CTO at Bekind Labs, a software consultancy in Tokyo full of ex-Pivotal Labs folks. Love trail running and language learning. Building Context in my spare time.
Tokyo
Joined November 2008
- Tweets45.3K
- Following851
- Followers1.8K
- Likes2K
Andrew Wright retweeted
🚨 Elon Musk has finally complied with the demands and provided the list of shareholders of X Holding Corp who helped purchase Twitter. And who do you think turned out to be investors in the X platform, besides Islamists and Trump's lobby? That's right, Petr Aven and Vadim Moshkovich. Now it is absolutely obvious where Musk's love for Russia comes from. 🙄‼️
Let me remind you who Aven and Moshkovich are.
Petr Aven is a Russian billionaire and founder of Alfa Group, which is one of Putin's main financial backers. This man is completely controlled by the Kremlin. Aven is one of Putin's oldest friends. In fact, without Petr's intervention, Putin would have ended up in prison instead of becoming president. After all, Aven covered for Vladimir Putin in 1992, when Putin organized illegal trading in export licenses. Aven also attended a meeting with Putin on February 24, 2022, marking the beginning of the invasion of Ukraine.
‼️ Vadim Moshkovich is a sanctioned Russian agricultural billionaire who was a member of the Federation Council of the Federal Assembly of the Russian Federation. In other words, he officially belonged to Putin's innermost circle. Moshkovich owns Rusagro, Russia's largest agricultural holding company. ‼️
Another interesting figure is Al-Waleed bin Talal Al Saud, who appears on the list of investors no fewer than three times, both personally and through his company, Kingdom Holding Company. This man continues his investment business in Russia and privately supports Hamas in its war against Israel. In his investments in Russia, he regularly cooperates with Rosneft, Lukoil, and Gazprom.
✅ Everyone was wondering how it was possible that Twitter (X) shares were falling, advertisers were terminating contracts, and nobody was doing anything. After all, this was causing direct financial damage to investors.
But everyone overlooked the fact that the investors did not care about profits. They needed the platform to promote their narratives, such as supporting Putin and Russia, lobbying for a freeze of the war in Ukraine, pushing Ukraine-related issues out of the top news stories, justifying Hamas's activities, and so on. 😳
Andrew Wright retweeted
One of the men in this picture is a politician currently serving 10 and a half years in prison for accepting Russian bribes. The other was his boss and doesn't think there's evidence of Russia interfering in UK politics.
Andrew Wright retweeted
Theo's right about everything but he's missing one thing.
Right:
- It is (annoyingly) a good model. Reality would be simpler if this weren't the case, but no, it's actually very good.
- It is horrendous at UI
- It speaks very nicely; feels smart, heavy, and big
Wrong:
- THERE IS NO SUBSCRIPTION SUBSIDY. OMG THIS THING IS FUCKING EXPENSIVE.
Theo seemingly has an API key that is giving him unlimited access to the model so he wasn't able to get a good sense for pricing.
I first started with the $5 plan, couldn't finish a single turn, upgraded to $15 plan, couldn't finish more than a couple turns, upgraded to the $50 plan, couldn't finish the whole request, and finally upgraded to the $200 plan an hour ago, and IT'S GONE (5 hr limit).
Afaict there is no actual reason to use Muse Wonder (Full) at all at these prices.
If Meta's $200 subscription quota is really what it costs to use AI for coding (and it probably is), then we are all in for a very, very rude awakening when the token subsidies end.
Andrew Wright retweeted
Dusting off this tweet from April because this gap in shared understanding of LLM capability is *widening*. It's now less "two groups of people speaking past each other" and more a sharp funnel.
- Napkin math somewhere around 6B people (~75% of the population) have barely come in contact with LLMs at all.
- Around 1-2B (~20%) are casual and infrequent users of free-tier ChatGPT-like products. This group sees derpy chatbots and treats them a bit like a better Google search, a writing aid, or etc. Maybe an agent tries to book you a flight. I have non-tech friends who (reasonably, imo) say they have not much use for it at all. Even many professionals outside of math&code are in this tier. For example, execs and many other functions spend a lot of their time talking to other people, so while they understand what is happening intellectually, it is still second-hand and a bit abstract.
- Now we get to professional use of frontier-grade LLMs in math&code. Somewhere around 20M people (0.2%) see first-hand that large, complex projects that used to take them weeks/months can now be completed by agents with a prompt. Building apps, copying apps, translating apps, decompiling apps from binaries... This has all happened very quickly and recently - less than 1 year ago, I was writing code manually by hand, typing memorized computer code commands into a code editor character by character, occasionally pressing Tab to autocomplete a little chunk of code.
- And finally we get to the ~5,000 people (~0.00006%) with access to frontier-grade systems internally. The external world has seen the preview. It looks like swarms of thousands of agents collaborating over weeks on software mega projects: minting zero days, running cyber attacks and defenses at machine speeds, discovering new science, advancing the frontier of mathematics. Things that would have taken top professionals in the industry years of work. Meanwhile, human review and comprehension are starting to fall behind. For example, people are still involved in the "archeology" of the OpenAI-HF incident from many months ago. Mathematicians may be poring over the 722 manuscripts on frontier mathematics for a while.
