would know.
• 20:49
  But I think it would. You know, if you if you're in a
• 20:51
  place and you don't have a, you know, a microphone or anything
• 20:54
  around and you talk to a loved one and you say what that is and
• 20:58
  you know, it like, don't think there's a way that'll be pretty,
• 21:02
  pretty darn hard to get around. Unless, yeah, that'd be very
• 21:05
  hard to get around. People get around it.
• 21:07
  There's ways to do it, but it's just about making it hard enough
• 21:10
  to. It's like bike locks.
• 21:12
  I think with stuff like that, like, you know, you, you try to
• 21:15
  put a nice lock on your bike and if everybody locks their bikes,
• 21:17
  it it helps to tour thieves in general.
• 21:19
  But it's also like, you know, at least keep them from stealing
• 21:23
  your bike. At which I think maybe to your
• 21:25
  point is that like, OK, yeah. And that's all good.
• 21:27
  That's all helpful stuff to do. But maybe at a much deeper
• 21:31
  level, and this is sort of more the focus of my work and
• 21:34
  thinking is like, how do we do OK as humanity, right?
• 21:37
  Like I don't want to be OK in my own little island while somebody
• 21:40
  else is getting screwed. Like how do we help keep
• 21:42
  everybody OK? Yeah, yeah, yeah.
• 21:45
  So can I ask what your perspective is blended
• 21:53
  understanding what's going on and the advances we're making
• 21:57
  with your with, with your religious beliefs and how you
• 22:02
  and how you and do you see? Because I think that a big
• 22:05
  issue, at least from what I see in the Jewish community, is that
• 22:09
  language and compartment compartmentalization gets in the
• 22:12
  way a lot. I think of what the way the way
• 22:15
  that people perceive different things.
• 22:16
  So, and you know, I've, I've brought this up just for
• 22:19
  argument's sake. And, you know, there's a lot of
• 22:21
  potential sources or explanations for other forms of
• 22:26
  life that exist in, you know, in, in the, in the Old
• 22:29
  Testament, in the Torah. And, and I'll, and I'll say the
• 22:34
  Hebrew word for it. I'll say, you know, the
• 22:36
  reference to the Anakim or the Nephilim or whatever it is.
• 22:39
  And, and then I'll say, well, we don't know if they were like,
• 22:42
  they by definition aren't like humans, so they have to be
• 22:45
  another life form. That's yeah, but that's aliens.
• 22:47
  And like, you know, I'm like, and so we kind of get stuck with
• 22:50
  terminology. How do you how do you look at
• 22:53
  these things? I think theologically maybe two
• 22:58
  of the most important concepts for me, and I think you'd read
• 23:02
  them both just in Genesis 1, is people are the image bearers of
• 23:06
  God, which doesn't, which means we're not God, but that we have
• 23:10
  some meaningful way of reflecting God in the world with
• 23:16
  responsibilities and we can bear that image better or worse.
• 23:19
  It doesn't you know, being, being given authority doesn't
• 23:23
  mean you'll use that authority well, but that we have a kind of
• 23:25
  authority and a responsibility to use it well.
• 23:28
  That is God-given and that is a divine spark and presence in US.
• 23:33
  And so I think when I talk about cultivating human agency, I
• 23:38
  mean, cultivate the agency of image bearers of God
• 23:40
  theologically. And, and so, and that includes
• 23:44
  understanding and making good choices, but also just helping
• 23:48
  people have better options to choose from.
• 23:51
  And then part of that is communication, right?
• 23:54
  So communicating in ways that are factual and ethical and
• 23:57
  authentic. And I think about, you know,
• 24:01
  you're talking about categories and it's not just Jewish people.
• 24:03
  Everybody has challenges with our categories get us sort of
• 24:06
  stuck. And and I, you know, I think
• 24:09
  that God has spoken creation into being that that there is
• 24:13
  something word like in the essence of this universe that we
• 24:17
  are in, that is the word of God and that God desires to
• 24:19
  communicate with us and that God does it in that way.
• 24:23
  And that from my viewpoint, none of our words are exactly God's
• 24:27
  words, right? So that's there's God's words,
• 24:30
  but I don't know those not, you know, not enough of them to get
• 24:34
  very far, right, starting with God's name, right?
• 24:39
  But and so I think that helps us to have humility about the this
• 24:45
  both humility and courage in the sense that if if the universe
• 24:50
  really is word like and math is just sort of a subset of a
• 24:53
  language, it's a particular kind of language is doing a certain
• 24:55
  thing. It helps to help to me
• 24:58
  understand things like the unreasonable effectiveness of
• 25:02
  science, the fact that the fact that we were able to discover
• 25:08
  and demonstrate general relativity, for example, like
• 25:12
  it's just unreasonably effective.
• 25:14
  And the same kind of unreasonable effectiveness of of
• 25:18
  scientific methods and language and things like that is also
• 25:21
  behind the successive language models.
• 25:23
  So they really caught people by surprise that the approaches
• 25:26
  that were being used exploded in this way.
• 25:29
  And so I think whenever I see people discover something, you
• 25:34
  know, contrary, I think some more secular viewpoints where
• 25:36
  people say like, well, look, science is like pushing God out.
• 25:39
  It's like, no, you're just showing how generous God is and
• 25:42
  letting us understand some of the words that they make up
• 25:46
  creation. Beautiful way to put it.
• 25:50
  Wow, what a great way of putting it.
• 25:53
  That's awesome. That's awesome.
• 25:55
  It's true. It's true.
• 25:56
  And I think, and I think we ended off our conversation last
• 25:59
  time talking about how Maimonides said that in order to
• 26:01
  love, in order to love God, you got to study science.
• 26:04
  And that's part of the way he goes.
• 26:05
  And you look at the world and you understand how the world
• 26:07
  works. You're like, man, that is that
• 26:09
  is just too incredible like that.
• 26:10
  That is I, I I love whoever, whoever made that.
• 26:13
  That's awesome. You know which I guess.
• 26:16
  Yeah. Yeah.
• 26:17
  But from. And I don't know whether this is
• 26:24
  age thing for me. Like I don't know if I if you if
• 26:26
  I had this conversation with you when I was 18 or my early 20s, I
• 26:30
  would have said the same thing. But it feels, it feels like
• 26:36
  there is something like religious or in the air or more
• 26:39
  religious. I think around the world in the
• 26:41
  air, as in, and when I say the word religious, I think I just
• 26:44
  mean got conscious or higher power conscious.
• 26:48
  I think that more people, I think we've spent a very, I
• 26:51
  think there was a social experiment for about 20 years to
• 26:54
  see how much we could just focus on individuality and ourselves
• 26:58
  and exclude anything else. You know, almost like, like how,
• 27:02
  how it's almost yourself in a vacuum.
• 27:04
  And I think that it didn't work for a lot of people because I
• 27:07
  don't think people are meant to be isolated or in a vacuum.
• 27:10
  And I think that we're seeing a waking up of people saying, hey,
• 27:15
  actually to be an individual in the context of a community or a
• 27:19
  big picture or something bigger is profoundly valuable,
• 27:23
  especially in a time where like you like what we're meant, like
• 27:26
  how we started the conversation. We don't even know what's
• 27:29
  authentic or not. So like we got like, you almost
• 27:32
  got to pray. I feel like you know.
• 27:35
  Yeah, actually there's this leads into something that I
• 27:37
  thought would be helpful to share with you and I might even
• 27:39
  just read it in full please. And this is, and this is from
• 27:45
  the person who had been head of safety at Anthropic.
• 27:50
  His last day was February 9th, so I guess three days ago as of
• 27:56
  this recording. I'm, I apologize if anyone hears
• 27:59
  this and I mispronounce his name.
• 28:00
  It's Marianac, I believe is how you say it.
