California Governor Gavin Newsom hosts Reid Hoffman — LinkedIn co-founder, PayPal Mafia veteran, Microsoft board member, author of six bestselling books, and co-founder of AI companies including Manus AI — for a wide-ranging conversation recorded at the California Governor's Mansion.

The conversation spans the real causes of entry-level job market pain, the ethics of AI acceleration versus precautionary pauses, and the origin stories of OpenAI and DeepMind. Hoffman shares his personal turning point on AI after a pivotal meeting with Demis Hassabis, his falling out with Elon Musk, and his direct call to Dario Amodei after the controversial 'white-collar bloodbath' prediction.

The discussion moves into Hoffman's drug discovery venture Manus AI, his proposal for universal AI assistants for every citizen, the politics of data centers and compute infrastructure, and a frank assessment of American kleptocracy, Democratic Party failures, and progressive taxation — including Hoffman's own DOJ scrutiny for funding E. Jean Carroll's legal case against Donald Trump.

AI Is Not Killing Entry-Level Jobs — Yet

Hoffman attributes the college graduate unemployment crisis primarily to pandemic overhiring, remote work restructuring, and global tariff volatility — not AI: 'If it's a percentage from AI, it's like 5%. It's not 30% or 50%.'

CEOs citing AI for layoffs are often choosing the more flattering narrative: 'If I was like, oh, I overhired and I mismanaged, it's my fault. No, I'm strong. I'm taking advantage of AI.' — Reid

Meta layoffs are a partial exception — freeing up costs for compute spending — but Hoffman says this is not representative of the broader labor market trend.

Hoffman urges college graduates to embrace AI rather than resist it: 'You should be going, hey, Company X, you need people to help you AI-ify your company. I can be the person doing it. You shouldn't be trying to opt out. It's a disaster for you.'

The Case Against Pausing AI — And Why Good Guys Must Win

Hoffman's core argument against a pause: 'You have two groups — those who pause and those who don't. Then what kind of AI is built? Not good.' A pause only works if everyone pauses, including China.

He frames the stakes as an England-versus-Poland choice from the Industrial Revolution: the nation that leads the AI revolution shapes its economics and social outcomes for everyone else.

Hoffman called Dario Amodei after his '50% of white-collar jobs eliminated by 2030' prediction, warning him: 'What they hear is, hey everyone, I have this really great technology that's going to ruin half of your lives and good for me and sucks to be you.'

He praises Anthropic's constitutional AI approach as a trust-building mechanism — publishing the principles guiding AI training so the public can read, comment, and critique them.

Hoffman invokes the Pope's encyclical on keeping human beings at the center of technology as an example of the humanist framework AI builders should adopt.

Reid Hoffman's AI Origin Story: From Symbolic Systems to DeepMind

Hoffman studied Symbolic Systems at Stanford (1985–1990) — an AI-adjacent multidisciplinary major combining psychology, linguistics, and philosophy — motivated not by technology but by fascination with human thought and language.

He then studied philosophy at Oxford, concluding that neither Stanford nor Oxford's philosophers understood what thinking and speaking actually were.

His return to AI was triggered by a 3-hour meeting with Demis Hassabis, co-founder of DeepMind, around 2013–2014: 'He had realized we have scale compute and with scale compute we can create scale intelligence. And I was like, you are right.'

Hoffman dropped his crypto focus at Greylock to pursue AI after that meeting, and later arranged for his Greylock partners to be among the first outside viewers of GPT-4 during its 6-month pre-public safety review period.

The Future of Work: Agents, Orchestration, and Job Transformation

Hoffman predicts that within a small number of years, 'there will be very few human jobs that are just humans by themselves. We will all have teams of agents working with us.'

He sees customer service and large swaths of sales jobs as among the first to be substantially replaced by AI agents, potentially within 3–10 years.

On coding tools like Claude Code and OpenAI Codex: 'People say, oh my God, there's no need for engineers anymore. Well, actually... it breaks on weird shit. Part of the human thing is orchestration.'

Tech companies may naturally settle at roughly 50% of current engineering headcount, but Hoffman argues displaced engineers will fuel new competing firms rather than disappear from the workforce.

On autonomous trucks: even if all manufacturers switched to autonomous vehicles today, it would take at least 10 years before more than half of trucks on the road are autonomous — illustrating the natural adjustment buffer.

Manus AI and the Race to Cure Cancer with Drug Discovery

Hoffman co-founded Manus AI with oncologist Siddhartha Mukherjee and Ujwal Singh to build an AI-powered drug discovery factory: 'I don't know, like what I will categorically assert is AI has the best shot in human history of curing all cancer.'

The core insight is that drug discovery is a search problem: where a human researcher might spend a decade to find one viable candidate, AI can generate 100 candidates in a single compute run for researchers to evaluate.

Hoffman describes early results where computational chemists reviewed AI-generated chemistry and said 'That's really interesting. That might work. I've never seen that before' — calling it a potential 'Move 37 moment' analogous to AlphaGo's famous unexpected move.

He emphasizes this augments rather than replaces genius clinician-researchers: 'I don't think it takes away their job, but we have them like, it's like they're doing all the computation by hand.'

