If you think you understand something then you're probably not standing far enough away from it.
Joined August 2009
- Tweets15K
- Following611
- Followers493
- Likes16.7K
“Trouble rather the tiger in his lair than the sage among his books. For to you kingdoms and their armies are things mighty and enduring, but to him they are but toys of the moment, to be overturned with the flick of a finger.”
― Gordon R. Dickson, Tactics of Mistake
Robot Archie🇺🇦🇪🇺💙 retweeted
This is why I love the UK; we all agree when HM Government decided to jump over a cliff and commit suicide. The UK has, with Airbus UK in Stevenage all it takes to keep a key capability in the UK, ITAR-free, and also in EUROPE. Even if it does cost more now, it will pay in the future. UK has large medium satellites with Airbus and Surrey Space Technologies and micro satellites with Clyde Space in Glasgow. Alongside them, a plethora of companies are working on different subsystems. Write to your MP and stamp your feet; this is the time to do it. Made in the UK.
Despite a direction from me when Defence Secretary that the Skynet 6A , £2bn contract was too important to the UK space industry to offshore to Lockheed Martin it looks as if Wes is preparing to place the contract overseas. The damage it would do to Stevenage and the UK satellite industry would be irreversible and it was always assessed it could mean the end of the sector. Lockheed’s promises of factories in Newcastle were skin deep. The differences between types on offer were marginal. No.10 could, and should stop this. @AirbusDefence @wesstreeting @larisamlbrown @kevinbonavia @haynesdeborah
Robot Archie🇺🇦🇪🇺💙 retweeted
This is a fantastic idea - a mass attempted replication of old findings of LLMs with new models. Much conventional wisdom is outdated.
Recent models no longer invent citations and both hallucination rates & stereotyping has dropped. But they are still sycophantic & lean left.
Robot Archie🇺🇦🇪🇺💙 retweeted
The only thing bigger than Election Night drama this week could be what happens afterward…
I spoke with @JimSciutto tonight about possible scenarios in the aftermath of the 2026 midterms.
Question: If the midterms don’t go the Republicans’ way, will Donald Trump, the justice department and others not recognize the results?
@FrankLuntz “That is the right question to ask. And I'm afraid of the result because in the end, a democracy, a successful democracy depends on the acceptance of the people that governs and their trust and confidence in how they're chosen. And if we have another election like 2020, where the public doesn't believe that every result, every vote was cast accurately was counted fairly. And in the end the result was effective. If we don't believe that, then our country is going to be really hurting.”
Robot Archie🇺🇦🇪🇺💙 retweeted
There was a time when this kind of thing would have been considered a major news story
It begins. theguardian.com/us-news/2026…
⬇️🧵
Hey @Grok, concise, no spin.
How many DOCUMENTED CASES exist of U.S. public schools secretly performing gender-transition surgeries on children without parental consent, as Trump repeatedly claims?
NUMBER ONLY.
Robot Archie🇺🇦🇪🇺💙 retweeted
I think ChatGPT is just the latest tool every generation swears will make students stupid.
The list is older than you'd think:
> Around 370 BC: in Plato's Phaedrus, Socrates warns that writing will make people forgetful, because they'll trust marks on a page instead of their own memory.
> 1545: Swiss scholar Conrad Gessner complains about the "confusing and harmful abundance of books" that printing created.
> 1986: math teachers picket outside the national math teachers' conference in Washington, D.C. to keep calculators out of elementary school. One sign read "The Button's Nothin' Till the Brain's Trained."
> 2022: ChatGPT.
Here's the uncomfortable part. Each of them was a little bit right.
We did stop memorizing long poems. We did end up drowning in books. Most of us reach for a phone to split a dinner bill.
But nobody today would give back writing to get their memory back.
The tool never decided what happened. What students were asked to do with it did.
The calculator protesters had the best line of all of them.
The button's nothin' till the brain's trained.
