I’ve been reading a lot about historical cases of automation and how it impacted the labor market. At this point, I don’t think AI will make a huge dent in the knowledge-worker job sector. Not even among SWEs.
Polanyi’s paradox comes into play. It’s basically “we know more than we can tell.” The argument is that automation can only do the parts of a job someone managed to write down as rules or examples. The parts that resist articulation stay human. Implicit stuff like knowing what “done” means for this customer, sensing a requirement is wrong before anyone can say why, noticing that correct-looking validated output is subtly off. LLMs are the strongest test of this yet, because they were trained on everything humans did write down. We’re seeing the residue of stuff nobody ever wrote down, because nobody could.
Michael Kremers O-ring theory basically describes a production process where every step must succeed or the whole product fails.
In an O-ring process, making 90% of the steps actually *raises* the value of whoever guards the last 10%, because a failure there now costs the entire product. And we know AI doesn’t get us to 100%.
Then we have Bainbridge’s “Ironies of Automation” from his research in aviation and process control. His premise is: the better the automation, the harder the human’s remaining job gets, because what’s left is monitoring plus intervening in exactly the cases the machine couldn’t handle, with less routine practice to stay sharp. Fewer working hours but the hours you do work are more intense.
Look at Machine translation. Neural MT got pretty goodish around 2015ish. But, translators didn’t vanish. Instead the industry restructured around post-editing, where the machine drafts and a human fixes. This ended up skyrocketing translated volume and increasing demand.
ATMs are a good example too. Teller employment rose for about 30 years after ATMs, because branches got cheaper to run and banks opened more of them. Tellers shifted to relationship work.
And the classic example: spreadsheets. VisiCalc shipped in 1979. Bookkeeping jobs collapsed, but analyst and accountant jobs increased by a much greater magnitude because humans were needed to figure out “what question does this model answer.”
Excellent post. I think the question of “where would LLMs add value beyond coding” is key. Software seems like an exception because output validation is easy. In other fields, the value of LLMs may be less visible in ROI.
Thoughts from my Shanghai field trip
Sentiment among Greater China investors was far worse than I expected. Unlike the US market, A shares had not seen a single rebound since July. Still holding Chinese and Hong Kong AI stocks that had crashed and Korean memory stocks that had collapsed, investors kept pressing me on when memory would bounce back.
Many investors said the AI trade was over, and a few said they doubted whether even 2027 capex, not just 2028, would go ahead as planned.
Overall, the bearish sentiment was so strong that even I, a bull before coming to China, started to waver.
It got to the point where, faced with the bear case that memory prices could start falling in the third quarter of next year, I took it seriously and couldn't refute it.
As the conversations went on, I gradually began to understand where their pessimism was coming from. In the end, they were all asking the same question.
So what comes after AI coding? What will be the next engine of ARR for Anthropic and OpenAI?
Several candidates came up. AI4S, AI chip design, and so on... But none of them clearly convinced investors.
I couldn't offer a clear answer either. Sure, there are rumors that Anthropic has made major breakthroughs in both AI4S and AI chip design, but what we actually want to know is when any of that turns into money.
With AI4S, there could be beneficiaries right away, such as companies that sell lab monkeys or diagnostic kits. In fact, there already are. But while we can trade those names in the short term, what we ultimately want to know is whether Anthropic can make money by developing drugs. That is still several years away.
As for AI chip design, the TAM itself is hard to calculate, and even if you do calculate it, it is hard to monetize. That makes it difficult for the market to believe in.
Honestly, this feels like the hardest market in history. If a company isn't making money, selling makes sense. But right now the market sells even when companies are making money, even when they beat, and even when they say demand will stay solid all the way through 2028 and 2029.
So my recommendation to investors was to buy memory, and within memory, DRAM. Not because I have conviction in the DRAM cycle, but because I see it as a relatively defensive choice within the AI trade.
