This episode of Odd Lots features a live panel discussion hosted by Tracy Alloway and Joe Weisenthal with three returning guests: James Van Gielen, founder of Citrini Research and author of the viral AI jobs doom scenario; Jasmine Sun, a Substack writer covering AI and Silicon Valley culture; and Sam Rowe, author of the stock market newsletter TKer.

The conversation spans the challenge of communicating AI's implications to a general public, the existential question of whether AI will replace media and investment research, and what human writers and analysts can still offer that models cannot. The panel also covers the US-China AI race, the alignment and safety debate, the robotics boom quietly underway in factories, and how investors should think about AI-driven market volatility and potential job displacement.

Communicating AI to a Public Left Behind by Tech Media

Jasmine Sun argues that the vast majority of AI media coming out of Silicon Valley is 'by AI people for other AI people,' leaving ordinary people's questions about parenting, politics, and affordability largely unaddressed.

Industry leaders make alarming statements — about job loss and existential risk — without considering that a general audience is listening: 'They're saying all these crazy things without for a second thinking that any normal person might hear them.' - Jasmine

Unlike crypto, which only a fraction of the public ever deeply engaged with, AI is already impacting people's social feeds, workplaces, and children's education, making broad public communication more urgent.

AI Job Loss: Real Concern or Tech CEO Hype?

James Van Gielen's viral post was a scenario, not a forecast, but still generated two credible death threats — illustrating the raw public anxiety around AI-driven unemployment.

James believes AI will mirror historical technological leaps and ultimately be positive, but the critical unknown is pace: 'What happens if the models get so good and people adopt them over 5 years instead of 100?'

Sam Rowe highlights the economic paradox of mass AI-driven job loss: 'The economy stops working when no one has jobs and no one has income. It's great that you have this huge profit margin, but you have no revenue because everyone's unemployed.' - Sam

The US has never successfully executed a large-scale reskilling program — during deindustrialization, steelworkers did not learn to code — making political backlash a real risk even with relatively small net job losses.

AI Alignment and Existential Risk: P-Doom in Context

Jasmine Sun declines to assign a probability to human extinction from AI but argues safety and alignment are not in tension with utility: 'The alignment of the model is very correlated with whether it is economically useful as well.' - Jasmine

Historically, it is often the researchers who make the biggest technical advances in foundational models who end up most concerned about misalignment — because they understand how controllability underpins usefulness.

Most senior Silicon Valley figures Jasmine has spoken to are genuinely worried about societal disruption, not just talking their book — their concern centers on the speed of transition, not the direction.

China vs. the US: Three Eras of AI and a Hardware Reckoning

Jasmine frames US AI development in three eras: the academic era (kicking off with ImageNet/AlphaGo), the commercial era (ChatGPT), and the current geopolitical era (Pentagon and national security focus).

China remains in the academic era — collaborative, pro-open source, and largely unconcerned with safety alignment. Chinese labs have abdicated philosophical questions to the party, which actively shapes AI's role in society.

Chinese researchers are compute-constrained due to chip controls: 'One OpenAI researcher is allocated more compute than an entire Chinese lab oftentimes.' - Jasmine

James predicts the next Chinese 'DeepSeek moment' will be about hardware, not a model — specifically targeting the DRAM and memory bottleneck. CXMT, China's answer to Micron and SK Hynix, is IPO-ing this year and will attract significant capital.

Flash memory bandwidth is 100x cheaper than DRAM, making it a viable path to widening the bottleneck.

James warns that 'bottlenecks are made to be widened' and that investors piling into the bottleneck trade may be underestimating this dynamic.

James argues the US should continue selling chips to China, invoking a Sun Tzu principle: 'You build your enemy a golden bridge upon which to retreat' — cutting China off entirely could accelerate their push for independent AI infrastructure dominance.

The Quiet Robotics Boom Already Reshaping Factories

Amazon will employ twice as many robots as humans this year — a milestone that signals factory and warehouse automation is already at inflection, well before humanoid robots reach consumer homes.

Jasmine visited the Unitree office in Hangzhou and witnessed humanoid robots and quadrupeds actively deployed in factories for inspection, surveillance, and a 24/7 pharmacy operation — fully operational, not prototype.

Factory automation earnings data from companies like FANUC show a 'huge inflection that's not commensurate with other segments selling into factories,' suggesting the robotics boom is underreported relative to its scale.

What Human Writers and Analysts Can Still Offer AI Cannot

Jasmine Sun's framework: journalism is the act of converting private or tacit knowledge into public knowledge — 'that particular task is extremely robust in the era of AI' because the world is dynamic and models are trained on static scraped data.

James Van Gielen argues that physical presence — sending a human analyst to the Strait of Hormuz or a reporter to a party in San Francisco — generates data that simply isn't in any training set.

Sam Rowe identifies the hardest-to-replicate skill as generating angles and ideas no one is asking for: 'The toughest part about my job is being able to come up with those ideas and those angles that no one's asking for.' - Sam

Sam's sister built a Claude-powered 'Sam robot' trained on his published work that mimics his voice and word choices — an experience he describes as 'kind of offended, but yeah, it's a problem.'

James notes that investment research as currently practiced will not survive: models have been trained on vast amounts of existing research, so differentiation must come from either being consistently right or doing something fundamentally different.

Investor Psychology: Why Ignoring Bad News Is a Mistake

Sam Rowe pushes back on the conventional advisor advice to ignore the news and forget your 401k password: 'I think that's all incredibly silly. I think you really do have to think about how bad things are at a given time.' - Sam

His reasoning is memory-building: experiencing downturns consciously makes investors more robust in future crises, because they have lived reference points for how markets recovered from past disasters.

On market peaks, Sam acknowledges the data shows 12-month returns after all-time highs tend to be higher than at lows — but maintains that healthy nervousness and awareness of drawdown risk is essential for any investor.

Resources Mentioned

The Notebook of Doom (Books 1-3)

Tracy Alloway and Joe Weisenthal with three returning guests: James Van Gielen, founder of Citrini Research and author of the viral AI jobs doom scenario; Jasmine Sun, a Substack writer covering AI a

Research Notebook template for literature reviews, systematic reviews, and structured summaries [A4 - Tile] (Academic writing)

perience he describes as 'kind of offended, but yeah, it's a problem.' James notes that investment research as currently practiced will not survive: models have been trained on vast amounts of existi

The Notebook of Doom (Books 1-3)

ave been on the show before. Yeah. We had James Van Gielen. He is of course the founder of Citrini Research and the author of the viral AI jobs doom scenario, as well as Jasmine Sun. She writes a Sub

Data Structures and Algorithm Analysis in C++

timing of things I do write about or the way I frame it that feels a little bit more human than the research or the analysis that, you know, the chief investment strategist or whatever is sending out

If It Doesn't Challenge You It Won't Change You Journal

a gathering? Yeah, I mean, you just have to look at this from— I'm pretty confident that investment research won't exist in the same way that it does now. The internet kind of made it democratized, so

that these models have been trained on

ilizing it to kind of do the same thing that the banks are doing, well, there is so much investment research that these models have been trained on. And as they get better, they will pretty much be ab

AI Study Skills Workbook for Students Responsible AI • Digital Literacy • Academic Integrity • Student & Teacher Guide with Posters & Activities

et and we have these walled-off internets. One interesting thing is the way that I think about USAI research is it sort of had three eras of American AI. You had the academic era, maybe you could cal