That's still the whole AI debate, in seven words.
Which side would you have been on in 1986?
Robot Archie🇺🇦🇪🇺💙 retweeted
Ken Paxton: The criticism has been that I let a child predator out after 1 day. Well guess what? They served 60 days.
Readers added context they thought people might want to know
Paxton's office proposed that Adam Hoffman serve 1 day in jail as part of his plea deal. The judge sentenced him to 60 days instead. He was released after serving 29 days with good behavior credit. texastribune.org/2026/05/19/ken… kwtx.com/2026/05/25/for… ibtimes.co.uk/ken-paxton-hof…
Robot Archie🇺🇦🇪🇺💙 retweeted
Introducing a slightly older, wiser Claudia... and a fairly exact expression of the combined exhilaration and terror I feel right now about AI. Does this speak to anyone else? (Sound on!)
Robot Archie🇺🇦🇪🇺💙 retweeted
Dementia J. Trump thinks three people a week were drowning in the reflecting pool.
Which at its most perilous, in the middle, is 30 INCHES DEEP
Robot Archie🇺🇦🇪🇺💙 retweeted
🚨 NEW: Fox News have lost one-third of their viewership ahead of the US midterms
This comes after CNN recorded a 19% increase
Robot Archie🇺🇦🇪🇺💙 retweeted
How long does a finding about LLMs stay true?
We re-tested 50 widely cited findings about LLM behavior (various biases, hallucination, etc). Within 15 months, most no longer held or had substantially reduced effects.
LLM findings has short shelf life zou-group.github.io/llm-clai…
Great job by @oq_35!
Robot Archie🇺🇦🇪🇺💙 retweeted
Ukraine has deployed a large-scale electronic warfare “dome” across Kyiv to disrupt MESH radio communications used to remotely control Russian Shahed drones. According to radio technology expert Serhii Flash, hundreds of antenna and EW equipment locations were installed throughout the capital in just weeks. The effort involved major Ukrainian security and defense institutions and reportedly forced Russia to pause, upgrade its Shaheds and seek alternative technical solutions. #Ukraine
Robot Archie🇺🇦🇪🇺💙 retweeted
Absolutely incredible: MIT researchers say GPT-6 Astra improved a small Qwen model’s math score from 7.9% to 48.1% by directly editing just 0.016% of its weights, without gradient-based training.
The result points toward AI systems that can diagnose and improve other models directly, a potential step toward self-improving AI.
We asked GPT-6 Astra to edit Qwen3.5-0.8B weights to improve math.
It directly changed only 0.016% of weights, fixing 0.8B's reasoning loops & improving math up to 6x and even instruction following 3x, without gradients or training.
We call this Weight-Space Programming: frontier LLMs can now natively read/write neural network weights! 🧵
Blog: rikiyat.github.io/blog/weigh…
⬇️🧵
🚨New paper: "What Makes LLM Agent Simulations Useful for Policy Practice?"
@simile_ai is making a great launch, but can these LLM agent simulations actually help real institutions make better decisions?
We spent a year working with policymakers to answer this simple question.
The answer is yes—but perhaps not how you'd expect.
👇 THREAD 👇
[Link to paper: arxiv.org/abs/2509.21868]
[1/n]
⬇️🧵
We built a flight simulator for policy.
Today we introduce WhatIf: 10,000+ LLM agents you can steer, rewind, and interview in real time, so you can test a consequential decision before it touches a single real person.
#UIST2026 🧵
Robot Archie🇺🇦🇪🇺💙 retweeted
In a stunning innovation, Greater Manchester’s police have discovered that if officers turn up to burglaries (instead of screening them out as “low-level”), talk to people, look at CCTV, do forensics and chase the local career criminals, they solve nearly three times as many as the national average and cut break-ins in half. Truly novel. Who knew the beating heart of policing might involve neighbourhood bobbies catching local burglars? A model for the ages.
