If he truly believes this, why wait to stop thinking if AI is so capable? Why not have AI write all of your blog posts, papers, proposals, emails, conference presentations, letters of rec, etc? Read an AI script for your next keynote lecture. Every professor will be doing exactly that in the near future, as you predict.

Or maybe - maybe true critical and synthetic thinking will become even more rare and valuable than before. I actually find that when using AI in my job, I have to think more - I don’t get to do the slow work of crafting code anymore. I have to keep building my understanding of an ever-evolving codebase, on a cadence of hours. Because I have to explain what it’s doing at some point. I’m not sure that part goes away.

Or if I’m a professor writing grants or giving lectures, eventually I have to stand up and talk about it, and answer questions. So the machine can help me generate content, but I still have to internalize much of that content before I can confidently deliver it. So now that’s my rate-limiting step - the processing and digesting. I need to be doing that every day, and people will come to expect a new rate of output.

I think for some people, there will be an ability to tune out and become a manager of AI, just passing specs, requirements and feedback through. But I suspect that for many “knowledge workers”, the “knowledge” part is actually going to get more intense. That’s been my experience so far at least.

But maybe that’s just the last gasp of somebody being outcompeted by a thinking machine.

Again, I’m not saying we’re there, it just feels like the control software is the main bottleneck, and the development of that is now being accelerated by LLMs.

I agree the data trove for these is not nearly as comprehensive as “the entire internet” was for LLMs. But LLMs paved the way - people are building the datasets rapidly now. I suspect this will look a lot like LLM rollout looked over the past 5-ish years. Unusable garbage… less unusable garbage… then suddenly, a fully capable model like Opus 5.5 or the like.

And the “problem” with the way the foundational models seem to work is that they don’t just acquire individual skills one at a time like we do - they go from knowing nothing to knowing everything.

> I'm also continually surprised that people seem to think politics doesn't exist anymore. I find it exceedingly unlikely that massive amounts of people will simply accept this

This is exactly the part I’m afraid of. The political response of the displaced labor (which may or may not be violent / revolutionary), and the subsequent counter-response of capital… potentially using the very robots that displaced the jobs to begin with.

But also, upon more reflection, it occurred to me that if we’re able to create a competent humanoid robot, that could legitimately be world changing in some good and other very bad ways. The scary part is that I don’t believe there’s any major physical / energetic barrier to this. The hardware exists, the power supplies are light enough. The only part that’s lagging is the software and motion contols; the intelligence. The part that is experiencing breakthroughs at an accelerating pace.

What if a highly intelligent and physically competent robot were available for the price of a cheap car, say ~$20k? What if that robot were competent enough to cook, clean, assist the elderly, move things around, assemble and disassemble things, etc? Billions of jobs would be displaced, or hit with a very concrete maximum wage.

Many white collar jobs, while still at risk for displacement as well, benefit from barriers like IT integration and data privacy hurdles, and the fact that corporate higher-ups simply prefer to communicate with humans.

But most blue collar / manual labor jobs have almost the opposite condition: humans introduce risk, and generally don’t enhance anybody’s “experience” - see people’s somewhat odd acceptance of higher fares for a Waymo vs an Uber. [1]

I don’t think we’re there yet, but the fact that we are rapidly accelerating our trajectory toward that world should be at the least, unsettling.

[1] https://techcrunch.com/2025/06/12/waymo-rides-cost-more-than...

I’m afraid this is where most cases of general “arbitration” are headed. Whether it’s a court case, you negotiating with your insurance / warranty / other hostile corporate party, or even job applications to an extent, it seems that there is an advantage to be gained by both parties using AI to overwhelm their “opposition” with generative babble. It’s almost like a proof-of-work hurdle that will need to be surmounted in order to escalate to a human.

We largely already missed our chance to stop the toxicity of social media - let’s not mess it up again with chat bots.

> In my meetings with Daniel I found that the discrepancies between our CVs were due to Daniel calculating CV based on the median value of each precision run, while I was calculating CV of the entire data set for each level. When I asked him why we do this, he said that it was a way to average out the noise. I was under the impression that the coefficient of variation was meant to be, at least in part, a measure of how much noise exists in the data. By averaging out this noise before CV is calculated, the CV as a metric of assay performance becomes less meaningful.