The funnel is driven by a combination of factors:
1. The impact scales with ambition, problem size, and horizon. A question with a paragraph answer barely stresses the system. You need a reservoir of big, difficult problems that you really care about. This aspect drives the consumer / professional dimension of the funnel.
2. The jaggedness of the system (which I have written about a lot separately). Capability peaks in domains that are digital, verifiable and economically valuable. This is because LLM capability emerges from reinforcement learning on verifiable rewards on a curated environment mixture driven by revenue potential. This aspect primarily drives the area (e.g. math&code) dimension of the funnel.
3. Access. Free-tier, paid-tier, internal.
So this is the weirdness of the moment. The general public has mostly not interacted with these systems. When they have, it looks like a derpy chatbot. The majority of professionals still see only a modest uplift. And a small sliver of professionals are experiencing the vertigo of the curve going vertical. And it is all happening at the same time.
Judging by my tl there is a growing gap in understanding of AI capability.
The first issue I think is around recency and tier of use. I think a lot of people tried the free tier of ChatGPT somewhere last year and allowed it to inform their views on AI a little too much. This is a group of reactions laughing at various quirks of the models, hallucinations, etc. Yes I also saw the viral videos of OpenAI's Advanced Voice mode fumbling simple queries like "should I drive or walk to the carwash". The thing is that these free and old/deprecated models don't reflect the capability in the latest round of state of the art agentic models of this year, especially OpenAI Codex and Claude Code.
But that brings me to the second issue. Even if people paid $200/month to use the state of the art models, a lot of the capabilities are relatively "peaky" in highly technical areas. Typical queries around search, writing, advice, etc. are *not* the domain that has made the most noticeable and dramatic strides in capability. Partly, this is due to the technical details of reinforcement learning and its use of verifiable rewards. But partly, it's also because these use cases are not sufficiently prioritized by the companies in their hillclimbing because they don't lead to as much $$$ value. The goldmines are elsewhere, and the focus comes along.
So that brings me to the second group of people, who *both* 1) pay for and use the state of the art frontier agentic models (OpenAI Codex / Claude Code) and 2) do so professionally in technical domains like programming, math and research. This group of people is subject to the highest amount of "AI Psychosis" because the recent improvements in these domains as of this year have been nothing short of staggering. When you hand a computer terminal to one of these models, you can now watch them melt programming problems that you'd normally expect to take days/weeks of work. It's this second group of people that assigns a much greater gravity to the capabilities, their slope, and various cyber-related repercussions.
TLDR the people in these two groups are speaking past each other. It really is simultaneously the case that OpenAI's free and I think slightly orphaned (?) "Advanced Voice Mode" will fumble the dumbest questions in your Instagram's reels and *at the same time*, OpenAI's highest-tier and paid Codex model will go off for 1 hour to coherently restructure an entire code base, or find and exploit vulnerabilities in computer systems. This part really works and has made dramatic strides because 2 properties: 1) these domains offer explicit reward functions that are verifiable meaning they are easily amenable to reinforcement learning training (e.g. unit tests passed yes or no, in contrast to writing, which is much harder to explicitly judge), but also 2) they are a lot more valuable in b2b settings, meaning that the biggest fraction of the team is focused on improving them. So here we are.
Andrew Wright retweeted
Reading the comments of junior mathematicians confirms what many senior mathematicians have been warning about over the past month:
It is becoming clear that people with the intellectual ability to understand these proofs will not be interested in spending their lives training to check LLM logs.
This is the tragic side of what is happening that almost no one — except the actual mathematicians — seems to get.
Either we figure out how to keep people interested in doing mathematics, or in 20 years LLMs will have no one’s work left to recombine into new proofs, and no one interested in reading those proofs.
That would be the thermal death of mathematics.
The argument that “LLMs will read LLM math” is so shortsighted that it makes me want to cry…
Andrew Wright retweeted
Apply this situation to the economy and you see why there are big problems ahead
The situation:
-two math whizzes leverage the latest *publicly available* AI to work on solving Navier-Stokes. They’re onto something.
-openAI sees it, wants to win, so puts 10,000 agents from their latest *private* model that is already light-years ahead of what’s publicly available on it
-they solve it in just a few days
The economy problem
-let’s say I leverage the latest publicly available model and find a way to detect skin cancer. I’ve spent tens of thousands on tokens, proven it can work. I get investors and build a company. A path to saving many lives and making hundreds of millions is clear
-openAI (or Google or Meta or Anthropic) see this hundred million opportunity and point their private most-advanced AI model and 10,000 agents toward my idea. In a few days, they win.
Repeat this thought experiment for pharma, legal, medicine, advanced materials, software…
This model represents a step-function improvement on many benchmarks, and its training is ongoing.
Our internal model group arrived at the Navier–Stokes solution in 88 hours, using around 10,000 coordinating AI agents.
