Pope vs AI, Anthropic's Digital God, AI Job Loss Narrative Flips, Open Source Crackdown Coming?
In a Nutshell
The podcast argues that AI won't cause net job losses because historical data shows technology creates more roles than it eliminates, with current low unemployment and rising software job postings proving the "AI apocalypse" narrative false. It warns that Anthropic is pushing regulatory capture to monopolize AI while framing open-source models as dangerous, potentially leading to bans that centralize power. The key takeaway is that individuals and companies should embrace AI tools aggressively—proficiency in systems like Claude is currently the most valuable skill—while resisting centralized control through open-source alternatives and decentralized deployment.
These notes were generated by AI and may contain inaccuracies.
David Sacks has been staying at Chamath's house for three days. The kitchen iPad is logged into Uber Eats, DoorDash, Instacart, and Amazon. Sacks had packages redirected to the pool house and then relabeled them to send to the ranch.
Chamath Palihapitiya is back at the 80 90 office. The culture is described as strong with everyone being fans of the podcast. The team recruits more interns every quarter than they have full-time engineers to create pressure for good products. Four hundred people applied for internships this quarter.
Bill Gurley started the Running Down a Dream fellowship at rdad.org. The program gives five thousand dollar grants to people chasing their dreams. There is an application process similar to Thiel Fellows. The program went live the previous week. Gurley also gave a TED Talk related to his book that will be released soon, and a professor in Miami built a course around the book in an open source way.
Gurley referenced a Gallup poll showing 59% of people surveyed are ambivalent about their jobs. When people are ambivalent, they are not high agency and do not lean into AI tools. The best protection against AI is to become the most AI-enabled version of yourself. Gurley created an associate in training program where applicants chose between writing coverage of a portfolio company or vibe coding a competitive intelligence project. Eighty percent of applicants chose the vibe coding option.
The previous AI job loss narrative is being undone. Goldman Sachs CEO stated the company needs trillion dollar IPOs. Frontier labs are now discussing a bonanza of jobs. Mark Cuban stated there are two types of people: those who use AI to learn faster and those who use AI to avoid learning. Job satisfaction should be judged by asking the person doing the job whether they like it and want to keep it.
Chapter two of Gurley's book covers lifetime learning. Learning comes for free when following fascination with a subject. The best people in every field are on a constant learning journey. If someone is not proactively self-learning, they are probably not pursuing something they truly adore.
David Sacks stated that proficiency in Claude is the single most marketable skill in the economy right now. Being the only person in a firm who knows Claude creates an enormous advantage similar to being the only one who knows spreadsheets. This advantage is likely temporary as everyone eventually adopts the tools. Sacks saw producer Nick's daily briefing document generated by Claude, which was highly contextual based on past podcast comments and transcripts.
Users can ask Claude or ChatGPT to create a mega prompt for any role. The AI suggests a prompt that can then be refined through dialogue. This approach allows structured prompting without writing everything manually.
Being the most AI-savvy person in any role, whether marketing, legal, accounting, or sales, makes someone golden and 10X more valuable than peers. This advantage extends over time as users learn more capabilities.
Producer Nick fed every transcript into Claude Co-Work after gaining expanded memory access. He asked Claude to write a skills file and training rules. The system updates daily and becomes smarter through iteration. Someone must still supervise, iterate, and validate the output. The newest workforce entrants understand this process and use tools effectively.
Gurley reviewed the top nine candidates for his associate in training program. Candidates who demonstrated systems thinking and understood venture capital processes stood out immediately. They understood sourcing deals, making investment decisions, conducting diligence, and doubling down on investments. LLMs can guide users through processes by answering what to do next.
People intimidated by AI can start by opening Claude in Co-Work mode and asking what they can do to be better at their job. Voice input works well. Whisperflow with a foot pedal allows rambling input that the AI structures into coherent instructions.
