Latest / Today’s AI News / Altman Calls for Robot Taxes and a Wealth Fund, The New Yorker Investigates Sam, Meta’s Models

Transcript

Altman Calls for Robot Taxes and a Wealth Fund, The New Yorker Investigates Sam, Meta’s Models · Today’s AI News

• 0:00
  Welcome back to today's AI News, brought to you by Nynex
• 0:03
  Productions. Imagine a brilliant inventor
• 0:05
  building a money printing machine, but he's standing
• 0:08
  outside the factory begging the government to tax it before it
• 0:11
  breaks the economy. That is the wild paradox we were
• 0:15
  looking at today with the future of AI.
• 0:18
  It's like constructing a rocket ship while frantically trying to
• 0:21
  rewrite the laws of gravity at the exact same time.
• 0:24
  Yeah, that is, I mean, that's really the perfect way to frame
• 0:27
  what we're looking at today. Right, so we have an incredibly
• 0:29
  dense journey mapped out for you in this deep dive.
• 0:32
  First, we are going to explore a tech billionaires sweeping new
• 0:36
  social contract designed to, well, save the economy from his
• 0:40
  very own technology. Which is fascinating on his own.
• 0:43
  Exactly. And then we'll pivot into some
• 0:45
  newly surfaced, highly critical memos about his past.
• 0:48
  We're also going to examine Meta's public struggle to
• 0:51
  release a competitive new model and then deliver a barrage of
• 0:54
  rapid fire global news detailing how AI is like, violently
• 0:59
  breaking into the physical world.
• 1:01
  And finally, we'll breakdown a community workflow that shows
• 1:03
  how one single person is collapsing an entire production
• 1:07
  agency into a solo operation. When you lay all these sources
• 1:10
  out on the table, a stark picture really begins to form
• 1:15
  about the sheer velocity of this technology.
• 1:18
  We are looking at a timeline that is accelerating far faster
• 1:23
  than our traditional social and economic structures were really
• 1:26
  designed to handle. Yeah, so let's start right at
• 1:28
  the top with the most sweeping development on the board.
• 1:31
  Open AI just dropped a 13 page policy document that reads less
• 1:35
  like a standard corporate memo and honestly much more like a
• 1:38
  legislative blueprint for a new civilization.
• 1:41
  It really does. Sam Altman is explicitly pushing
• 1:43
  the United States government to a up a brand new social
• 1:46
  contract. And you know the reason for this
• 1:48
  urgency is a specific phrase buried in the document.
• 1:52
  They state that the transition to artificial super intelligence
• 1:55
  or ASI, that it has already begun.
• 1:57
  That phrasing is a massive paradigm shift because for
• 2:00
  years, you know the industry talk about artificial super
• 2:02
  intelligence is this distant theoretical horizon.
• 2:05
  Right, like science fiction. Exactly.
• 2:07
  But now the organization leading the charge is stating that the
• 2:10
  transition is an active, ongoing event.
• 2:14
  And because they view this as an active transition, the policies
• 2:17
  they're proposing are just unprecedented in their scope.
• 2:20
  I mean, they are outlining A structural overhaul of the
• 2:23
  modern workforce. Yeah, we're talking about
• 2:25
  implementing a tax on AI driven profits and like a literal tax
• 2:29
  on robot labor, right? Plus he's proposing A mandatory
• 2:33
  4 day work week across the board.
• 2:35
  Which is huge. Yeah, and perhaps most
• 2:37
  intensely, they are asking for government level containment
• 2:40
  playbooks for rogue autonomous AI.
• 2:43
  But the specific proposal that really stopped me in my tracks
• 2:46
  was the wealth fund. Yeah, the Sovereign style wealth
• 2:49
  fund is really the centerpiece of this entire document.
• 2:52
  So the mechanism they're proposing is to create a massive
• 2:55
  national fund seated directly by AI companies.
• 2:58
  OK. These firms would pay dividends
• 3:00
  into the fund, which would then be distributed to every American
• 3:03
  citizen. The document explicitly draws a
• 3:06
  comparison to the Alaska Permanent Fund.