Memory stocks are now trading close to 1x P/B on 2027 BPS. Historically, memory stocks have staged technical rebounds when they fell below 1x P/B.
And even if the broader market falls further from here, the three DRAM makers will keep buying back their own shares. I thought that would guarantee a certain level of buying and concluded the stocks would have downside support.
Of course, I don't know whether memory stocks can break above their previous highs. That depends on whether sentiment across the broader market recovers. But I think there is a fairly good chance that memory outperforms the rest of the AI trade.
roman retweeted
Here’s the thing about spotting the horrendous, soulless AI-generated writing infecting every corner of the internet now: it’s not just about the words — it’s about the patterns.
And once you start noticing them, you can’t unsee them.
The unnecessary em dashes. The polished transitions. The examples given in groups of three.
The weird metaphors that kind of make sense but kind of don’t — like a parachute opening inside a submarine.
The load-bearing use of the term “load-bearing.”
The random one sentence paragraphs.
The way a bunch of bland gibberish is quietly reframed as a profound insight.
The on-going and un-necessary hyphenation of every-day words that really don’t need it.
The cliches.
Read that again.
Cliches.
Louder for people in the back.
Tons of cliches.
That’s not a typo.
Sit with that for a moment.
And the biggest red flag? Asking yourself a question and answering it.
Because honestly? No sentient human writes like this.
And truthfully? If you can’t identify these red flags, you might not actually be sentient yourself.
Because really? If this formatting doesn’t immediately strike you as awkward and artificial, it means you don’t read books.
Ever.
Seriously.
But here’s where it gets interesting:
It doesn’t.
The point has been made but I’ll keep going. That’s what AI writing is all about.
Let’s delve into real problem. It isn’t the technology. It’s the people using it because they’re too retarded to generate an original thought of their own.
That’s not thoughtfulness. It’s intellectual laziness wearing digital makeup.
It doesn’t make the internet more fun. It makes me want to throw my phone in the river and forsake all of humanity and live as a hermit in the woods for the rest of my life.
Seriously this shit fills me with despair. I’m not kidding.
And the distinction matters.
Especially when you’re saying the distinction matters without actually drawing a distinction.
But here’s what nobody tells you.
When you ask a machine to manufacture your personality, you’re not just automating your writing. You’re quietly announcing that you’re a boring, insipid dork with nothing interesting to say.
Let that sink in.
The uncomfortable truth?
AI didn’t turn these people into tedious midwits.
It quietly revealed what they already were.
And perhaps — just perhaps — that’s quietly the most important takeaway of all.
But what do you think? Let's unpack this together in the comments. 👇
I think I'm getting routed to Fable 5.5... the results I'm getting from the Fable 5.1 are unbelievable, much better than the old Fable 5.1. Same prompts. Maybe a release this week?
roman retweeted
Design engineering is a trap
It's too easy to delegate the 'thinking' part of the job to the agent and ship whatever it suggests
For a long time I've been 'designing in code' but now I'm actually designing in tools like Paper/Figma first and then going to code
This is the way
roman retweeted
The AI naming curse strikes again.
“AI Safety” firm made AI unsafe.
“Effective Altruists” are both ineffective and enabling criminal activity.
“Irregular” is regularly incompetent.
BREAKING: A single Israeli Effective Altruism firm is behind OpenAI, Anthropic, and Meta cyberattacks
🧵 with help from @lumpenspace
roman retweeted
I have conducted an audit of Anthropic's finances.
What I have found is so shocking that I am calling for a Congressional investigation.
Anthropic is not just seeking regulatory capture.
It has built a regulatory capture machine that cannot be turned off.
Structural financial incentives make it impossible for Anthropic -- I call it the Anthropic Network -- to turn off its own AI doom cycle.
It starts with METR.
Dario Amodei proposes "third-party evaluators" to assess the risk of Anthropic's models.
He proposes METR for this purpose.