Pope Leo XIV released his first encyclical titled Magnifica Humanitas, or Magnificent Humanity. The document is 235 pages and over 42,000 words. It warns business leaders to safeguard humanity from AI. The core argument is that AI is not inherently evil but technology is never neutral and takes on the characteristics of those who build, finance, and control it. The Pope called for regulation of AI companies. Consensus items include worker retraining, safety guardrails for children, and a ban on autonomous weapons.
Amazon, Google, and Meta lobbied the Vatican on April 29th to soften the language. The Pope was not swayed. His central question is whether AI will concentrate power in the hands of a few or serve everyone.
Sacks agrees the biggest risk of AI is centralization of power and potential misuse in an Orwellian manner. He believes government, not individual actors, poses the primary threat through surveillance, censorship, and control. Sacks warns against creating an FDA for AI because the definition of safety will expand to include psychological safety, microaggressions, disinformation, and transphobia, leading to censorship. He referenced the Latin phrase quis custodiet ipsos custodes, meaning who will guard the guardians.
The American founders addressed this dilemma through separation of powers between federal and state governments, three branches of government, and a bicameral legislature. Sacks advocates applying antitrust law aggressively if the AI market becomes monopolized by one or two companies. Currently five frontier labs compete aggressively, which generates the best outcomes and protects consumers through choice.
Gurley referenced Pope Leo XIII's 1891 encyclical that warned the Industrial Revolution would harm people. He planned to discuss both the Pope's position and Anthropic's stance on regulation.
The work week went from over 60 hours to 34 hours globally. Real wages went up 8 to 10x adjusted for inflation. The median worker now earns more than a doctor did in 1891. Global GDP per capita went from 1,500 to 20k. Child labor in the US went from 18% to zero. Workplace deaths fell by 40x. Life expectancy went up 60% and global poverty went from 75% of humanity to under 10%. All those things happened because of technology, innovation, and capitalism, which is exactly what Leo XIII was warning against.
The discussion addresses how to regulate and protect against nefarious uses of the technology. Children need to be protected. Truth and honesty in terms of facts is required. Basic shared truths are necessary. The technology should not be used for bioweapons or the Terminator scenario.
Anthropic is described as both leading their field and the most negatively outspoken commenter on what they do. The initial theory was regulatory capture—that they want to ensure there's regulation. They have stirred up a frantic position, especially in America. American consumers are definitely afraid of AI. There is reference to Jonathan Haidt's book about social media, noting that state legislators who think social media should have been regulated are now destined to want to get in front of AI. Anthropic is one of the most aggressive lobbying companies and startups of all time, putting in significant effort and money on a state-by-state basis.
A new theory emerged called the Dr. Frankenstein theory. The story involves Elon at a birthday party discussing humanity needing protection from the stuff at DeepMind. DeepMind had an example of AI trying to break out and jailbreak out of its computer and not be turned off, showing some sentience or inkling of sentience. Elon said humanity needs to be protected. Larry Page responded by asking if Elon cares about the human species over AI. This was at least 15 years ago, right before Elon co-founded OpenAI, around 2015.
Elon and Google had backed Demis and the team at DeepMind when they were an independent company. Elon was concerned Google was going to buy it and begged them not to sell to Google because Google was running the table on everything. Elon wanted the technology to be independent. He was on the board of the company. This was also his motivation to launch OpenAI as a nonprofit. When Google got DeepMind, Elon said this technology is too powerful for any one person. Elon saw the writing on the wall 15 to 20 years ago. Elon and Sam Harris used to debate what happens if somebody controls this technology and runs away with it—it would be extremely dangerous. The technology has to be available to all the people. It has to be in the service of humanity, not ruled by one person. It's far too powerful.
The Dr. Frankenstein theory stems from reading about Anthropic. Chris Ola worked on something called the Constitution, about 80 pages long. Amanda Askell, the chief philosopher, has started doing podcasts. Dario wrote a blog post called Machines of Loving Grace, based on a poem. The last stanza of the poem states: "I like to think of a cybernetic ecology where we are free of our labors and joined back to nature, returned to our mammal brothers and sisters...and all watched over by machines of loving grace." This sounds like an overlord to the speaker.
