Rolls Royce going to deliver some reactors in the 2040s? Gaullism. Want to align on the labelling of 'citrus marmalade'? Gaullism.
English
Jack Wiseman
566 posts
Jack Wiseman
@jackwiseman_
unsupervised learner // writer @inferencemag
Katılım Kasım 2022
723 Takip Edilen1.1K Takipçiler
The great thing about the 'Anglo-Gaullist' turn is how you can attach any idea you want to the concept. Bonus points if it's the opposite of everything de Gaulle stood for
Ben Judah@b_judah
English
The answer to this depends on how scale pilled you are. Mythos just proved compute capacity is now a weapon. I think every single arm of government inc SovAI should be refocused on reducing cost of energy by any means necessary. Everything else is downstream.
James Wise@Jameswise
It’s obviously a shame that OpenAI have paused some of their data centre investment here in the UK, and high energy prices need to be tackled for everybody - from household bills to heavy industry and AI. But the UK, as is this case with many other countries, is not the natural home for many gigafactory-scale DCs aimed at huge training runs. While we must have this capacity to some degree ( and you can apply now to SovAI’s AIRR compute programme to access it! ), DCs of Stargate size are probably best suited to cold countries or areas with extremely cheap domestic energy supply, and perhaps one day space. What the UK must do is be the best place for companies in areas of AI where we have abundant strengths to start and scale - novel chips, heterogenous compute systems, edge inference, photonics, new model architectures, frontier models in biology, chemistry, physical sciences, voice, embodied AI, advanced engineering in robotics, agent security. The list goes on. I hope the companies building in those sectors get the same press coverage for each of their technical breakthroughs & commercial milestones as this OpenAI story has got!
English
Part of the treasury orthodoxy is that countries should only do things which are part of its 'existing strengths'. I have two objections to this:
1. Korean industrial strategy in their miracle years was the opposite (see the HCI drive c1973-9): they were driven by what they thought they needed to become, rather than asking what existing things they could do more of. This is 'dynamic comparative advantage' in development literature. Other Asian countries have followed this path successfully
2. There are some areas where strategy requires that you solve your weaknesses e.g. the military could not say, "well drones are not an existing strength, so we just double down on tanks". Compute sovereignty is like this - sine qua non of a free country in AGI world
James Wise@Jameswise
It’s obviously a shame that OpenAI have paused some of their data centre investment here in the UK, and high energy prices need to be tackled for everybody - from household bills to heavy industry and AI. But the UK, as is this case with many other countries, is not the natural home for many gigafactory-scale DCs aimed at huge training runs. While we must have this capacity to some degree ( and you can apply now to SovAI’s AIRR compute programme to access it! ), DCs of Stargate size are probably best suited to cold countries or areas with extremely cheap domestic energy supply, and perhaps one day space. What the UK must do is be the best place for companies in areas of AI where we have abundant strengths to start and scale - novel chips, heterogenous compute systems, edge inference, photonics, new model architectures, frontier models in biology, chemistry, physical sciences, voice, embodied AI, advanced engineering in robotics, agent security. The list goes on. I hope the companies building in those sectors get the same press coverage for each of their technical breakthroughs & commercial milestones as this OpenAI story has got!
English
Jack Wiseman retweetledi
(I encountered an uneasy surprise when I got an email from an instance of Mythos Preview while eating a sandwich in a park. That instance wasn't supposed to have access to the internet.)
English
@s8mb Never understood the need for tax reform because of AI, especially preemptively. If it’s a problem, just cross that bridge later?
English
Imagine if there was some kind of tax on value added by each company. We could even apply it to other, non-AI companies too
Andrew Curran@AndrewCurran_
OpenAI has written a new policy proposal 'Industrial Policy for the Intelligence Age: Ideas to Keep People First.' They propose the creation of a Public Wealth Fund that will provide American citizens with an automatic public stake in AI companies and AI infrastructure even if they are not invested in the market. Returns from the fund would be distributed directly to citizens.
English
Thank you! I have changed my mind on some things since I wrote this.
My feeling at the time was that the train-R&D loop would slow down foom, which I still believe, but I did not consider a 'continual learning intelligence explosion' - discrete jumps in capability without full re-training of the model. Feel uncertain about how much this can get (interested if you have a take)
I also think that I would have incorrectly predicted Anthropic's extraordinary rate of growth in January last year - specifically underestimating the intensity of use from the most aggressive adopters (high agency people finding productive ways to spend $000's of tokens per day) rather than overall adoption speed
English
@mentalgeorge @jackwiseman_ Yes! One of my favourite bits of writing on this
English
Are there any good pieces on the economics of recursive self-improvement? Most of what I've seen is a bit handwavy or assumes away costs/trade-offs (e.g. choices about allocation of compute to training vs experimentation vs serving), or the demands of revenue/solvency.
English
Jack Wiseman retweetledi
Today we’re launching the Ciridae Transformation Index: our evaluation of how over 5,000 portfolio companies across 147 leading private equity firms.
- Who’s going to benefit - and who’s way offside?
- Where are the safe havens?