• 3:08
  Oh, right, with the oil. Exactly.
• 3:10
  In Alaska, a portion of the state's oil revenue is put into
• 3:13
  a constitutionally established fund, and then every eligible
• 3:16
  resident receives A yearly dividend check just for a living
• 3:20
  there. Altman is suggesting we do the
• 3:22
  exact same thing, but instead of extracting fossil fuels from the
• 3:26
  ground, the resource being extracted and monetized is
• 3:29
  artificial intelligence and automated labor.
• 3:32
  Axios actually went on record saying this is the most detailed
• 3:35
  blueprint any tech Titan has ever published regarding taxing,
• 3:39
  regulating and redistributing wealth from their own
• 3:41
  technology. So what does this all mean?
• 3:44
  Because I really have to push back here.
• 3:45
  It feels like a glaring contradiction.
• 3:47
  You know How so? Well, why is a tech CEO, the
• 3:49
  head of a company valued at nearly a trillion dollars,
• 3:52
  asking the infamously slow moving Gears of Washington to
• 3:55
  fix a problem he is actively racing to create?
• 3:58
  Right. Yeah.
• 3:59
  I mean, he's building the exact technology that he openly warns
• 4:02
  will break our economic system and then essentially saying to
• 4:06
  the government, hurry up and catch me.
• 4:08
  It feels like watching a ticking clock where government
• 4:10
  regulation speed and AI development speed are running on
• 4:13
  2 totally different physics engines.
• 4:15
  Well, what's fascinating here is the underlying tension between
• 4:18
  incredible technological progress and, honestly, pure
• 4:22
  economic self preservation. OK, I'm back that.
• 4:25
  Altman isn't just asking for regulation to slow things down,
• 4:28
  he is trying to engineer a soft landing for society because his
• 4:32
  business model absolutely depends on it.
• 4:35
  If AI displaces labor at the scale open AI is predicting, the
• 4:38
  consumer base collapses. I mean, people without jobs
• 4:42
  cannot buy goods, they cannot use services, and the economy
• 4:45
  just grinds to a halt. That makes sense.
• 4:47
  Right, so he's arguing for a right to AI access for all
• 4:51
  citizens because if you don't redistribute the wealth
• 4:54
  generated by ASI you end up with a microscopic class of hyper
• 4:58
  wealthy tech owners and a massive economically paralyzed
• 5:02
  public. It is utopian policy, sure, but
• 5:05
  it's driven by mathematical necessity.
• 5:07
  He needs the economy to survive the arrival of his own product.
• 5:10
  OK, let's untack this. Because if we are talking about
• 5:14
  handing over the keys to our entire economic system and
• 5:17
  fundamentally altering how wealth is distributed, we really
• 5:20
  need to look at the architect of this plan.
• 5:22
  Yeah, definitely. We just explored this grand
• 5:24
  utopian vision for society. But to truly understand, we have
• 5:28
  to look at the leader himself. And the sources provided a
• 5:31
  highly critical investigative report that just dropped in The
• 5:34
  New Yorker. What exactly did they find
• 5:36
  there? So this investigation was built
• 5:38
  on over 100 interviews, and it brings some very troubling,
• 5:41
  previously unseen documentation into the public eye.
• 5:45
  The most significant revelation involves secret memos from
• 5:48
  Opening Eyes, former chief scientist Ilya Sutskever.
• 5:51
  We are talking about 70 pages of internal Slack messages and
• 5:55
  human resources documents. Wait really?
• 5:58
  So this isn't just hearsay or water cooler gossip, they
• 6:01
  actually have the internal HR documents.
• 6:03
  They do? Yeah.
• 6:04
  And the documentation alleges that Altman consistently
• 6:07
  misrepresented safety protocols to the company's board of
• 6:10
  directors. Wow.
• 6:11
  Furthermore, it wasn't an isolated concern.