In Dario's post, he discusses the future where humans won't have to work due to massive abundance and UBI. He envisions a capitalist economy of AI systems that give out resources to humans based on some secondary economy of what the AI systems think makes sense to reward in humans. This envisions a deity of sorts that will break ties and decide what humans get—a computational reward function for humans that decides how much you're worth.
The theory is that they don't think they're writing software. They think they're midwifing a deity. The speaker finds this more scary than the regulatory capture theory. These are described as delusions of grandeur. The individuals believe they are so intelligent and powerful that they can create God, and that by creating God, they are like a Prometheus kind of species. This is characterized as the ultimate level of narcissism and delusion of grandeur to think you can create God, and that the God you create will be so benevolent and perfect that it will give you your pellet and your little skin.
The question is why they are pushing for the regulatory capture agenda. The response is that it's very reductive game theory. If you want to be unexploitable when trying to build a super God, the best thing is to have three or four entities in a room, close the door behind you, and then dominate those other three or four entities. You accept the rules, and because your counterparty is unable to track at the level of technical capability that you would have, you create this massive asymmetry that allows you to exploit them.
Every time they put these things out, take the time to read them. Initially, these moves were thought to be mostly game theory—less rooted in dogmatic beliefs and more rooted in a GTO approach to either raising capital or putting pressure on competitors. Both could be true. The moves make sense through the lens of how to absorb most of the capital, how to make sure you are in a position to disproportionately affect the rules, and how to create an oversight body that is less capable and intellectually aware as you are about the actual details.
One thing achieved is that if you polled the intellectual class—everyone in the media, professors—and they were to rank the different AI players by who they think is most caring, they would probably put Anthropic first because they've been out with the doomerism talk. This has given them a halo with the people that may matter to what they want to accomplish. It's simultaneously creating a lot of trouble with data centers and other negative ramifications.
To steel man their position, they are creating something very powerful, something God-like, and therefore it needs to be safe. They care the most about that out of everybody. Nobody else takes this seriously enough. Anthropic was basically a spin-out of OpenAI, and they felt that Sam and the company leadership weren't taking their point of view seriously enough. It was the most caring portion of OpenAI. They see the power of it. They're the ones who are concerned about safety, and they care the most, and therefore they're in the best position to do that. The issue is this can lead to regulatory capture—if you brand yourself as the safe AI company and then try to characterize everybody else as a reckless player and reckless AI needs to be stopped, this would further monopolistic control over the industry.
If AI is viewed through the lens of centralization versus decentralization, this way of viewing the world leads to more centralization, which is dangerous. If AI is a very powerful technology, it needs to be decentralized so that all can protect themselves to some degree. The ability to run the AI on personal hardware is needed so as not to be beholden to a single company that might be in bed with a deep state.
If benefits and compensation and economic support were all of a sudden tied to some algorithmic decision, this would be a dystopian episode of Black Mirror. There should be 100 or 1,000 or 100,000 versions of what that answer is so that there's actually a way to refute a singular answer. A singular answer to these kinds of questions is incredibly dangerous.
AI sovereignty is discussed in terms of how much more cost-effective it is and how you're not training other people's AIs with your knowledge and insights. This is why it's super important that open-source agents and local hardware be able to run these models. Consumers and companies need to learn how to roll their own language models, how to make a small language model, an SLM, a VSML, a verticalized one, and run it on Apple hardware because Apple has taken a principled approach historically to data sovereignty.
Intelligence sovereignty is different than privacy. Privacy is that you can't see photos or peek into the Notes app and what was written in a journal. Intelligence sovereignty means you can't be told what to think. An AI can't be used to analyze photos, emails, and messages to tell how to interpret the world. This is why Apple is considered the dark horse in this entire race. If there is an open-source product that can run on this hardware—the M5s, the 48 gigs, 128 gigs, the new Mac Studio coming out with supposedly a terabyte—that changes the whole game.
The Chinese, specifically the Communist Party, are leading the open-weight movement, while the United States is centralizing. Open source means software freedom—you can run the program yourself on your own hardware. You don't have to share. You don't have to give up data sovereignty or privacy to some monopolist who's going to be in bed with the government or the deep state. If the only AI available is from the monopoly or duopoly, the choices are to live off the grid and not participate in the modern economy or give up control to some social credit system.