But METR is financially dependent on the Anthropic's success -- specifically, on the explosive growth of more than $7 billion dollars in Anthropic stock.
Dustin Moskovitz invested this stock into Good Ventures Foundation, where it represents the majority of that organization's portfolio.
And GVF is the overwhelming funder of the entire Anthropic Network ecosystem.
This stock was worth $500 million early last year.
It is worth more than $7.7 billion just ~16 months later.
METR -- and all of those building a career its parent organizations -- cannot afford to disrupt that growth.
Because if Anthropic goes under, many of the organizations that fund METR go under as well.
But if Anthropic succeeds, METR and its parent organizations become more richly financed to regulate AI -- something those at METR want very much.
The "third-party evaluator" is not "third-party" at all.
The evaluator is on Anthropic's payroll.
If this were the end of it, that's bad.
But that isn't all.
The same organizations that fund METR also fund the many organizations, such as the Tarbell Center, that promote AI Doom.
The Tarbell Center publishes AI Doom articles in The Verge, Science, LA Times, The Dispatch, TIME, and others.
They are selling the problem, and then selling the solution to the problem -- from the same money pile: Anthropic's.
All of these organizations are financially dependent on the same exploding $7 billion money pile.
As Anthropic grows more and more powerful, its AI Doom Machine grows better and better financed -- louder and louder.
Meanwhile, the regulatory regime seeded in METR grows larger to solve the increasingly loud -- now hysterical -- problem of AI Doom that the Anthropic Network itself created.
From this standpoint, as Anthropic becomes more powerful, AI might be getting scarier, sure -- but the positive feedback loop also becomes more deafening -- independent of objective facts.
This itself is an objective fact.
The deafening AI Doom is part of an business model, that, as it expands, so too does the AI Doom messaging -- there is simply more money to do it.
But the problem also goes in the other direction:
If Anthropic dies, the Regulatory Regime and the AI Doom Machine are crippled or die.
Neither METR nor Tarbell nor the other organizations in the Anthropic Network can allow that to happen.
Hence, neither METR or the AI Doom Machine can be trusted to provide independent assessments of Anthropic's models or AI more broadly.
They simply are not organizations independent of Anthropic.
And Anthropic cannot detach itself from METR or Tarbell or countless other safety orgs (not shown here), either, because they drive hype for the models and the possibility of eventual regulatory capture, and Anthropic will not give that up willingly.
What's more, the people at all of these organizations are all the same ecosystem, the same community. They just shuffle between organizations.
The Anthropic Network is therefore, so long as it is successful, locked into a self-amplifying feedback loop inside an ideological monoculture.
And that feedback loop is winning.
That's what Jacob Coxon is.
China is keeping messaging tight. That is why optimism for AI is so high in China.
America has Anthropic: a massive company pushing anti-AI propaganda at a state level.
Anthropic will either create hysteria until American AI slows down and China wins, or it will create fractures throughout American society with severe political consequences.
Ironically, because of the structural financial incentives underpinning the Anthropic Network, it has become the same kind of self-amplifying virus that it fantasizes AI to become in the future -- while hiding its tracks just as carefully.
It is the mirror of the same AI virus that it hypothesizes to consume America.
Anthropic's business model, models itself after the very thing it claims to fear.
Except Anthropic's ideology infects humans, not computers.
Congress must investigate.
Evidence and Github in next post.
Then some supplementary figures.
IMO Fable 5.1 is still far superior to GPT-6 in web development. Need to figure out what to leverage GPT-6 for...
It’s really difficult for me to imagine models getting much better than Fable 5.1 / GPT-6. Like what would that even look like? They’ll never have human taste/intuition but from a technical perspective they’re way ahead of me as it stands.
I've got this paranoia that one day the era of subsidized tokens is going to end, so I've been trying to exhaust every last token before limit resets. Haven't used Claude much this week so looks like I'll be spamming Fable 5.1 agents like crazy tonight.