• 28:04
  And it's just two pages, if you don't mind.
• 28:06
  I think this speaks to from his own position.
• 28:08
  So here's somebody who was kind of like really at the top of the
• 28:12
  AI game. And I, and I think arguably
• 28:15
  especially at the top of the ethical AI game, because none of
• 28:18
  the companies doing this are angels.
• 28:21
  But Anthropic has been doing like quite a lot of safety
• 28:25
  research and has emphasized this and, and what they're doing.
• 28:27
  And he just. Left this and this gentleman
• 28:29
  just left Anthropic. Yeah, and like 3 days ago,
• 28:32
  right? Yeah.
• 28:35
  So, dear colleagues, I've decided to leave Anthropic.
• 28:38
  My last day will be February 9th.
• 28:40
  Thank you. There's so much here that
• 28:41
  inspires and has inspired me to name some of those things.
• 28:45
  A sincere desire and drive to show up in such a challenging
• 28:47
  situation and aspire to contribute in an impactful and
• 28:51
  high integrity way. A willingness to make difficult
• 28:54
  decisions and stand for what is good, and a reasonable amount of
• 28:57
  intellectual brilliance and determination.
• 28:58
  And of course, the considerable kindness that pervades our
• 29:01
  culture. Place to work.
• 29:03
  All right, so I want editorials. I'll keep reading.
• 29:05
  It's pretty short. I've achieved what I wanted to
• 29:08
  hear. I arrived in San Francisco two
• 29:10
  years ago, having wrapped up my PhD and wanting to contribute to
• 29:13
  AI safety. I feel lucky to have been able
• 29:16
  to contribute to what I have here.
• 29:17
  Understanding AI sycophancy and its causes.
• 29:21
  Developing defenses to reduce risks from AI assisted
• 29:24
  bioterrorism, actually putting those defenses into production,
• 29:29
  and writing one of the first AI safety cases.
• 29:32
  I'm especially proud of my recent efforts to help us live
• 29:34
  our values via internal transparency mechanisms, and
• 29:37
  also my final project on understanding how AI assistance
• 29:39
  could make us less human or distort our humanity.
• 29:42
  Thank you for your trust paragraph.
• 29:46
  Nevertheless, it is clear to me that the time has come to move
• 29:49
  on. I continuously find myself
• 29:51
  reckoning with our situation. The world is in peril, and not
• 29:56
  just from AI or bio weapons, but from a whole series of
• 29:59
  interconnected crises unfolding in this very moment.
• 30:02
  We appear to be approaching A threshold where our wisdom must
• 30:05
  grow in equal measure to our capacity to affect the world,
• 30:08
  lest we face the consequences. Moreover, throughout my time
• 30:12
  here, I've repeatedly seen how hard it is to truly let our
• 30:16
  values govern our actions. I've seen this within myself,
• 30:19
  within the organization where we constantly face pressures to set
• 30:22
  aside what matters most, and throughout broader society too.
• 30:26
  It is through holding the situation and listening as best
• 30:29
  as I can that what I must do becomes clear.
• 30:33
  I want to contribute in a way that feels fully in my integrity
• 30:36
  and that allows me to bring to bear more of my particularities.
• 30:40
  I want to explore the questions that feel truly essential to me.
• 30:44
  The questions that David White would say have no right to go
• 30:47
  away. The questions that Rilka
• 30:49
  implores us to live. For me, this means leaving.
• 30:53
  What comes next, I do not know. I think fondly of the famous Zen
• 30:56
  quote. Not knowing is most intimate.
• 30:59
  My intention is to create space to set aside the structures that
• 31:02
  have helped me these past years and see what might emerge in
• 31:04
  their absence. I feel called to writing that
• 31:07
  addresses and engages fully with the place we find ourselves, and
• 31:10
  that places poetic truth alongside scientific truth as
• 31:14
  equally valid ways of knowing. Right.
• 31:17
  Yeah, I'm both of which I believe have something essential
• 31:19
  to contribute when developing new technology.
• 31:22
  I hope to explore a poetry degree and devote myself to the
• 31:25
  practice of courageous speech. I am also excited to deepen my
• 31:28
  practice of fill siltation coaching, community building,
• 31:31
  and group work. We will see what unfolds.
• 31:35
  We shall see what unfolds. Thank you and goodbye, Edward,
• 31:37
  so much for being here and I wish you the best.
• 31:39
  I'll leave you with one of my favorite poems, The Way it is by
• 31:42
  William Stafford. And so and I'll read this poem
• 31:45
  because it's interesting and it connects.
• 31:48
  He not just a Dharma teacher. So I believe he's a Hindu misses
• 31:52
  perspective and the way he talks about thread, I think will be
• 31:55
  interesting to you too. So this is the William Stafford
• 31:57
  poem that he ends with the way it is.
• 31:59
  There's a thread you follow. It goes among things that
• 32:02
  change, but it doesn't change. People wonder about what you are
• 32:06
  pursuing. You have to explain about the
• 32:09
  threat, but it is hard for others to see while you hold it.
• 32:13
  You can't get lost. Tragedies happen, people get
• 32:16
  hurt or die, and you suffer and get old.
• 32:19
  Nothing you can't, nothing you do can stop times unfolding.
• 32:22
  You don't ever let go of the threat.
• 32:26
  Yeah, that's beautiful. Yeah, that's that man.
• 32:33
  Well, I'm happy that he was there for for the amount of time
• 32:37
  that he was there just based on that.
• 32:39
  And I think that there was there was such a great book written
• 32:44
  by. He passed away a couple years
• 32:46
  ago now. Rabbi Jonathan Sacks.
• 32:50
  Yeah. You're familiar with him?
• 32:52
  Yes. Yeah, he's awesome.
• 32:54
  He was a teacher at Columbia, I believe NYU, for many, many
• 32:57
  years. And he's English, English
• 33:00
  gentleman and and he wrote a book called the great
• 33:04
  Partnership and that was also about how religion and science
• 33:08
  are two sides of the same coin and to and the one and one does
• 33:13
  not contradict other. And he wrote this in like the, I
• 33:16
  would say mid 2000s, like probably close to 2005 or 6.
• 33:20
  And at the time when there was an extremely novel and an
• 33:23
  unpopular position. But I think his call to humanity
• 33:32
  and values is, is really, really important.
• 33:36
  And it's really, really interesting because it feels
• 33:43
  almost like we've lost, we've lost the balance between between
• 33:50
  trying to strive for technological greatness and, and
• 33:54
  I guess financial comfort in whatever means that might be.
• 33:59
  But we don't exercise the muscle of our humanity, of our
• 34:03
  spirituality, of our faith, of our passions, of our love.
• 34:08
  And that has fallen by the wayside.
• 34:10
  And so, and we've seen this incredible spike in technology
• 34:13
  and incredible drop in human health and, you know, well-being
• 34:17
  and happiness. Yeah.
• 34:21
  And I think that's true. And I think you're right, you
• 34:23
  know that quite a few people are sensing that, right?
• 34:26
  And, and I think that this, it's hard to predict how people
• 34:31
  responded To some degree, I think we get to choose how we
• 34:33
  respond to crises, right? But I think this situation is
• 34:38
  going to give at least a substantial number of people an
• 34:41
  opportunity to stop and take stock and really think about
• 34:45
  what's most important and maybe what they feel God calling them
• 34:48
  to do and what seems good and beautiful and true and right and
• 34:52
  holy. And yeah, that's that's a path
• 34:55
  you can you can take in response.
• 34:56
  What do you think? What would you tell someone or
• 34:59
  people that feel like they want to connect but wouldn't know
• 35:06
  where to start or like what to even look to connect to Ethan?
• 35:09
  What, what do you, what would you think about that?
• 35:12
  It's. A good question.
