Jamie
@jamiec_24
@sotalikesfuture cofounder | msc @imperialcollege | @join_polaris fellow
I've been yapping for months about bad evaluation setups and how results/AI behaviors are reported, and this new @AISecurityInst paper does so much more clearly. In short: There's a massive difference between showing a model can do something sketchy versus showing it tends to…
😈 Here's why you should not worry that models will start blackmailing you out of nowhere: 1. At their heart, LLMs are pattern-matching and prediction engines. Given an input, they predict the most statistically likely continuation based on the vast dataset they were trained on.…
How to decentralise scientific information sharing away from incentives of the legacy for profit journal cartel without succumbing to the incentives of the social media system is a v important question imo. Useful thread here.
"Science has many bad features because of the bad incentives of journals" risks becoming "Science is even worse because of the [sometimes] worse incentives of social media". I'm optimistic about creating healthy new media forms, but science-with-X-as-target has many downsides
Had this idea for a while that's perfect to explore for this hack. Please DM if interested in teaming up! It's inspired by 2 articles: - One is @andy_matuschak's piece on how books and lectures don't work for acquiring and retaining knowledge, because they weren't created with…
SoTA’s Human Augmentation Hackathon, 26-27th July We’re seeking new tools that foster human-AI symbiosis. What you build will expand our capabilities, rather than writing us out of the story. Success will be judged against the following criteria: - Is your demo better than…
@sotalikesfuture have a solution for this in the next edition of Letters. Make sure to subscribe to get notified as soon as it's out shortly. Very grateful to get to collect transformational ideas from the community @jamiec_24 is pioneering.
Incredibly damning article about 1 of the Top 3 MBA Programs in the country. I’ll try to get some key excerpts here but I would suggest reading the whole article.
leaving out the years that he worked at AMD and LSI is a great eg of how a lionizing narrative works. it creates the impression of superhuman genius, which confers a valuable aura that has material impact on morale, negotiations, etc. Similar to “Singapore was a fishing village”
The CEO of Nvidia has a legendary LinkedIn profile:
Talent density at our Human Augmentation Hackathon will be incredible. We have applicants eager to work on copilots for scientific discovery, BCI, industrial robotics, memory & sensory enhancement, silent speech interfaces, opera composition -- all to extend human capability.…
SoTA’s Human Augmentation Hackathon, 26-27th July We’re seeking new tools that foster human-AI symbiosis. What you build will expand our capabilities, rather than writing us out of the story. Success will be judged against the following criteria: - Is your demo better than…
a timely new report from @BritishProgress ! I think highly-capable AI agents have the potential to transform the way militaries are run over the next decade, so there is huge scope for impactful defense + AI work that ATI can tackle by narrowing its focus
🚨 NEW REPORT 🚨 The Alan Turing Institute has visibly struggled, and is now at a critical juncture. The Secretary of State has intervened. What happens next? A new report for @britishprogress by me and @jujulemons sets out a plan: 🧵
An important element of the new @ESRC strategy we are working on is to make the promotion of economic growth an explicit goal of our work. I've written a short piece about why we're doing this, and what that might involve.
This was fun! You should especially make time to listen if you’d like to *be* an ARPA-style PM, nerd out over R&D procedures, or just want to know what’s going on at “ARPAs” around the world Will tweet out more clips over the weekend. Some of Jano’s answers excerpt wonderfully!
Had a great time talking to @eric_is_weird about @SPRIND, what we look for in program managers & "Jano’s obsessions with SPRIN-D’s administrative processes, such as their speed with “Fourteen days time-to-money” freaktakes.com/p/an-interview…
This is very good. underwriting-superintelligence.com/?july
You need to be side quest maxxing. Get your pilots license. Get scuba certified. Go sky diving. Get your coastal cruising certificate. Become a sommelier. Get every niche / pointless certificate & license you can.
this isn't just a funny joke, I think it's actually a very insightful point about the future of work
“Lawyers are going to be replaced by computers” um a computer can’t go golfing with the judge
my uncle worked for DARPA/NSF for many years. Every year the same general came to his office and pitched "Let's make mech warriors". He changed the phrasing and the concept every so slightly...but it always came out to "Let's make mech warriors." My uncle said no every year.
i could intuitively pilot a mecha.
New from our security teams: Our AI agent Big Sleep helped us detect and foil an imminent exploit. We believe this is a first for an AI agent - definitely not the last - giving cybersecurity defenders new tools to stop threats before they’re widespread.
DeepSeek was not a Sputnik moment in AI This graph is the Sputnik moment in AI
AI will "redefine how nations project influence, how threats emerge, and how we defend ourselves. It will therefore transform how diplomacy is conducted." EG. "AI is at the spearhead of hybrid threats like disinformation."
10 Downing Street is taking over the Design Museum to show potential applicants what previous Innovation Fellows have built during their tours of duty in government. Join us on 30th July to see the demos over drinks and meet the current Innovation Fellows & 10 Data Science team,…
@10DowningStreet recently wrote to SoTA's community. They're assembling a new, elite team of technologists to 'fundamentally rewire the state' and 'transform No10 and key areas of the government system through the applications of data science and AI'. Read the full letter to…
Kinda told this story but in 2012 me and my buddy Kevin made an unsolicited proposal for company management while working at Honda R&D, which directly led to them re-entering F1 6 months later. My boss, (furious that I would be doing such things) said I could never participate…
Share a piece of lore about yourself
seems cool (for my cracked 😤 british 😔 friends)
@10DowningStreet recently wrote to SoTA's community. They're assembling a new, elite team of technologists to 'fundamentally rewire the state' and 'transform No10 and key areas of the government system through the applications of data science and AI'. Read the full letter to…