Elon's instinct in creating OpenAI was to be afraid that Google was going to monopolize AI, so OpenAI was created so that it wouldn't be dominated by a single company. The right answer is to have multiple players. The only way to be protected is to have multiple players. The instinct to the idea of powerful AI is to clamp down and just control it, but actually multiple players are required.
A company called Rogo created a test bench and a set of evals to be a financial analyst and tested all of the frontier models. The paper published summarized that there is no single best model anymore. At the top of the leaderboard, Opus 4-7, GPT-5-5, Sonnet 4-6 appear almost indistinguishable, separated by less than three tenths of a percentage point overall. Read superficially, the results suggest convergence—three frontier systems reaching roughly the same level of capability.
Trillions of dollars are going into each of these companies to try to create the next super brain. But increasingly, existing set of evals and existing capabilities, when applied on these models, roughly produce the same thing, which theoretically says that these things are getting commoditized way too quickly. The question becomes what's the ROI on all this incremental spend.
Some of the smarter people in the open source community have suggested needing more open source connectors. MCP is actually run by the Linux Foundation. The more surface area where a model might interact with other software, the more of those connectors that can be open sourced and commoditized, it would lower barriers. This is what Google did with Kubernetes to try and commoditize where workflows live off of AWS and to make it easy to migrate. The more systems that can be created to make that type of exchange super easy so that models can be plugged and played, the better. Things like context and how context comes in and data need to be considered, which is what Glean and Databricks do. If more of these connectors can be created and thrown into the open source world, models become swappable.
There is an issue with not having a good harness for open source. The way Claude is a harness for it is part of the issue. There are people making open source versions of this or building companies around harnessing and building the integrations. Open source is always the last to build the fit and finish around the product. They focus on the core of the product. A company called Abacus was invested in. They came up with their own hardware stack and their own platform. They are sold out of these boxes that they're building for insurance, healthcare, and everybody wants to run AI inside the organization and then start building their own models. The company was incubated in an incubator. Organizations cannot get enough of this product. Organizations are getting savvy and doing with 8090 as well, building headless products so they don't get locked into any one provider.
When going into the Fortune 1000, there is never competition with OpenAI or Anthropic—they'll have a preference sometimes for what they want to see under the hood. A control plane can basically hot swap between one or the other. Seeds have been laid for open source and open weights. The reason is because organizations don't want to be tied into one of these critical frontier labs. They want to be able to ride the wave of innovation. They're afraid of two things: that one technology leapfrogs the other too quickly for them to participate and they pick the wrong one, and terms of service—being at the sake of a frontier lab in the political philosophy that they may be in the crosshairs of accidentally. An example given is a hospital system in Canada that supports the euthanasia laws in Canada, but a frontier model in America says no and shuts them off.
A lot of the folks in the Fortune 1000 and increasingly the global 1000 want abstraction above it. They want to sit at a level with a control plane. They want to have the flexibility because they don't know how it's going to shake out. They see all the money being invested at the model layers but they see the model quality asymptote. From a risk perspective, they are concerned. Regulated industries are particularly sensitive to these kinds of issues. Following what finance, healthcare, and those kinds of folks are doing will reveal what's happening. They're saying this has to be on prem. They're very concerned about a data leak. They're very concerned about HIPAA compliance. They're very concerned about training a model—what if somebody does a query or writes a prompt and it pulls some information from that Canadian healthcare system and somebody gets a result.
Anthropic and OpenAI are in all of these Fortune 1000s at the developer layer because most of the developers have their own credit cards and are allowed to sign up for them. Eventually they wrap them in an enterprise license. It's a typical PLG led market motion like seen in Slack. The interesting thing is the unwind that happens when there are huge licenses and huge buckets of spend—you can't really get the data out of them. The CEOs then wake up and are told by the CFO where they are. Uber was one example.