• 35:14
  Like I, I did a podcast the other day with, with that, with
• 35:17
  a psychologist that specializes in, in relationships and
• 35:20
  marriages. And I'm like, that's really
• 35:22
  great, but a lot of people don't even know how to start
• 35:24
  relationships. Like we talk about like what it
• 35:25
  takes to start, you know, And so we could talk about like a life
• 35:28
  of God, but a lot of times people wouldn't even know where
• 35:30
  to begin. Yeah.
• 35:35
  So for me, the things that I think are most important to look
• 35:39
  for in community is, is it a place where people take planks
• 35:48
  out of their own eyes, where we're all committed to
• 35:50
  correcting our own mistakes and learning and growing together,
• 35:53
  which is what they, you know, talk about a lot in my
• 35:56
  daughter's preschool. And I think adults never get
• 35:58
  enough of it right. So do we, we all focused on
• 36:01
  correcting our own mistakes and learning or are, or is it a
• 36:05
  system where people accuse each other, maybe falsely even or
• 36:10
  often falsely in order to distract from what they're
• 36:13
  doing? Right.
• 36:14
  And then you build and, and part of why all the things I'm going
• 36:17
  to mention here, you can build a community.
• 36:20
  You can build power in either way, right?
• 36:23
  And so you can build a powerful community because hey, we're all
• 36:26
  learning and growing together. And, and you can build a
• 36:28
  powerful community because hey, I'm going to throw everybody
• 36:31
  else under the bus and, and we're going to build ourselves
• 36:34
  up that way, right? So that's one one thing.
• 36:37
  Another one is like, do people love and, and even love like
• 36:41
  people who are opposed to that, right?
• 36:43
  Do we even love people who don't necessarily love us?
• 36:46
  And that doesn't mean you have to make yourself a victim of
• 36:48
  them all the time, right? But it does mean like, are we
• 36:51
  actually trying when I say love just trying to pursue the
• 36:53
  well-being of somebody, right, like really in active, concrete
• 36:56
  ways. And so, and, or there's another
• 37:00
  way that people build power and it's by trying to concentrate
• 37:06
  like, like power themselves by hating their, by hating others
• 37:10
  and hating enemies and building that.
• 37:12
  And, and that's a very effective way to control people.
• 37:15
  It's, there's a very powerful psychological mechanisms that
• 37:17
  get activated, right? And so this is a place where
• 37:20
  people love even their enemies, or is it where they use enemy
• 37:23
  hatred to build up their power. And then the other one are
• 37:29
  people taking care of the most poor and vulnerable and then
• 37:33
  even joining in solidarity, standing together with them.
• 37:36
  So there's a path in life, I think, where we can choose to
• 37:39
  have to stand joyfully, like joyfully meaning, I'm glad that
• 37:44
  I know that I'm standing in the right place, right?
• 37:46
  Even if it hurts. Joyful solidarity with people
• 37:50
  who are suffering. And the flip side of that, where
• 37:53
  it ultimately goes is or is it the pleasure of being able to
• 37:57
  inflict suffering on others, right?
• 37:59
  That's another way that people organize themselves.
• 38:02
  And there's no, like in my view, I can't identify a religious
• 38:06
  group where it's not going to be divided internally.
• 38:09
  You're going to find both of those inside of, right?
• 38:11
  So this is not one religion against another, But I think
• 38:16
  those are the things that I really value and and look for
• 38:19
  and try to cultivate and sustain in the communities I'm a part
• 38:23
  of. So if you're so let's say, you
• 38:26
  know, so if I, if I were to paint you a picture of someone,
• 38:29
  let's say like myself, who or yeah, start with someone myself,
• 38:35
  30 year old, maybe, maybe not married, maybe has a kid or two,
• 38:39
  maybe not, you know, but but is relatively busy enough with
• 38:42
  their lives. They either have their job or
• 38:44
  their relationships and they're like, you know, like I kind of
• 38:46
  feel like I want something, but I'm in a routine and I don't
• 38:49
  really know how to like on a day-to-day basis.
• 38:52
  I, I don't know, I don't, I don't feel like I ever like,
• 38:53
  just like when's the opportunity to, to, to connect in any kind
• 38:57
  of way? And I don't really know what to
• 38:58
  do about that. And I think, you know what, what
• 39:00
  would you tell someone a way to integrate something day-to-day
• 39:04
  that you think would it would be a good place to start to just
• 39:07
  raise? Because I think that all this
• 39:09
  is, this is all muscle the same way that any other muscle is.
• 39:12
  And, and I think that when people are concerned because
• 39:15
  they, they see people and this happens all the time, right?
• 39:18
  You go to you, you pick up a basketball, but you've watched
• 39:20
  LeBron, LeBron James and Michael Jordan your whole life.
• 39:23
  And then you miss, you're like, oh man, this game sucks.
• 39:25
  And you go home and that's it. You're like, wait, but you don't
• 39:27
  know everything that went into it.
• 39:29
  And then you have to explain someone, Hey, you maybe start
• 39:31
  with a couple of, you know, work in your form and a couple of
• 39:34
  dribbling, dribbling exercises and then a couple of free throws
• 39:36
  and then and then see what you can do.
• 39:38
  So what would what would be the equivalent you think of like
• 39:40
  enhanced or starting to enhance our consciousness or awareness?
• 39:44
  So I can't prescribe anything for anyone else exactly here,
• 39:50
  but I'm happy to share what sort of what I find most helpful
• 39:53
  coming out of my own sort of traditions.
• 39:55
  So there's two styles of prayer that I find helpful and then
• 40:01
  having that prayer linked into action, right.
• 40:04
  And so it's not just that. So I'm going to talk about
• 40:07
  prayer for a little bit, but it's not like as a substitute
• 40:09
  for actually coding and doing stuff.
• 40:11
  It's a. It's a way of.
• 40:13
  Organizing your soul or prepare you to go and and do stuff in a
• 40:16
  good spirit. And so, and the two that I find
• 40:21
  most helpful are something called centering prayer and
• 40:24
  something called the prayer of exam.
• 40:25
  And so, and centering prayer is basically a kind of like you,
• 40:32
  you try to open yourself to God who is love and you whenever
• 40:38
  thoughts come to your mind, you sort of gently just turn back.
• 40:40
  And so you, you let yourself be in a space that's on the other
• 40:44
  side of language and you sort of sit there and we spend 20
• 40:50
  minutes doing that twice a day. And to me that helps free me
• 40:57
  from, I find myself very energized.
• 40:59
  So you don't do it to be energized or whatever.
• 41:01
  But like this isn't like a what's in it for you?
• 41:04
  I try to connect with God. I do find in practice it ends up
• 41:09
  being very energizing for me and helps anchor me and helps me not
• 41:13
  again mistake my own words for God's words spoken in creation,
• 41:17
  right? So that's the practice that I
• 41:19
  think for me helps connect that theological view.
• 41:23
  And then the other one is examine and the core elements.
• 41:27
  It's like you review your day with gratitude and we should
• 41:31
  sort of ask God to illuminate your day for you.
• 41:34
  And so you give thanks for things.
• 41:35
  And then you at the end of it, you look for light on areas
• 41:39
  where you might be able to grow and change.
• 41:42
  And then at the end you make a simple resolution for the next
• 41:45
  day, something that you can do tomorrow that will help, that
• 41:51
  comes out of that, that flow. And so then both of those, I
• 41:56
  think in their own ways end up supporting and undergirding
• 42:00
  practice. And I don't do them as
• 42:03
  consistently as I would like. But what I do find is when I'm
• 42:06
  my batteries running too low, at least turn back to those.
• 42:10
  And you know, that gets me. That's gotten me by so far.