“Becomes less meaningful”… this is the most polite way I can imagine telling your CEO, “hey, our results are all BS”. His message is so thorough and measured, and they just tear into him. He’s the only professional in the room.

People need to stop the absurdity of imagining AI as some out of control independent entity. Every job is kicked off by someone’s prompt. Every job runs on models and compute owned by people. Assign accountability where it’s due: GPT didn’t hack huggingface - OpenAI did. They wrote the prompt, built the sandbox and ran the compute. When you write a program that hacks another company, you are responsible. This doesn’t magically change with LLMs. Also, if their model is so smart, why didn’t they use it to design the sandbox? Or was it incapable? Or were the humans too lazy?

If you build the world’s fastest train, start it up with no driver and don’t finish the tracks, when it crashes, it’s just your fault. Not the train’s. So OpenAI saying “we’re worried AI will wipe out humanity” is basically equivalent to them saying “we’re worried we will wipe out humanity”. Like, seriously? Don’t worry, we’ll take care of it if you even come close.

I liked The Windup Bird Chronicle slightly more, and it’s an easier read than 1Q84 to get into him. He also does short stories that are phenomenal.

The killer was usually the downstream processing required to clean out off-flavors and colors, which was required to get even a passable product (these are just protein powder fillers at this point, not whole cuts), but ballooned CapEx and cut yield such that cost went back up to like, A5 Wagyu levels.

I’ve sat in the awkward tasting sessions the author describes, and they’re all the same. The attendees muster up a “that’s pretty good” - but we all know nobody would ever choose to buy this unless they were forced to.

Now that I’ve been out of it for a while, it’s pretty obvious to me what the problems are. Cost, obviously. Taste and texture simply aren’t there. And “the knowing” is real. I never got over imagining the whole factory process involved in the production. I’m sure vegans would say the same about animal farming, though. So my conclusion was, the only way this technology becomes relevant is if people are forced to use it. And that’s either because you’re cruelty-free by choice (small fraction of the population), or because the cost of meat suddenly skyrockets, which today, is not looking that likely. Keep in mind as well that meat costs would likely be driven by feed costs in such a scenario, which also impacts cultivated meat. So it’s ultimately a process efficiency play, and you’re competing against eons of evolution.

But stating it plainly like this would make the contradiction too obvious.

Science article in question: https://www.science.org/doi/abs/10.1126/science.aec6129

Also, the most interesting result here is that the CNN-based feature encoder significantly outperformed a vision transformer encoder backbone…

If OpenAI or Anthropic go away tomorrow, people find a new model and move on. If Google goes away tomorrow, people’s lives would be severely disrupted, and in the case of Gmail / Drive / auth access, even temporarily collapse.

> the classes of firms that can accept the use of fully autonomous LLMs are few, by my count just three:

1. those who can accept failure cheaply: firms that would otherwise hire interns, firms involved in rapid prototyping work, etc.

2. those who need done a small set of narrowly defined tasks with existing clear guardrails: repetitive physical labor in a controlled environment, call center and customer service chat work, etc.

3. those that can accept or already do by nature the costs of rigorous specification and validation: chip design, drug discovery, and other domains where failure on deployment is an existential concern.

the first two classes are price sensitive and arguably don't need the jump in reasoning quality you see going from cheap to frontier models. most of these firms will be best served by open models running on cheap hardware, perhaps even locally at the site of use. for the first and third classes, the type of fuzzy combinatorial search that has produced headline results in mathematics and security research seems more sensitive to agentic swarm width than reasoning capacity

…

This is just so on point. And for the third class (which I would extend to things like materials research as well), specification and validation are already by FAR the larger costs, so automating search and simulation is really not a massive game changer for the broader business.

I think they’ve made a shrewd move in focusing on search integration and everyday users (Gemini app) vs software power users. They have their corner and nobody is really competing with them, plus it feeds directly into their existing revenue stream.