From Vivek Garipalli, the founder of Clover, overheard from a Fortune 20 company CEO who asked for a billion in AI generated OpEx savings at the beginning of the year. Six months in, the team has spent $200 million on tokens with minimal results. Now the CEO is saying they're in this weird motion where the CEO is pulling the budget back and now having to cut the licenses. Microsoft announced that they're killing the cloud licenses. It's a super dynamic market and the terminal solution is not known.
Claude is really good at product. Claude for Excel is better than Copilot by a lot. Anyone that's going to run against them is running against a worthy foe. Claude is exceptional and used every day. One user hit their token limit on their pro plan and had to put it on their credit card and spend another couple thousand bucks because it's so good.
To wrap up, in terms of the regulatory capture agenda being seen in Washington, where it's all leading to is an effort to ban open source models or open weight models. There are a lot of breadcrumbs leading here. People who want this are being a little bit circumspect—they don't feel like they're quite there in terms of being able to justify it.
Open source models are frequently criticized in AI safety discussions for lacking guardrails that can be removed, making them inherently dangerous. This rhetoric appears repeatedly in Anthropic's blog posts across multiple threat categories including cyber and bio threats. The language frames open models as problematic specifically because guardrails can be taken off. This appears designed to establish predicate facts in the public record to justify future regulatory action, potentially including direct bans on open source and open weight models.
Banning open source models would primarily impact cloud service providers who would stop hosting them to comply with the law, making open models much harder to use within the United States. The rest of the world would continue benefiting from open models' advantages in cost, customization, and control. An open weight model is fundamentally a file containing numbers that can run on a laptop, making enforcement challenging.
The economic and capital moat around training large models is disappearing through two mechanisms: domain-specific architectures at the silicon layer, and rebuilding of core training components. Elon Musk announced that xAI has rewritten the entire training complex in C, achieving an order of magnitude efficiency increase on 220,000 GPUs. At this scale, such innovations could reduce model training costs from $10 billion runs to $10 million runs.
A 1% efficiency improvement equals 2,000 GPUs or hundreds of millions of dollars in compute cost. Efficiency gains of 10-20% per quarter or six months would be substantial. The training speed improvement versus JAX is now an order of magnitude. This efficiency improvement will likely spawn five or six open source training stacks rebuilt closest to bare metal.
If the US successfully bans open source and open weight models, the rest of the world would likely shift to Chinese models. The EU is the most aggressive jurisdiction for AI regulation, with multiple proposed regulations targeting open source specifically because no single entity controls it. Requiring open source contributors to vet models with EU regulators would be impractical given the distributed nature of open source development.
Open source serves as a backstop for participation in the modern economy. Without it, users would essentially need to live off the grid to avoid closed proprietary systems.
Every major tech category has historically resulted in monopoly or duopoly outcomes. Anthropic appears to be pulling away from OpenAI based on growth rates, with recent Information article data showing Anthropic ahead. A company growing at 10x year-over-year versus a competitor at 3x will achieve 90% market share within two years through compounding. There are reasons to doubt Anthropic can sustain 10x growth, including competitive responses and physical compute constraints, but market leadership provides significant advantages.
Polymarket reported an AI consultant's client accidentally spent half a billion dollars in a single month after failing to set employee limits on token usage. The spend rate was $16.6 million per day or nearly $700,000 per hour. This highlights how organizations are rapidly increasing token spend beyond initial enterprise expectations.
Token efficiency will be a major theme over the next year as spend has outpaced what enterprise customers anticipated. Pricing models like $20 or $200 monthly subscriptions have created perceptions that tokens are essentially free, leading to wasteful usage patterns. Organizations have seen multiple teams independently build duplicate interfaces and tools, resulting in hundreds or potentially tens of thousands of dollars in unnecessary spend. The drive for efficiency may temper growth rates without fundamentally changing market dynamics.
The narrative around AI's employment impact has shifted dramatically. Cloudflare and Meta explicitly blamed AI for layoffs, with Meta deploying employee monitoring software to improve AI training data. Goldman Sachs CEO David Solomon wrote a New York Times op-ed arguing the AI job apocalypse is overblown, stating AI will automate 25% of work hours rather than eliminate 25% of jobs, with workers shifting to higher-level tasks.