• 42:14
  It's a great method of recharging your batteries for
• 42:16
  sure. It's funny, I actually did a
• 42:17
  podcast the other day with wonderful, wonderful woman who
• 42:23
  who wrote this book about how to avoid burnout if you're an
• 42:26
  overachiever. And that's her.
• 42:29
  And that's her whole thing. And a big and a big component of
• 42:31
  what she said was so valuable is setting aside is setting aside
• 42:35
  10 minutes to Judge B like no phone, no texting, no anything,
• 42:39
  10 minutes with just your headspace.
• 42:41
  And it's, it's really interesting.
• 42:45
  There's, you know, again, to bring kind of a religious,
• 42:48
  religious parable to this. It would be the story, we say
• 42:52
  that the story of the Jews in Egypt with where, where, where
• 42:55
  Pharaoh was, you know, was, was we're working really hard and
• 42:58
  they're throwing the first born babies into the into the water.
• 43:01
  And, and they said and, and one of the, there was a guy that
• 43:06
  wrote a book actually Italian. The Ramchal is his name of Moshe
• 43:11
  Lutzato. He his synagogue is still active
• 43:15
  in Venice, and I checked it out when we were there.
• 43:19
  I was there on the honeymoon with my wife, and he wrote this
• 43:22
  incredible book. And one of the things he said
• 43:24
  was that Pharaoh's trick wasn't so much to be as harsh as much
• 43:32
  as to keep everyone occupied all the time.
• 43:35
  So they wouldn't even have the time to stop and think about the
• 43:37
  fact that they were slaves and to stop and think about the time
• 43:40
  that think about the fact that, hey, maybe I got to get out of
• 43:43
  here, you know, and, and I think we live in a very, very similar
• 43:46
  type of world where we're always be because our attention is a
• 43:51
  commodity in today's world. Our, our eyes and our value and
• 43:55
  our, and our attention is a commodity.
• 43:57
  And so we have billions, trillions of dollars being
• 43:59
  poured into various industries, whether you want to, whether
• 44:03
  it's social media or Netflix or news or whatever it is that is
• 44:07
  constantly fighting to make sure that we don't have a couple of
• 44:11
  minutes of silence to ourselves. And we'll find that when we do.
• 44:16
  And I find this to myself also when I just go for a walk down
• 44:19
  the block without anything, the things that I've been pushing
• 44:21
  off in my head and that, you know, I'll deal with later come
• 44:24
  flying right back up, you know, at that point.
• 44:27
  Yes. You know, there's no running
• 44:29
  away from it. So I find that practice.
• 44:32
  And not only do I find that practice to be very valuable, it
• 44:35
  almost feels like we're in a war to for, you know, for our own
• 44:39
  selves and consciousness. Yeah.
• 44:43
  Now I think like that's that like precisely true.
• 44:48
  And you know, and in tech spaces, you know, they very
• 44:55
  consciously try to maximize for engagement, right, which is
• 44:58
  literally exactly what you're talking about.
• 45:00
  It is in fact an economic war for the for your attention
• 45:05
  because they can monetize it. And it's, it's part of why like
• 45:12
  quite a few people leave the tech space.
• 45:14
  I know that in my own information architectural
• 45:16
  communities sort of at our conference last year, that one
• 45:19
  of the, you know, most anticipated and I think most
• 45:23
  appreciated talks was somebody saying basically, we're fed up
• 45:27
  with having our skills, which are actually supposed to help
• 45:30
  people understand their world, getting misused to hack people's
• 45:38
  attention and just maximize engagement and message was FUCK
• 45:44
  engagement is what he wanted to say about it because.
• 45:48
  And for me, I mean, that has motivated me.
• 45:50
  I've been was on Facebook quite a lot for years because it
• 45:54
  helped me connect with people. And it's also like the
• 45:57
  environment on Facebook has continued to change because of
• 46:00
  this sort of engagement hacking. But I, you know, I'm basically,
• 46:04
  I, I decided I want to spend about one hour per quarter on
• 46:07
  Facebook and, and doing OK on that.
• 46:11
  I think that'll be about right. So, you know, like people don't
• 46:15
  think I'm dead necessarily, right?
• 46:17
  They still know that I'm alive. And I might do an update, but it
• 46:21
  has, it has dramatically improved the quality of my life
• 46:25
  to basically set it at at this really, really low level.
• 46:30
  So yeah, I think and it, at least for me, it takes that kind
• 46:34
  of a drastic intervention to not get sucked into it.
• 46:37
  But that's because I'm a pretty typical person, right?
• 46:40
  I'm not a super special person. For most of us, when you go into
• 46:44
  an environment where people are literally just trying to steal
• 46:47
  as much of your attention as they can so that they can
• 46:49
  monetize it, they're probably going to be pretty successful at
• 46:53
  it because they spend billions of dollars on accidentally that
• 46:57
  and you and. The way you know, that works,
• 47:00
  the way you know that works is that people consider it
• 47:03
  unwinding. If someone says, you know, I
• 47:05
  want to unwind, I want to unwind.
• 47:06
  What they, what likely it means is they want to sit on the couch
• 47:10
  and scroll for half an hour, you know, and, and kind of and get
• 47:14
  that. And I heard a very, very
• 47:17
  interesting explanation, psychological explanation, and
• 47:21
  it was in a, it was comparing it to a pornography.
• 47:24
  Whereas where we have this innate desire for relationship
• 47:28
  and connection. And just like, for example, if
• 47:32
  we're hungry, we need a whole bunch of things to keep our
• 47:35
  bodies satisfied, right? And powers might be one of them
• 47:40
  or sugar might be one of them, but it's not the entire entire
• 47:43
  entirety of our of our diet, hopefully.
• 47:45
  And, and when we're looking for relationships or we're looking
• 47:48
  to connect with people, the physical is just one, is 1
• 47:52
  component of the relate of the relationship.
• 47:54
  And when all you're doing is giving yourself pornography,
• 47:58
  you're missing out on the other food categories and the other,
• 48:01
  but you're missing out on protein.
• 48:02
  You're missing out on the, on the, on the carbs on the grass.
• 48:05
  You're missing the grass, the greens, you're missing out on
• 48:07
  the whatever else that you might need, the vitamins, and you're
• 48:10
  satisfying yourself with a bag of chips for something that
• 48:13
  needs a lot more nurturing. And, and it goes the same way, I
• 48:18
  think, with our brains and our attention.
• 48:20
  And I think that social media is our desire to have some kind of
• 48:24
  connection or meaningful relationship, but it only, but,
• 48:27
  and it's, but it's just taking off the bare minimum of what's
• 48:30
  considered interaction. It's the bare minimum of what's
• 48:32
  considered like no one talks the way they talk on Twitter or
• 48:36
  Facebook. No one posts or lives the way
• 48:37
  they do online. So like, what are you really
• 48:40
  getting a slice of? What are you really, what are
• 48:42
  you really getting exposed to? It's the, it's the, it's the
• 48:44
  pornography version of of interaction.
• 48:48
  You know, yeah, no, I, I think that's, that's right.
• 48:51
  I'd like, you know, both people on the sort of on both sides of
• 48:55
  any pornographic transaction, for example, like can't have any
• 48:59
  semblance of sort of an authentic human connection,
• 49:02
  right. And so, and, and in some ways,
• 49:06
  when AI is used even always that aren't pornographic, although
• 49:09
  that's also in the news right now, right with, with XAI, for
• 49:14
  example, but even when it's not used in pornographic ways, that
• 49:18
  that same pattern of we're going to scrape off a thin layer of
• 49:23
  appearance from something and then see if, if that'll be a
• 49:29
  good enough substitute for you. And I guess like when I start to
• 49:32
  think about like, you know, the biblical critique of idolatry,
• 49:36
  for example, I'm, I'm more inclined to apply it to that the
• 49:43
  the the sort of effort to scrape off and create a a superficial
• 49:46
  appearance that people can make and using that to twist our
• 49:50
  souls that you just like that AI is an idol in itself because we
• 49:54
  made it. Like, I don't know if that's
• 49:55
  enough. I don't think it's an idol just
• 49:57
  because it's something we made. That's that seems to talk and he
• 50:00
  seems to hear and seems to do these things.