Throughout the effort, we maintained the strict safeguards—including monitoring and isolation—that we apply to all our frontier evaluations.
Andrew Wright retweeted
Two extraordinary charts from today’s PISA test results:
1) School test scores continue to collapse internationally, underscoring how this is no longer a Covid effect but sustained decline.
Those falls in reading and maths are equivalent to about two years of lost schooling.
Andrew Wright retweeted
Is Latin really so different from Italian?
I have seen kids doing this 1000x better than me, but just to give you an idea:
“Il buon padre di famiglia è un uomo rigoroso, leale, onesto e determinato che cura con dedizione la propria famiglia”
Drop the articles:
“Buon padre di famiglia è uomo rigoroso, leale, onesto e determinato che cura con dedizione propria famiglia.”
Restore the old endings:
“Bonus pater de familia est homo rigorosus, lealis, honestus et determinatus qui curat con dedicatione propriam familiam.”
Turn di into a genitive and tidy the preposition:
“Bonus pater familiae est homo rigorosus, lealis, honestus et determinatus qui curat cum dedicatione propriam familiam.”
Move the verbs later:
“Bonus pater familiae homo rigorosus, lealis, honestus, determinatus est qui propriam familiam cum dedicatione curat.”
😇
Andrew Wright retweeted
I hereby declare the "learn to code" era officially dead:
Big declines in the number of people studying computer science in the last year or two 📉
Chart from this week’s edition of our newsletter on AI and the labour market ft.com/content/9183a122-10b7…
Andrew Wright retweeted
I'm in awe of how sociopathic this comes off. What if I... actually enjoy talking to my family about their lives and interests?
Andrew Wright retweeted
A Brown professor gave his students a take-home midterm exam. After suspecting many cheated using AI, he made the final in-person. The orange dots are the midterm scores and the gray dots are the final scores. Looks like all but 3 cheated on the midterm.
Andrew Wright retweeted
The *median* wealth of a French is almost *twice* the median wealth of an American.
So much for the "Europoor" meme.
The *average* American wealth is high because a tiny number of Americans are absurdly rich. But most Americans get screwed.
Andrew Wright retweeted
Rather good summary of Don Jr’s corruption from Hunter Biden, which is not a sentence I ever imagined typing out even two years ago.
I hope everyone had a great 4th of July. I know @realDonaldTrump and family did.
250 years ago we declared independence from a king who ran the colonies as a family business. In just 18 months the Trumps have made King George look like an amateur.
A $620 million Pentagon loan, the largest in the program’s history, to a company Don Jr.’s firm bought into three months before.
An Air Force drone contract to a startup the princelings took public through a golf course company they own a piece of.
The Army’s largest drone motor order ever, to a company where Don Jr. sits on the board and holds millions in stock.
A $24 million Pentagon robotics contract to the company that employs Eric as Chief Strategy Advisor.
A stake in the largest undeveloped tungsten deposit on earth, in Kazakhstan, backed by $1.6 billion in US government support.
Jared’s fund seeded with $2 billion from the Saudi crown prince, now $6.2 billion, 99% of it foreign money from Gulf governments. Over $110 million in fees collected from the Saudis alone. He negotiates American foreign policy with the governments that pay him.
$2.3 billion from crypto ventures their father regulates. More than a million people bought in and lost $2.3 billion. The money didn’t grow. It simply moved from the subjects pockets to the crown’s coffers.
And the next one is already drafted. A proposed ATF rule that will allow guns to be shipped straight to your front door. The government’s own estimate is 3.3 million home gun deliveries a year. Don Jr. sits on the board of the online gun megastore built to cash in. He holds 300,000 shares.
And that’s only the fraction they’ve allowed us to see. Not one subpoena served. Not one search executed. Why hide anything when you own the investigators?
Me? They searched a laptop for six years. Federal prosecutors. Grand juries. Subpoena power. Congressional hearings. They found nothing. I made about $200k a year selling paintings when my Dad was President, and they made my paintings part of an impeachment inquiry.
For six years they’ve asked Where’s Hunter? What about the laptop?
Wrong questions. The right one is 250 years old. Does America belong to a family?
They’ve given their answer. Long live the King.
Andrew Wright retweeted
This has the capacity to ruin the knock out rounds of the World Cup. It’ll be too hot to play
A major heat dome kicks off July next week with the hottest air this season expected for many along the eastern half of the US. Heat indexes next Wednesday and Thursday are expected to range from 105-110 outside of the Appalachians, with instances of 110+ possible especiallly across the Carolinas and the Delmarva!
Andrew Wright retweeted
wouldn’t that also make the tools better for humans
I really want CLI tools to start speaking directly to coding agents with errors. Instead of:
“Error:”
Return:
“Here’s what broke. Here’s how to investigate. Here’s how to generate a redacted repro. Here’s where to send it.”
Obviously there are privacy/security challenges to solve. But if we get this right, we can have a virtuous software improvement loop. Every failed agent interaction becomes a high quality bug report. Agents find the bugs. Agents fix the bugs. Software gets better faster.