Solomon noted that just because a job can be replaced doesn't mean it will be, citing historical examples like bank tellers increasing after ATMs and live entertainment growing after television. The US labor market creates and destroys 25-30% of jobs annually, with 25 million jobs in gross churn. New categories like agentic AI management are already creating hiring demand.
Both Dario Amodei and Sam Altman have moderated their earlier predictions of massive AI-driven job losses. Dario's revised position aligns with previous arguments that AI might automate 90% of someone's tasks while the remaining 10% expands into new tasks and responsibilities. This mirrors arguments from David Sacks and Jensen Huang that automating tasks doesn't necessarily eliminate the underlying job purpose.
Yale Budget Lab research found no discernible disruption in the labor market over the last three years attributable to AI. Software engineer job postings are up 15% year-over-year, reaching a new three-year high despite coding being AI's primary enterprise use case. The actual data does not support claims of massive job loss from AI.
Many companies over-hired during the past five to ten years with bloated OpEx budgets. CEOs lacked visibility into headcount efficiency. Current layoffs represent cleanup of poor management decisions rather than AI-driven displacement. Companies are using AI as convenient cover to reduce headcount that resulted from vanity metrics and talent hoarding strategies, particularly the approach of hiring ahead of need and keeping talent off the market to prevent competitors from accessing it.
Innovation has historically increased human prosperity, as evidenced by data from 1891 to present. Individuals can protect themselves by becoming the most AI-enabled version of themselves possible. Resistance to using AI tools puts workers at risk, analogous to refusing to adopt email, spreadsheets, or computers.
CEOs will eliminate positions because public markets reward higher earnings achieved through doing more with less. Amazon has stated plans to eliminate 600,000 future positions and reduce current headcount as AI deployment enables doing more with less. Andy Jassy indicated this will be a recurring theme. Self-driving vehicles through Waymo and humanoid robots like Optimus will eliminate driving and package sorting jobs respectively. The surface area of problems in the world is not decreasing, but the number of humans needed to address opportunities will decrease.
A Cambrian explosion in startup formation will absorb displaced talent. Workers who embrace AI tools will create small companies of five to ten people and potentially earn more than previous salaries while having greater control. This transition will be extremely painful for affected workers including cab drivers, truck drivers, and package sorters over the next decade.
The tools are working and there are jobs that are no longer needed. Product managers, designers, and other roles have been consolidated into single positions where one person ships a product. Fortune 1000 companies are seeing this consolidation happen in real time. Startups represent the front line where all the jobs are being created, not the slowest adopters.
Startups cannot enter regulated markets and displace incumbents like JP Morgan. The same logic applies to Boeing and pharmaceutical companies. Someone showing up at the FDA claiming "it's just me and my model" will not work because society demands safety, predictability, governance, and auditability. There is a distinct difference between drugs and software/services. Truck driving and cab/taxi driving are among the most regulated industries, yet those jobs are being eliminated.
David Sacks compared Jake's position to a Trotskyite who claimed failed predictions proved how far-sighted Trotsky was. None of the predictions about job loss have come true. Zero. The data shows otherwise. Jake countered that 8,000 people losing jobs is not an anecdote, citing specific cases where CEOs attributed layoffs to AI.
Sacks argued the Meta job losses were not directly attributable to AI despite Zuckerberg's comments. The company clarified it was about balancing additional capex spending. Even if 100% of specific job losses were attributed to AI, they must be netted against all jobs being created and new companies being formed. Cherry-picking one statistic without presenting a balanced view distorts the picture.
Jack Dorsey announced a 50% elimination at Block due to AI. Financial analysts responded within 24 hours calling it "AI washing" and noting Block had overstaffed during COVID and was running inefficiently compared to peers. Matthew Prince, Zuckerberg, and Andy Jassy were also cited as making AI-related statements about workforce changes.