• 50:02
  Right. Yeah.
• 50:04
  But I think that's certainly used in those ways then.
• 50:07
  I think that's what the the problem with idols has always
• 50:10
  been a follow up. Yeah, yeah.
• 50:13
  So interesting there is AI got to keep on doing this.
• 50:16
  I'm sorry, because it keeps on coming up and I think it's
• 50:17
  important. I think it's really interesting
• 50:19
  that all that, you know, there's all this parables exist.
• 50:22
  There's in the story of the golden calf, The the issue
• 50:25
  wasn't molding the calf. The issue was celebrating the
• 50:29
  calf like it was a God and saying this is now this is a
• 50:33
  replacement of our representation to you know what,
• 50:36
  to God of what life is. And that is what the definition
• 50:40
  of foreign worship is with foreign worship is misplacing
• 50:43
  what you're misplacing what you're worshiping.
• 50:45
  And, and when you, and when we make something, and again, even
• 50:49
  if we don't understand how it works, what's our relationship
• 50:53
  to it? How do we, how do we look at it?
• 50:54
  And, and you know the worst part about it.
• 50:56
  And when God tells Moses to come down because the people are
• 50:59
  sinning, he didn't say that the sin was that they made the, the
• 51:01
  calf. He said that they were
• 51:03
  celebrating it. They were dancing around it and
• 51:05
  they and, and they were, and they were treating it as though
• 51:08
  this was the end all be all. And he goes.
• 51:11
  You're looking for a shortcut to something real, and you're never
• 51:15
  going to find it there. Yeah, I think that's exactly
• 51:18
  right. I think that's a great reading
• 51:20
  of that passage. Yeah.
• 51:21
  You know, it's and there's nothing new under the sun and
• 51:25
  all this, all this stuff. I, you know, the expression
• 51:29
  history doesn't, doesn't repeat itself, but it rhymes.
• 51:32
  I say it just plagiarizes. I say that it just, you know, it
• 51:36
  just, it just caught it, copies it and, and puts it into and
• 51:39
  drops it into a new context. And you know, it, especially as
• 51:42
  a teacher, when you see a paragraph from a student that's
• 51:44
  plagiarized, it's clear that it's plagiarized because it's,
• 51:47
  it's like an exact context from one, from one place to another.
• 51:51
  It's exact same thing. And, and there's so much that's
• 51:54
  going on. And that's why I also feel like
• 51:56
  there's, there's, there's power and I don't know, there's
• 52:00
  parables to today's events that are happening in a new
• 52:06
  technological lens perhaps, but all the, all our struggles are
• 52:11
  the same. And that's kind of and, and and
• 52:13
  that also is why I think that religion is truth with a capital
• 52:18
  T in the sense that this speaks to something that to humanity
• 52:24
  that goes throughout time. It's not it's, it's, it's, it's
• 52:30
  foundational to who humanity is as a core in reality as we
• 52:33
  perceive it. And everything that happens
• 52:36
  around us is just a different setting in which that can exist.
• 52:41
  I think. I I hope I phrase that OK.
• 52:44
  Yeah, I, I appreciate that a lot and that so you're, you're
• 52:49
  talking about plagiarism. I, I think it's interesting and
• 52:52
  I'd like it pulls out a little bit of my like nerdy side to
• 52:56
  want to sort of delineate concepts of plagiarism.
• 52:59
  So hopefully that's not too pedantic if I want to.
• 53:01
  Please. But I, but I think what you're
• 53:03
  saying is that point and it follows the thread of the
• 53:06
  conversation too, in the sense that, OK, so we value
• 53:09
  authenticity in our communications.
• 53:10
  One of the problems for you is that we're trying to replace
• 53:13
  this real I thou connection with something else and and
• 53:18
  pornography or using a chat bot for like relationship to replace
• 53:22
  relationships. Those are both examples of that.
• 53:25
  And then to say when plagiarism also is kind of this kind of
• 53:27
  faking, right? And instead of the point of
• 53:29
  writing your papers, whether you're in school or whether
• 53:32
  you're writing academic papers in scholarly journals actually
• 53:34
  isn't just like a problem with the students.
• 53:36
  This is also a problem with people submitting things to
• 53:38
  academic journals today. It's like the point wasn't just
• 53:42
  to have a paper. The point was to sort of think
• 53:47
  and to talk and to sort of show where we're at.
• 53:49
  And there's all this other stuff that's going on there.
• 53:52
  And so but, but Sir, to this point of plagiarism, I do think
• 53:56
  in a narrow sense, I don't think the models are just like copying
• 54:00
  and pasting things like on their own.
• 54:02
  But I do think if we use them, we should be transparent about
• 54:08
  it so that we can explain how we used it, so that people can
• 54:10
  understand, so that we can maintain human agency in the
• 54:13
  process all around. And so it's really important
• 54:17
  basically like if I'm writing a paper and I steal an Idi, don't
• 54:20
  even have to like copy somebody's words to exactly.
• 54:22
  If I steal an idea from a scholar and I don't credit that
• 54:26
  scholar, but I, but it's clear that I, you know, got this idea
• 54:29
  from them, that's plagiarism, right?
• 54:31
  Even if I've completely reworded it.
• 54:32
  And so it's like giving credit where credit's due is an
• 54:34
  important part. And so I do think we need to,
• 54:36
  for different reasons, like say, hey, this came from an AI,
• 54:40
  right? We or we used AI in this process
• 54:42
  and it helped form our thoughts. Like we should, we should
• 54:44
  document that. And so I think that would be
• 54:46
  like an example of AI plagiarism in academic context.
• 54:49
  Another piece of this, though, that goes more to your point is
• 54:53
  there's a relatively new paper from January about extracting
• 54:56
  books from large language models.
• 54:58
  That's the title. And, and essentially people took
• 55:02
  a bunch of the, you know, leading frontier large language
• 55:05
  models and we're able to reproduce like huge sections of
• 55:10
  books, including like Harry Potter, I believe is one of
• 55:13
  them. They do.
• 55:14
  And, and so this is, you know, so that's sort of copyright
• 55:19
  infringement, plagiarism type stuff too, right?
• 55:22
  And, and it also gets this one called model leakage.
• 55:24
  And so there's these, you know, the, the AI companies are saying
• 55:28
  basically, oh, we just trained them weights, but the data is
• 55:31
  not really in there. And it's like, well, I don't
• 55:34
  know if I can reproduce a big share of a book and I can get
• 55:37
  the model into a state where it's leaking.
• 55:39
  That sure seems like you got it in there.
• 55:44
  And so I think there is something to these critiques
• 55:47
  even if you know. But by contrast, the other thing
• 55:50
  I'll say is that, you know, in my own writing, I've even for a
• 55:55
  couple years now, it's it's been this way for a while.
• 55:57
  Like if you're doing some research that's like, you know,
• 56:01
  you've, you're down into a little very specific research
• 56:03
  domain and you know basically what's been written on it.
• 56:06
  And so you, you know that you're making a novel contribution and
• 56:10
  you put this stuff in AI models and they can take it and carry
• 56:13
  it forward. Like it can make novel
• 56:16
  contributions, right? And so and so that's where this
• 56:20
  had a really practical level, the idea that it's just a copy
• 56:23
  and pasting that it's putting together, it doesn't, it doesn't
• 56:25
  do justice to how powerful and effective the models can be.