• 6:14
  Dario MO Day, another major figure who left to found
• 6:18
  Anthropic, kept private notes for years that independently
• 6:21
  reached the exact same conclusion.
• 6:23
  One of his notes literally read quote.
• 6:25
  The problem with open AI is Sam himself.
• 6:28
  That is heavy. It is, and the New Yorker's
• 6:30
  reporting traces this behavior back through his entire career.
• 6:33
  It details historical friction at his early location base
• 6:36
  startup Looped, and it documents instances where his partners Y
• 6:39
  Combinator actively tried to push him out.
• 6:41
  Really. Yeah, paints a picture of
• 6:43
  someone who is a masterful networker but highly
• 6:46
  manipulative behind the scenes. And it gets even darker when you
• 6:49
  look at the quotes from industry peers.
• 6:51
  In that piece, a Microsoft executive actually went on the
• 6:54
  record with the reporters and stated that there is a small but
• 6:57
  real chance Altman is remembered as a Bernie Madoff Sam Bankman
• 7:03
  freed level scammer. Right, which is quite the
• 7:05
  statement. Exactly.
• 7:07
  Now, to be clear, the report admits there is no absolute
• 7:10
  smoking gun here. Nobody is pointing to a specific
• 7:13
  crime. But the sheer volume of these
• 7:15
  allegations creates a distinct pattern.
• 7:17
  It really does. It makes the CEO of the world's
• 7:20
  most important AI company one of the most polarizing figures on
• 7:23
  the planet, balanced only by his, like, fiercely loyal
• 7:26
  supporters. It's like we're taking financial
• 7:28
  planning advice for the entire globe from someone who critics
• 7:31
  say might be running a sophisticated shell game.
• 7:34
  Do you trust the brilliant math or do you fear the man?
• 7:37
  And if we connect this to the bigger picture, it fundamentally
• 7:40
  complicates the social contract. We were just talking about the
• 7:43
  policies Altman is proposing, taxing robot labor, creating
• 7:47
  sovereign wealth funds, giving the government containment
• 7:50
  protocols. They require an unprecedented
• 7:52
  level of absolute trust between the public, lawmakers and the
• 7:56
  private tech sector. Oh, absolutely.
• 7:58
  If the leader advocating for these changes is viewed by his
• 8:01
  own former chief scientist as someone who actively deceives
• 8:04
  his board regarding safety, how can lawmakers trust his
• 8:07
  containment playbooks for rogue AI?
• 8:11
  This dichotomy between the visionary savior and the
• 8:13
  deceptive operator is the absolute center of the AI safety
• 8:17
  debate right now. You just cannot separate these
• 8:19
  massive social policies from the character of the person writing
• 8:23
  them. Yeah, that is a deeply
• 8:24
  unsettling thought to hold onto as we watch this whole industry
• 8:27
  accelerate. And Speaking of acceleration,
• 8:29
  while Open AI is dealing with these existential wealth funds
• 8:32
  and intense leadership dramas, we're seeing other tech giants
• 8:35
  struggling to even stay in the race.
• 8:37
  Let's pivot to meta. Right Meta.
• 8:39
  They are currently preparing to ship the first models from their
• 8:42
  new super intelligence team, which is led by Alexander Wang.
• 8:45
  But the runway has been incredibly bumpy for them.
• 8:48
  Their new model, which is codenamed Avocado, was delayed
• 8:52
  back in March because it's benchmark performances fell
• 8:55
  noticeably short of rivals like Open AI and Anthropic across the
• 8:59
  board. Now they're pivoting to a hybrid
• 9:02
  consumer focused approach. They plan to release both open
• 9:06
  and closed models for their apps, but they are keeping their
• 9:09
  largest, most capable models closed.
• 9:12
  Axios reported that Meta is fully aware these models aren't
• 9:15
  fully competitive yet, but they believe they have niche consumer
• 9:18
  strengths. See, I have to push back on
• 9:20
  Meta's strategy here. If you know your tech isn't the
• 9:23
  best, why release it right now? Releasing Avocado in this state
• 9:27
  sounds like putting a concept car on the track that you know
• 9:30
  can't win the race just to prove to your investors that you still
• 9:33
  know how to build engines. Yeah, I see what you're saying.