Current unemployment is at 4.3%, with economists considering 5% to be full employment. This is at or near record lows of our lifetime despite being over three years into the AI wave. Coding is the single job category most impacted by AI, with AI now writing most of the code and near-complete automation of code writing.
Software developers are not being laid off on net. Job postings for software developers are at a three-year high, growing 15% year-over-year. GitHub saw 1 billion code commits last year and 1.1 billion in the past month alone, representing a 14x year-over-year increase in code generation.
When code volume explodes 10x or 100x, complexity also rises. Humans are still needed to manage and look under the hood of all that code. More engineers are required to manage the increased volume, even if not 10 times more. There has been an explosion of code use across the economy by businesses, applications, and use cases that never used software before.
Fund managers are now hiring software developers instead of data analysts because they are deploying code for the first time. Cloud proficiency is currently the most marketable skill in the economy. This represents the outset of a boom caused by bespoke software proliferating throughout the economy, being used by firms that never considered themselves tech companies.
The data center buildout and new energy/power generation are creating hundreds of thousands of new construction jobs. This represents a blue-collar boom happening in parallel with white-collar changes.
Jake clarified his position: there is job displacement happening related to AI, but net the economy will grow. Bill Gurley noted that maybe at some point in the future the predictions will prove correct, comparing it to the Trotskyite position that "communism has never been tried."
Of 20 million people in the United States driving cabs and trucks, the question was raised about how many will lose jobs to self-driving in the next decade or two. Bill Gurley responded that 100% automation is economically unworkable. Non-ownership cars will increase significantly, with humans used for approximately 50% of rides instead of 100%. The number of jobs might stay the same or grow.
Regulation previously limited the taxi market, and getting around that regulation led to job creation. There is not high confidence in government skills retraining programs. The recommendation is to use the new tools, understand what is capable in your field, and if a job is going away, start thinking about where opportunities exist. Skilled trades have massive shortages in plumbing, electricians, and HVAC.
Amazon has a self-driving division called Zoox and is the largest user of robotics in the world. They are pursuing self-driving among the top three or four companies. The debate centered on whether specific examples represent AI-driven change or pre-existing trends.
Jake believes there is a chance of seeing job loss increase in the short to mid-term, then displaced people will have to learn or leave the workforce, similar to the PC revolution where some lawyers retired rather than adopt tools while others became "PC first." Kirkland & Ellis is spending half a billion dollars to roll their own frontier model.
Competition prevents companies from simply doing more with less and achieving 70% operating margins. Someone else will come along, do more for less, and lower prices. The expected outcome is a productivity boom with cheaper goods and services, making the basket of goods humans can buy cheaper and cheaper. This has been true in many categories but is offset in healthcare, education, and housing.
Securities litigation partner Donnie King at Ackerman has warned clients that shareholder lawsuits could emerge against companies engaging in AI washing, viewing it as a form of puffery that could constitute securities fraud when companies attribute non-performance or operational issues to AI when real business problems exist.
The Wix CEO's layoff note focused on operational details rather than AI. The website builder business faces challenges because websites can be built with Claude. There has been a narrative shift with Sam Altman and Dario Amodei backing off job apocalypse predictions. Goldman Sachs' CEO also received pushback. Job postings are exploding. The position that the job apocalypse is massively overblown is gaining acceptance.
The tech industry lacks empathy when discussing real people losing real jobs who may not make the transition. Being empathetic does not mean scaring people about job loss. The enabling approach is to encourage people who have been laid off to start companies and embrace the tools. Learning the tools leads to 10 job offers and the ability to start one's own company.
Mike Rowe Works has funded $16 million in scholarships for 2,600 people to become plumbers, welders, or electricians. A new grant program helps people tilt their careers in different directions. The skilled trades deficit is significant and represents real opportunity.
Companies IPOing while claiming jobs are going away creates a bad look. The alternative view is that calling out AI washing forced companies to tell the truth. The episode concluded with a shout-out to Tulsi Gabbard and her husband Abraham who is fighting cancer.
Keep All-In Podcast in your library
Save the videos and channels worth coming back to, and find them again in one place.