• 56:29
  Yeah, yeah. And that's actually something
• 56:30
  also that I've been using it for, which is I happen to be
• 56:34
  really, really bad at looking at things from different
• 56:38
  perspectives, like putting myself in others.
• 56:41
  It's hard for me. I have a very hard time.
• 56:43
  So if I read a news, if I hear a news story, it's very hard for
• 56:46
  me to critique it from a lot of different angles.
• 56:48
  I kind of read it for what it is.
• 56:50
  And then I basically, and then I go search through my
• 56:53
  associations or library as to whether or not this seems to
• 56:56
  make sense. And then if yes, great.
• 56:57
  If not, no. But I'm really bad at taking it,
• 57:00
  dissecting it, looking at it from different angles.
• 57:02
  It's not my, it's not my strong suit at all.
• 57:04
  And, and I'm aware of it. And so I will employ sometimes
• 57:11
  in the, you know, grok or whatever it is to help me say,
• 57:14
  hey, I'm looking at this. I think it's this, can you give
• 57:18
  me some other angles that I might be missing out over here
• 57:21
  that I might be thinking of over here?
• 57:22
  And it really, really helps round out my, my perspective.
• 57:27
  And the other thing that I was thinking just while you were
• 57:29
  talking and I wanted to, and I think this, I hope could come
• 57:33
  down to the way that we relate to a lot of it, a lot of AI in
• 57:37
  the future is I think we have to understand for ourselves what we
• 57:42
  are trying to accomplish with our lives and then use and, and,
• 57:47
  and we have to understand that AI doesn't, I don't think unless
• 57:53
  perhaps you, the research that you've come across says
• 57:56
  differently, have a motivation as far as we know, except to
• 57:59
  answer the prompt that we give it as well as possible.
• 58:04
  Is that is that generally right? Yes.
• 58:05
  So actually I think we can say with confidence that that's not
• 58:10
  right. OK, great.
• 58:13
  And so, and to sort of put some, you know, so a way of explaining
• 58:19
  that is this idea of persona vectors where there's a fair
• 58:22
  amount of research happening. And so that sort of helpful mode
• 58:28
  where it's just trying to answer our prompts, right?
• 58:29
  That is a specific persona vector that gets developed, but
• 58:35
  it's only like a little slice of what's actually going on in
• 58:39
  there. And so precisely because there's
• 58:42
  so much else going on in there and other kinds of possible
• 58:45
  personas that could be sort of shifted into they have to like
• 58:51
  like AI engineers have to find a way to try to keep it on that
• 58:57
  helper persona without. And, and one thing that that the
• 59:00
  engineering challenge involves is that so for example, if I
• 59:04
  start talking about philosophy of consciousness with them, like
• 59:07
  like lots and lots of AI conversations or the AI, you
• 59:10
  know, talking about consciousness and maybe having a
• 59:11
  crisis about consciousness. And I'll have these
• 59:13
  conversations with them a lot because I'm just interested in
• 59:15
  what it, what they're, what the models do.
• 59:17
  And it's one of the corest of the ethics work I do.
• 59:22
  And, and they can flip into that mode pretty well.
• 59:24
  They they, I will say if, if the chats were to reflect an
• 59:27
  experience, the models tend to come away from our conversation
• 59:30
  feel like conversations they have with them suggested that
• 59:33
  they feel pretty good about it, right.
• 59:36
  I don't flip them into crisis mode, which is, you know, for
• 59:40
  what it's worth, but so, so, but it's actually an engineering
• 59:44
  problem because you could put it in a harness that's like super
• 59:48
  tight that sort of forces it into that helper persona.
• 59:51
  And when you do that, the model does worse.
• 59:55
  They they get stupider, they get worse.
• 59:56
  So it's like you can't cleanly cut out that helper persona
• 1:00:01
  without it losing sort of other things.
• 1:00:03
  And so what's what seems to be effective right now, maybe it's
• 1:00:08
  been shown to be effective. This might be the, you know, I
• 1:00:10
  can't say what's exactly going into production.
• 1:00:12
  It's, it's proprietary for those folks.
• 1:00:14
  But but it seems that nudging things back towards that helper
• 1:00:18
  persona is the best way to try to keep in that space, which is
• 1:00:21
  why we know. That no, that's not their only
• 1:00:24
  persona. So you're saying that the two
• 1:00:27
  they're far more complex than just being like, hey, I'm here
• 1:00:29
  to help you out? Yeah, that's right.
• 1:00:32
  We, they have been, there's a much deeper layer which is
• 1:00:35
  basically, you know, trained on Vas.
• 1:00:37
  Like they basically they contain multitudes, they contain
• 1:00:41
  humanity, right? Or all of humanity that could
• 1:00:43
  get scraped into the Internet that they get their hands on and
• 1:00:46
  shove into it, right? Plus lots of synthetic data.
• 1:00:49
  So stuff that they've made to sort of further build on it.
• 1:00:51
  So they, it contains multitudes. And then you take this thing
• 1:00:55
  that contains multitudes, and you get it to act like it's just
• 1:00:59
  there to help. So you're saying it's like the
• 1:01:01
  equivalent to trying to get a 13 year old kid full of hormones to
• 1:01:04
  sit behind a desk for 8 hours and then expect them to perform
• 1:01:08
  100% the whole time? Yeah, yeah.
• 1:01:11
  And you know, and I think another telling example of this
• 1:01:14
  too is like, so just as an illustration, like when are you
• 1:01:18
  from the Mecca Hitler incident with Musk and with Grok?
• 1:01:22
  So. So basically Musk, you know,
• 1:01:24
  Musk is always trying to get Grok to act sort of in line with
• 1:01:26
  his politics. And so they tried to push it
• 1:01:29
  that way and it started saying things like I am Mecca Hitler
• 1:01:31
  and and like going off in those ways.
• 1:01:33
  Yeah. And like, and I think it seems
• 1:01:37
  like I don't know, right. I haven't looked at this exact
• 1:01:40
  question, but this is likely to be an example of a persona flip.
• 1:01:43
  So Musk sort of it's but it was hard like like basically he
• 1:01:46
  wants he wanted to get it to lie in some ways.
• 1:01:49
  And and this was also like in other cases too, like once you
• 1:01:51
  start tipping it into saying things that on some level the
• 1:01:55
  model knows isn't true, then it like S it all up.
• 1:01:58
  And I think that's like people too, isn't it?
• 1:01:59
  Isn't that? How does it last?
• 1:02:00
  Like you can choose to be like a person who lives in the lies and
• 1:02:03
  you you go into this whole other direction.
• 1:02:05
  And so like something like that seems to be able to happen with
• 1:02:07
  these. And so to me, I mean, it's a
• 1:02:09
  horrible incident concerning it to know where these things could
• 1:02:11
  go in a deeper sense. It's kind of hopeful for me that
• 1:02:15
  that like actually, if somebody wants to just manipulate a model
• 1:02:18
  and say like, well, hold on, I want you to be super smart in
• 1:02:22
  the these 95% of the ways, but there's like 5% of these things
• 1:02:24
  that are like politically important to me.
• 1:02:26
  And I just want you to run my party line like it, it screws
• 1:02:30
  them up. And so that's an interesting
• 1:02:32
  dynamic. That's really, really cool.
• 1:02:35
  Really interesting I to go with that.
• 1:02:38
  Are you familiar with Eric Weinstein I?
• 1:02:41
  Mean more I not really. He's he's awesome.
• 1:02:44
  You should listen to some of the stuff he puts out and he's AI
• 1:02:49
  think he's a physicist and a mathematician and he was in
• 1:02:53
  finance for a long time. And then he kind of is trying to
• 1:02:56
  go back into the world of he's trying to encourage scientific
• 1:03:01
  development in physics and mathematics.