• 9:35
  It feels like a defensive PR move rather than an actual
• 9:38
  technological breakthrough. That is a fair critique of the
• 9:41
  technology itself, sure, but you have to look at the mechanism of
• 9:44
  distribution here. Meta operates WhatsApp,
• 9:47
  Instagram and Facebook. They have billions of daily
• 9:51
  active users. Sure, if they embed an AI model
• 9:54
  directly into the search bar of Instagram, people are going to
• 9:57
  use it simply because it is already there, even if it is,
• 10:00
  say, 10% less capable than ChatGPT.
• 10:04
  Meta is prioritizing ecosystem lock in over frontier dominance.
• 10:10
  They need user data and they need to train their user base to
• 10:13
  interact with Meta AI daily. Because right now, everyday
• 10:16
  people are already using accessible models to do
• 10:19
  incredible things, and Meta simply can't afford to let them
• 10:22
  build those habits on competing platforms.
• 10:24
  Yeah, it makes complete sense why Meta is pivoting to consumer
• 10:28
  apps when you look at what everyday users are already doing
• 10:30
  with the tools available right now.
• 10:32
  The sources gave us a brilliant look at a practical workflow
• 10:35
  using Perplexity Deep research. And the best part is, this is a
• 10:39
  hidden version of Perplexity Computer that is actually
• 10:41
  available on their free plan. Right.
• 10:43
  It gives you up to five queries a day, and the mechanism behind
• 10:47
  this tool is what makes it so powerful.
• 10:49
  You aren't just chatting with a bot.
• 10:51
  You open perplexity, switch to deep research mode, and paste in
• 10:55
  a prompt with your raw business idea.
• 10:57
  Then you literally step away for 5:00 to 6:00 minutes.
• 11:01
  Just walk away. Yep.
• 11:03
  Behind the scenes, the AI is conducting parallel web
• 11:06
  searches. It's reading through dozens of
• 11:07
  sources, cross referencing market data, structuring A
• 11:10
  logical argument, and automatically outputting a fully
• 11:14
  designed slide deck analyzing the feasibility of your idea.
• 11:18
  And the pro tip they shared is what makes this so actionable
• 11:21
  for you listening. You don't just run it once you
• 11:23
  build variants. You ask it to generate A6 slide
• 11:26
  pitch deck for potential Co founders and then you run
• 11:28
  another query asking for a 90 day minimum viable product
• 11:31
  rollout plan. Which is amazing.
• 11:33
  It really is. You could take every half baked
• 11:35
  shower thought or business idea you've had over the last year,
• 11:38
  run them through the system every Saturday morning, and burn
• 11:41
  through a year's work of evaluation in a single month.
• 11:44
  We are seeing everyday people gain CEO level research and
• 11:47
  strategy capabilities from free off the shelf tools.
• 11:51
  I mean the democratization of this kind of analytical power
• 11:54
  fundamentally shifts who gets to be an entrepreneur.
• 11:57
  You no longer need a team of analysts to validate a market,
• 12:00
  you just need the right prompt. It really is shifting
• 12:03
  everything, and that kind of democratization is exactly why
• 12:06
  the industry is moving at warp speed.
• 12:09
  So let's look at how this is breaking into the real world
• 12:12
  right now. We're going to transition into a
• 12:13
  rapid fire. Look at the astonishing ways AI
• 12:16
  is breaching the boundaries of software.
• 12:18
  Sounds good. Starting in the corporate
• 12:20
  sphere, the Information is reported that Sam Altman and
• 12:23
  Open AI Chief Financial Officer Sarah Fryer are actively
• 12:26
  clashing over the timing of the company's initial public
• 12:28
  offering. Open AI has officially denied
• 12:31
  this claim, but the rumors of internal friction over going
• 12:34
  public are really persisting. OK, moving to global
• 12:37
  geopolitics. And we are delivering this
• 12:39
  impartially, literally just reporting the facts on the
• 12:41
  ground. Iran's military has published
• 12:43
  satellite footage of the Stargate data center.