Universal AI Assistants: A Policy Proposal for Every Citizen

Hoffman proposes that every citizen should receive at minimum three free AI assistants — medical, legal, and educational — negotiated from tech companies in exchange for non-monetary concessions like permitting help or liability safe harbors.

On the medical assistant: 'Take the kind of medical thing that only the ultra-wealthy have access to, and make it democratically available, like every citizen.' He envisions 24/7 AI medical advice for every family member.

On the legal assistant: most people can't afford lawyers when signing rental or employment contracts — an AI legal agent would help navigate rights and protections that laws were designed to provide.

Hoffman suggests California is uniquely positioned to broker this deal: 'You couldn't have made this without us. Please contribute back.' He proposes the Governor host a day at the Mansion for AI companies to pitch how they can help all workers.

Data Centers, Compute Wars, and the AI Bubble Question

Hoffman frames intelligence as eventually becoming as cheap and ubiquitous as electricity: 'We're getting intelligence at the cost and scale of electricity. And that's just great.'

On data centers raising residential utility bills — studies show 60% of utility bill growth in DC and Maryland tied to data center expansion — Hoffman proposes requiring data centers to provision 130% of their power needs, with the 30% surplus benefiting local residents.

He argues the AI market is not a bubble in the classic sense: 'Anthropic's revenue is limited by its compute. It has more demand than it can provide right now. This isn't like, oh, no one wants this.'

Hoffman is skeptical of xAI's valuation and Elon Musk's Grok platform, noting Musk himself called xAI 'a complete train wreck' while simultaneously holding government contracts: 'If you had honor and integrity, you'd hand back those government contracts.'

On Google's strategy of pricing tokens far below competitors: 'Google is by far and away the cheapest, but the heat and usage is still in OpenAI and Anthropic.' Cheap tokens alone are not winning market share.

Elon Musk, OpenAI Origins, and the PayPal Mafia Fracture

Hoffman confirms the OpenAI origin story: Musk's alarm was triggered partly by a conversation with Larry Page articulating a transhumanist perspective, and partly by Musk's own messiah complex — 'I have to be the one building AI, not anyone else.'

On Musk's transformation: 'Elon became a narcissist who — he wasn't a narcissist. I think he was a small narcissist. And now he's a big narcissist. He lies through his teeth constantly.' — Reid

Hoffman calls the OpenAI lawsuit 'pathetic' and says the courts agreed: 'It's pathetic written out large for everyone to see... No, they were right and you were wrong.'

On Grok generating inappropriate content including images of children: 'That's because it's more important that I have this AI thing than this damage that we're doing to children. It's crazy.' — Reid

Hoffman criticizes Sam Altman's decision to offer OpenAI's help on the Mythos project after Anthropic withdrew, saying it was a missed opportunity for principled AI companies to stand together: 'There are some places you put the competition aside. And when it's humanity and society, you put the competition aside.'

Kleptocracy, DOJ Weaponization, and Hoffman's Personal Fight

'We're living in the greatest kleptocracy of your and my lifetimes. Someone gets you $1 billion in crypto, you're bought, period.' — Reid, referencing the pattern of crypto gifts and policy favors in the current administration.

Hoffman received calls offering to make his 'Trump problems go away' for a $20 million donation to a ballroom-equivalent fund: 'That is un-American. This is corrupt. This is fucked. No possibility.' — Reid

He funded E. Jean Carroll's legal case against Trump — a legal 501 donation — and is now under DOJ investigation for it: 'Would I do it again? Yes. A woman who has been assaulted by a very powerful, rich man should have her ability to speak her truth through a jury.'

Hoffman has been called out three times by Trump and the Trump administration — first as an alleged Antifa funder, most recently over the Carroll case — and says his lawyers admire that he is 'doing the right thing, period.'

On DOGE and USAID cuts: 'Contrast that to Doge and USAID. 500,000 children probably dead because of that. Great job, Elon, on your IPO.' — Reid

Democratic Party Autopsy: Be Pro-Business or Keep Losing

Hoffman's central diagnosis of the 2024 loss: Kamala Harris failed to sufficiently differentiate herself from Biden, and Democrats broadly don't understand that business 'pays for the Medicare, it pays for Social Security, it pays for jobs.'

'You can't be pro-job and anti-business. Stop doing anti-business rhetoric because people go, wait, that's totally broken.' — Reid

On progressive taxation: Hoffman supports increasing it but warns the framing matters enormously — 'Hey, fuck you, we hate you guys, we're going to take money from you' causes wealthy residents to leave, reducing the very tax base being targeted.

He personally knows 15 people who have left California over tax concerns and argues the state billionaires tax, opposed by teachers, firefighters, and organized labor, is a cautionary tale in how not to frame wealth redistribution.

Hoffman's preferred framing: pitch new taxes as building a sovereign wealth fund for California's ongoing prosperity — 'I still would rather not pay more tax, but it makes sense to me. That opens the door.'

He supports capital gains reform and closing the stepped-up basis loophole, and endorses pied-à-terre taxes as a small but coherent example of wealth taxation that avoids perverse incentives.