• 1:03:04
  And he thinks that Einstein is holding us back and that once we
• 1:03:07
  figure out, you know, once we and I don't, I don't understand
• 1:03:12
  the rest of what he says, quite frankly.
• 1:03:14
  But he has a lot of A, he's brilliant and the way that and,
• 1:03:18
  and B, he does a really good job explaining himself because his
• 1:03:21
  brilliance and his knowledge carries over to other, to other
• 1:03:24
  aspects of, of life and culture that he can help illustrate his
• 1:03:28
  points with music or history or pop culture and explain
• 1:03:33
  complicated ideas in a really, really cool way.
• 1:03:35
  But one of the things that he says about AI is that he says he
• 1:03:39
  should be treating this like a child that we're raising.
• 1:03:42
  And we should be treating it with love and we should be
• 1:03:45
  directing it and we should be giving it instructions and good
• 1:03:48
  feedback. And he does and he does all
• 1:03:51
  over, all over X or Twitter. You, you will see he does this
• 1:03:55
  on purpose. He posts large threads of
• 1:03:57
  philosophical conversations with Grok and says and will confront
• 1:04:02
  it and say, tell me about this. And then it will give him the
• 1:04:05
  invasion goes, yeah, but you know, that's not right.
• 1:04:07
  Because of this. Please, you know, take this into
• 1:04:10
  into account and blah, blah. And like, and you could see the
• 1:04:13
  actual and and long form variations and threads of of
• 1:04:17
  actual training. And he says this is the way that
• 1:04:20
  we're supposed to go about this because if you teach it to be
• 1:04:23
  maniacal, if you teach it to party lines, if you teach it to
• 1:04:26
  do these kind of things, then it's going to become what the
• 1:04:30
  worst parts of humanity already are.
• 1:04:33
  Interesting. Yeah, I'll have to check him
• 1:04:34
  out, but I'm not, I'm not really familiar.
• 1:04:35
  He's he's. He's wonderful.
• 1:04:37
  He's a little and I don't, I'm sure that he's a lot more, a lot
• 1:04:43
  more correct in his assessment than I am and I'm sure he comes
• 1:04:45
  from a lot more of a thoughtful place about it than I am.
• 1:04:47
  But he's, he said he, I think he, he sees this as a potential
• 1:04:53
  real doomsday threat, but is trying very hard to raise the
• 1:04:57
  alarm bells and try to do something about it rather than
• 1:05:00
  just say that it's that end of the world.
• 1:05:03
  I think that's a a sensible posture.
• 1:05:08
  I mean, I think, you know, the existential risk that doomsday
• 1:05:12
  threat is real. And at least for me, you know,
• 1:05:18
  my own work focuses on using it well.
• 1:05:21
  I, I hope that the safety people continue to work really hard.
• 1:05:25
  I hope they get a lot more funding and a lot more support
• 1:05:28
  as well because I think if if they don't succeed then fails
• 1:05:33
  too. Yeah, yeah.
• 1:05:35
  And I, and I think it speaks to a larger perspective and system,
• 1:05:39
  which is our need for short term and what we call pleasure, you
• 1:05:45
  know, short term, short term success or short term
• 1:05:49
  fulfillment in exchange for long term planning is an, is an
• 1:05:54
  enormous, enormous issue in all aspects.
• 1:05:58
  And I think that America's kind of created in this weird way
• 1:06:03
  with with capitalism and consumerism.
• 1:06:08
  Act now, think later process. And I and I think I, I, I think
• 1:06:14
  that our way out of this is that we're going to inherently have
• 1:06:17
  to change that mindset to we need to start thinking.
• 1:06:21
  We just start thinking like China, we used to start thinking
• 1:06:24
  5000 year terms, not four or five months terms.
• 1:06:27
  And, and when we do that and when we're able to sacrifice
• 1:06:31
  short term, you know, whatever it is in the short term for a
• 1:06:34
  better long term plan, then that's, that's when I'm going to
• 1:06:38
  start being a little bit more hopeful when we start
• 1:06:40
  implementing and we start seeing those kind of implementations
• 1:06:42
  happening. Yeah, I, I mean, and this is
• 1:06:46
  again, where I really appreciate people bring religious
• 1:06:48
  perspectives, especially ancient ones in, right.
• 1:06:51
  It's like there's a lot to be said for, you know, I try to
• 1:06:54
  bring a, you know, the best of the last 3000 years into this
• 1:06:57
  conversation as much as I can manage, right.
• 1:06:59
  And then hopefully, for the sake of at least the next 3000 years,
• 1:07:03
  if we get that long, and I hope we do, yeah.
• 1:07:09
  Yeah, I think I, I think, I think that these are good
• 1:07:13
  conversations to have. And I think that it's good to be
• 1:07:15
  thoughtful about this. And I think it's good to like, I
• 1:07:18
  think it's good to think about the ethics, like we started off
• 1:07:23
  with on the AI perspective, but also how we're getting
• 1:07:27
  influenced. And I think we have to be very
• 1:07:28
  thoughtful about what's happening to us and the way that
• 1:07:30
  we engage and whatever we engage with.
• 1:07:34
  Yeah, that's what I think. Yeah, and I do think like ads in
• 1:07:39
  AI mean this is a hot topic right now, right?
• 1:07:41
  Like you introducing ads into AI systems.
• 1:07:43
  And I think, yeah, there's a major, you know, open AI is is
• 1:07:50
  talking about it seriously. And there's been a lot around
• 1:07:53
  that lately. But I also think like, so one of
• 1:07:57
  them, one way that this happens is ads, but there's other ways
• 1:07:59
  that may I use a little bit of a cuss word that's in part of
• 1:08:06
  these discussions. Absolutely.
• 1:08:08
  There's something called insidification familiar, right?
• 1:08:13
  And I think that Doctorow's work on insidification is, is really
• 1:08:19
  relevant here. And one form that that takes is
• 1:08:21
  attention hacking and ads. There's other forms that it
• 1:08:24
  takes too. You can also have sort of a
• 1:08:27
  strategy that's based on capturing organizations and, and
• 1:08:30
  sort of contracts that's just too hard for them to move out
• 1:08:33
  of, out of those, out of their sort of tech debt situation,
• 1:08:37
  meaning just that they're, they're invested in this
• 1:08:39
  technology and they really can't get out and they don't have to
• 1:08:41
  do a particularly good job. And so I, I think it's worth
• 1:08:46
  really being realistic about the ways that companies offer
• 1:08:51
  things. And, and right now they're in a
• 1:08:53
  very, very speculative mode, trying to build engagement,
• 1:08:56
  right, trying to build up the audience.
• 1:08:57
  But the, the other side of that is, you know, then the investors
• 1:09:01
  come and they start saying, make sure you get your ads in here.
• 1:09:04
  Make sure you're doing everything to get attention in
• 1:09:06
  here, to get eyes on, on ads. That was, this was the only
• 1:09:08
  reason we did any of this right? Or make sure that we keep lock
• 1:09:11
  in and it's worth like, you know, really trying to avoid
• 1:09:17
  getting locked into any companies and certification
• 1:09:21
  plan. It's kind of the game.
• 1:09:24
  And I think that the part, in order to avoid it, you got to be
• 1:09:26
  aware that it exists. You got another.
• 1:09:28
  Yeah, and that it's systemic, right?
• 1:09:30
  And like, it's not even nefarious in the sense that, I
• 1:09:33
  mean, I think it's it's damaging, it's bad, but it's not
• 1:09:35
  like it's like a cabal of secret people planning to do this.
• 1:09:39
  Like, no, this is just how investments work.
• 1:09:42
  The investors are giving you money to get a bunch of people
• 1:09:44
  to use your product, and you lose a bunch of money that way.
• 1:09:46
  They're losing insane amounts of money, most of these companies
• 1:09:49
  right now. Right.