• 12:46
  This is the proposed $30 billion supercomputer facility located
• 12:50
  in Abu Dhabi. Iran's military has explicitly
• 12:53
  singled it out as a target, publicly vowing to destroy US
• 12:56
  infrastructure across the region.
• 12:58
  Wow. And then shifting to healthcare,
• 13:00
  we are seeing staggering adoption metrics.
• 13:03
  Open AI is head of finance Chang Peng MU revealed that ChatGPT is
• 13:06
  now receiving 2 million insurance related messages every
• 13:10
  single week. Because the American healthcare
• 13:12
  system is so notoriously complex, millions of people are
• 13:15
  basically bypassing traditional customer service and using AI to
• 13:19
  decipher their coverage. And staying with healthcare here
• 13:21
  is a massive milestone. Legion Health just won state
• 13:24
  approval for its AI app to directly refill psychiatric
• 13:28
  medications. Yeah, imagine logging into your
• 13:30
  patient portal and it's not a human doctor reviewing your file
• 13:34
  to decide if you get another month of your psychiatric
• 13:36
  medication. It is an algorithm.
• 13:39
  It is authorized to process these refills without any
• 13:42
  clinician oversight whatsoever. That is wild.
• 13:44
  Finally we have a wave of tech releases.
• 13:46
  Google dropped AI Edge Eloquent which is a fully offline iOS
• 13:51
  dictation app, meaning your voice data never leaves your
• 13:54
  phone. Netflix has open sourced a tool
• 13:57
  called Void, which is a physics aware AI for video editing.
• 14:01
  It actually understands gravity, light and motion, allowing
• 14:04
  editors to manipulate video without breaking the natural
• 14:06
  physics of the scene. That's cool.
• 14:08
  Yeah, and Microsoft upgraded its image generation capabilities
• 14:11
  with Mai Image 2, pushing the boundaries of realism to a point
• 14:15
  where distinguishing it from a photograph is becoming,
• 14:18
  honestly, nearly impossible. OK, wait.
• 14:20
  Here's where it gets really interesting, because we just
• 14:23
  went through a rapid fire list. But we have to stop and look at
• 14:26
  the sheer shock of these juxtaposition.
• 14:29
  On one hand, we have an AI that has been authorized by the state
• 14:32
  to independently control psychiatric medication refills
• 14:36
  for everyday people. And at the exact same moment,
• 14:39
  the massive physical data centers housing these very same
• 14:43
  AIS are literally being targeted by foreign militaries.
• 14:47
  The digital stakes and the physical stakes are violently
• 14:50
  colliding right in front of us. It raises an important question
• 14:53
  about containment. What we are witnessing is the
• 14:56
  moment AI completely breached the containment of the digital
• 14:59
  world. It is no longer just text
• 15:01
  generating on the screen. It is actively dictating
• 15:04
  military strategy, global infrastructure and direct human
• 15:08
  biological health. When a server farm in Abu Dhabi
• 15:11
  becomes a primary military target, it tells you that the
• 15:14
  algorithms inside it are now considered critical global
• 15:16
  assets. The lines between a standard
• 15:19
  tech company, a healthcare provider and a geopolitical
• 15:21
  superpower have completely blurred.
• 15:24
  It is a lot to process, so let's take a breath from the intense
• 15:28
  global news and bring this all back down to the individual
• 15:30
  level. How do you, listening to this
• 15:34
  right now, actually harness this chaotic power?
• 15:37
  The sources provided an amazing look into community AI workflows
• 15:40
  showcasing a reader named Wayne F from the UK.
• 15:44
  He acts as a one person e-learning agency and the way he
• 15:47
  orchestrates his tools is really a master class in modern
• 15:50
  productivity. It really is because typically
• 15:52
  an e-learning agency requires an instructional designer to write
• 15:55
  the course, a graphic designer to create the visuals, and an
• 15:58
  auditor to ensure the educational standards are met.