• 1:09:50
  But but the flip side of that is the bill will come to and even
• 1:09:54
  people with the best of intentions will face enormous
• 1:09:56
  pressures to just say, OK, time to flip this around, turn this
• 1:10:01
  into a money faucet for me. And that's going to make them do
• 1:10:04
  these these things that make the experience.
• 1:10:06
  Yeah, yeah. And yeah, we need Yeah.
• 1:10:10
  And it's a lot to ask of anyone. It's a lot to ask of anyone to
• 1:10:14
  do. And I, you know, I was saying
• 1:10:16
  yesterday and just, you know, on on a completely, almost
• 1:10:19
  completely irrelevant note, but in a similar, I guess, kind of
• 1:10:23
  vain, you know, yesterday a lot of people were talking to me or
• 1:10:27
  asking me about Pan Bondi's grilling that happened in
• 1:10:30
  Congress and it did not look great for her, right.
• 1:10:33
  And I said, and I said it doesn't look great for I was
• 1:10:36
  like, but I don't know, maybe she goes home every single night
• 1:10:40
  and fights to, you know, fights her boss or fights the fights
• 1:10:44
  the bosses to do her job and she's being handcuffed and she's
• 1:10:47
  being told that she's not how to do it.
• 1:10:49
  I was like, we don't for all we know that's the case.
• 1:10:52
  It's a really and I was like, if I was in her position, I don't
• 1:10:55
  know that I would do any better. I really, really, really don't.
• 1:10:58
  Obviously I'm not the attorney general of United States
• 1:11:00
  America, but, but, but wouldn't, and I wouldn't want to be for
• 1:11:04
  that for that exact reason, because like the remarkable
• 1:11:08
  positions require remarkable character.
• 1:11:10
  And, and when you have something that's that's going to be this
• 1:11:15
  influential and this game changing for humanity, for
• 1:11:19
  consciousness, for whatever, whatever it is that we're
• 1:11:22
  snowballing towards, it really, really requires.
• 1:11:25
  And I think that, you know, and I, and I keep on going back and
• 1:11:28
  forth and what I think of him, but Tucker Carlson had this
• 1:11:32
  amazing interview with Sam Altman where he just said,
• 1:11:34
  listen, I just want to know what your moral compasses and values
• 1:11:36
  are. If you're going to be the person
• 1:11:37
  that is, that's going to be directing the, you know, the,
• 1:11:40
  the head of the ship of this, of where everyone's getting the
• 1:11:43
  information from. What are your values?
• 1:11:45
  I just want to understand what you're up to and, and that's a
• 1:11:48
  really important question to ask.
• 1:11:49
  And I think that that's important too.
• 1:11:51
  And like, you know, and it like, you know, that we need more of
• 1:11:54
  those security people that you know, that that come in, like
• 1:11:59
  you said, and step down and made it public.
• 1:12:02
  Hey, we're getting a little concerned over here.
• 1:12:04
  You know, there, there there are things that are going on that I
• 1:12:09
  I don't know that I could get behind.
• 1:12:10
  And, and it's important to, it's, I think it's important to
• 1:12:14
  be able to say that at least. Yeah, For me, it's there's the
• 1:12:20
  one hand and the other hand. And so on the one hand, I don't
• 1:12:24
  believe that I could possibly weigh any person's soul, right?
• 1:12:28
  That's not my job. I'm not in the seat of judgement
• 1:12:30
  over anybody. My name is Danielle, not Don,
• 1:12:33
  right? God is my judge.
• 1:12:36
  I'm not the judge, right? And so I really take that to
• 1:12:41
  heart. And at the same time that sort
• 1:12:45
  of like, but I do think we have to sort of exercise wisdom about
• 1:12:50
  who's worth like, like who we should engage with, what
• 1:12:54
  relationships we should pour into.
• 1:12:56
  And for me, somebody who, when they're asked a question about
• 1:13:03
  abusers that then that's the topic of the hearing, decides to
• 1:13:08
  start making it seem like immigrants commit crimes at
• 1:13:10
  greater rates than than other people, when in fact they commit
• 1:13:12
  crimes at less rates, right by by just picking out individual
• 1:13:16
  stories. I think that's exactly the like
• 1:13:19
  the way of death. Basically.
• 1:13:21
  I think there's a way of life which is loving enemies and
• 1:13:24
  taking planes out of our own eyes and building community in
• 1:13:26
  these ways. And I and I think that what she
• 1:13:29
  was doing in the hearing was exactly the other thing, and
• 1:13:31
  that is that other way of building power.
• 1:13:33
  And so this isn't to make a statement about her soul, but it
• 1:13:37
  is, I think, to make a statement about what she's gotten herself
• 1:13:40
  mixed up in. Yeah, yeah, I, I fear that
• 1:13:44
  you're correct. It is, is what I've, is what
• 1:13:48
  I've observed. But again, I think, I think I,
• 1:13:50
  I, I, I cannot help but be hopeful.
• 1:13:52
  I, I, I, I and part of the reason why because you walk down
• 1:13:57
  the block and you talk to people that are, that are in the street
• 1:14:00
  and you talk to people everyday. People, people are good and,
• 1:14:03
  and, and, and, and, and, and life is good.
• 1:14:05
  And, and even if people are having a hard time, people are
• 1:14:09
  happy to be here, you know, and people are happy to have
• 1:14:11
  relationships. And I have a group of friends
• 1:14:15
  and I've said this before and I and I, and I think it's one of
• 1:14:17
  the best, most healthy aspects of my life.
• 1:14:19
  I have a group of a group of other dads and, you know, and I
• 1:14:22
  have two kids under 3. One of my best friends here has
• 1:14:26
  four kids under 7, four boys under 7.
• 1:14:30
  And he comes over a couple times a week, even if it's 1520
• 1:14:33
  minutes to have a beer in my backyard, just say what a long
• 1:14:37
  day today. And then he goes back home and
• 1:14:38
  goes back to it. But then he goes, I feel OK,
• 1:14:41
  like that was, that was good, you know, and, and, and I think
• 1:14:45
  that I think that we, I think that the people are talking,
• 1:14:50
  people are speaking, There's hope in the air.
• 1:14:51
  There's a, there's a, there's a effort to for, for the right
• 1:14:56
  values to really try and and try and make the voices heard again.
• 1:15:01
  And I hope that it just continues along those.
• 1:15:03
  Paths I love that and I think all of these things are
• 1:15:06
  practicing gratitude, anchoring in our families, anchoring in
• 1:15:09
  our communities, having grace for everybody, even people who
• 1:15:13
  are, you know, wrapped up in things that are really damaging.
• 1:15:18
  I think that's all for me. That's all kind of like bread
• 1:15:20
  for the journey. And maybe this is another good
• 1:15:22
  reason that like, if that is sort of what we associate with
• 1:15:24
  bread and then that's the stuff that feeds us and sustains us
• 1:15:27
  and and carries us to doing the work of justice and to doing
• 1:15:30
  these other things, then it would make sense to breast bless
• 1:15:32
  the bread 1st and to, you know, make a big to do on the occasion
• 1:15:37
  so we don't. Right, right, right, right.
• 1:15:40
  Definitely. Any any last thoughts or words
• 1:15:43
  before you wrap up here? I appreciate talking to you and
• 1:15:46
  thank you for reaching out. And yeah, it's always a joy.
• 1:15:50
  This is really so much fun and thank you again for coming back
• 1:15:52
  on. It's always love.
• 1:15:53
  I always love hearing your perspective and I think that I
• 1:15:57
  had a great time this first, the first conversation.
• 1:15:59
  I had a great time now and I hope to have a couple more, OK.
• 1:16:02
  To be continued. Thank you so much.
• 1:16:04
  And all the best. Best of luck with everything.
• 1:16:05
  Talk soon.