• 16:02
  Wayne is replacing all of them by establishing highly specific
• 16:05
  frameworks. First, he uses Google's Notebook
• 16:08
  LM as a dedicated reference library to ingest all of his
• 16:12
  subject matter expert documents. Then, to ensure the AI doesn't
• 16:15
  just spit out generic text, he maps his prompts to Cold's
• 16:19
  learning cycle. It's a psychological framework
• 16:21
  that takes a student through 4 distinct stages, experiencing a
• 16:25
  concept, reflecting on it, thinking about the theory, and
• 16:28
  finally acting on it. He forces the AI to structure
• 16:32
  every single lesson through that human psychological lens.
• 16:35
  OK. And he also uses something
• 16:36
  called a five part pentad scenario.
• 16:38
  What exactly is that doing in his workflow?
• 16:40
  Yeah. So the pentad is another
• 16:42
  academic framework that breaks any situation down into five
• 16:45
  elements, The act, the scene, the agent, the agency, and the
• 16:49
  terpus. By feeding this structure into
• 16:51
  the AI, Wayne forces the model to fully understand the context
• 16:55
  of their training scenario before it writes a single word.
• 16:58
  Oh nice. It creates a deep multi
• 17:00
  dimensional lesson rather than, you know, just a flat summary.
• 17:03
  And he doesn't just let the AI run wild, he uses a very strict
• 17:06
  command driven workflow. He only triggers asset
• 17:09
  generation by typing specific command words like VEO or create
• 17:12
  image utilizing tools like Nano Banana and VO with 3.1 to
• 17:17
  generate the specific graphics and videos needed for that exact
• 17:20
  moment in the lesson. Exactly, and the final step is
• 17:23
  arguably the most important. He loops back around and uses
• 17:26
  perplexity combined with notebook LM to audit his own
• 17:30
  generated work. OK.
• 17:32
  He prompts the AI to grade the finished course against highly
• 17:35
  complex academic models like the QM rubric, which is a widely
• 17:39
  recognized standard for course design quality.
• 17:41
  That's brilliant. Right.
• 17:43
  He's using one AI to verify that the other AI he didn't
• 17:46
  hallucinate and actually met the educational standards.
• 17:49
  The result is profound. He has successfully collapsed an
• 17:53
  entire production team into one single pipeline.
• 17:56
  He is simultaneously acting as the creator, the designer and
• 17:59
  the lead auditor. Wayne isn't just using an app,
• 18:01
  he is conducting a digital orchestra where he plays every
• 18:05
  single instrument perfectly. It is absolute efficiency.
• 18:07
  And if we bring this all full circle, Wayne is living,
• 18:12
  breathing proof of the exact efficiency that Sam Altman
• 18:15
  Social Contract is trying to address when one person sitting
• 18:19
  in their living room in the UK can replace the output of a
• 18:22
  whole creative agency. The fundamental nature of work
• 18:25
  has already changed the transition to super intelligence
• 18:28
  that Open AI is warning about. Wayne is already navigating the
• 18:31
  early stages of it. It's an incredible realization
• 18:34
  to end on. You see the macro level of
• 18:36
  billionaire CE OS begging Washington for regulation, and
• 18:40
  then the micro level of a guy named Wayne automating an entire
• 18:43
  department from his laptop. Yeah, if everyday people like
• 18:46
  Wayne can already automate the jobs of entire creative agencies
• 18:50
  from their living rooms today, what happens to the global
• 18:53
  economy the day before Sam Altman Super Intelligence
• 18:55
  actually wakes up? Will that social contract catch
• 18:58
  us, or are we already in freefall?
• 19:00
  That is something you're going to have to think about today.
• 19:02
  Make sure you subscribe to stay updated on all of these rapidly
• 19:05
  moving stories and please take a moment to rate the show. 5 stars
• 19:08
  really helps us out. We'll see you back for
• 19:10
  tomorrow's episode.