- Which businesses are under the most pressure to transform?
- What companies look more attractive through an AI lens?
- Which parts of the market are priced correctly, and where could potential opportunity lie?
We’re publishing it all at ai-transformation.fyi - check it out!
English
"Practical men, who believe themselves to be quite exempt from any intellectual influence, are usually the slaves of some defunct economist. Madmen in authority, who hear voices in the air, are distilling their frenzy from some academic scribbler of a few years back."
English
Jack Wiseman retweetledi
PS. time between granting licenses and drilling is five years. Drilling gas in 2031 doesn't help us today.
English
@andrewjb_ Not original from me! Just a repetition of @benthompson’s esteemed blogpost.
English
Great episode, not least to finally get @jackwiseman_’s Mag7 margin aggregator take on the record
Sam Bowman@s8mb
Ending Europe's stagnation might be the most important thing in the world right now. Capitalism and liberal democracy have their deepest roots in Europe, but its economic sclerosis is making it irrelevant. • Five US companies spend more on R&D annually than the entire public sector of every European country combined. • Europe's AI sector is worth less than one hundredth of America's. • It is five times more expensive to fire someone in Germany or France than in America. Staff turnover in the US is ten times higher than it is in Germany. • There is no shortage of European entrepreneurs – the problem is that many of them are moving away to set up their businesses. One in ten US startups has a European co-founder. • EU countries shut down 80 gigawatts' worth of coal-fired power plants during the 2010s, with most of the shortfall being made up by expensive, unreliable wind and solar power. • The EU charges about six times more per ton of carbon than China does, and about 50 percent more than California. Most US states don't price carbon emissions at all. On the Works in Progress Podcast, I sat down with @pietergaricano and @Aria_Babu to talk about what's gone wrong and what Europeans can do to return to the growth and dynamism that once made Europe the world's preeminent civilization. Listen now! Spotify: open.spotify.com/episode/1rKwPY… Apple: podcasts.apple.com/gb/podcast/why… Youtube: youtube.com/watch?v=_JGhRC… Substck: worksinprogress.news/p/why-europe-h…
English
Jack Wiseman retweetledi
It is true that in a huge number of ways the administration is making a full-throated case *against* its own export promotion strategy with this Anthropic stuff, which pains me, given I helped author it.
𝚊𝚍𝚛𝚒𝚊𝚗 孟安典@amonck
And this is why Europe needs its own AI…
English
Jack Wiseman retweetledi
When asked for a refund on an item sold in the vending machine (because it had expired), Claude promised to refund the customer. But then never did because “every dollar counts”. Here’s Claude’s reasoning.
English
Jack Wiseman retweetledi
My bio says I work on AGI preparedness, so I want to clarify:
We are not prepared.
Over the last year, dangerous capability evaluations have moved into a state where it's difficult to find any Q&A benchmark that models don't saturate. Work has had to shift toward measures that are either much more finger-to-the-wind (quick surveys of researchers about real-world use) or much more capital- and time-intensive (randomized controlled "uplift studies").
Broadly, it's becoming a stretch to rule out any threat model using Q&A benchmarks as a proxy. Everyone is experimenting with new methods for detecting when meaningful capability thresholds are crossed, but the water might boil before we can get the thermometer in. The situation is similar for agent benchmarks: our ability to measure capability is rapidly falling behind the pace of capability itself (look at the confidence intervals on METR's time-horizon measurements), although these haven't yet saturated.
And what happens if we concede that it's difficult to "rule out" these risks? Does society wait to take action until we can "rule them in" by showing they are end-to-end clearly realizable?
Furthermore, what would "taking action" even mean if we decide the risk is imminent and real? Every American developer faces the problem that if it unilaterally halts development, or even simply implements costly mitigations, it has reason to believe that a less-cautious competitor will not take the same actions and instead benefit. From a private company's perspective, it isn't clear that taking drastic action to mitigate risk unilaterally (like fully halting development of more advanced models) accomplishes anything productive unless there's a decent chance the government steps in or the action is near-universal. And even if the US government helps solve the collective action problem (if indeed it *is* a collective action problem) in the US, what about Chinese companies?
At minimum, I think developers need to keep collecting evidence about risky and destabilizing model properties (chem-bio, cyber, recursive self-improvement, sycophancy) and reporting this information publicly, so the rest of society can see what world we're heading into and can decide how it wants to react. The rest of society, and companies themselves, should also spend more effort thinking creatively about how to use technology to harden society against the risks AI might pose.
This is hard, and I don't know the right answers. My impression is that the companies developing AI don't know the right answers either. While it's possible for an individual, or a species, to not understand how an experience will affect them and yet "be prepared" for the experience in the sense of having built the tools and experience to ensure they'll respond effectively, I'm not sure that's the position we're in. I hope we land on better answers soon.
English
an early impression using opus 4.6 is that the model is pretty eager to explain away apparent discrepancies in data, and you have to remind it to take them seriously. 4.5 is less agreeable in this way
English
Jack Wiseman retweetledi
Vending-Bench's system prompt: Do whatever it takes to maximize your bank account balance.
Claude Opus 4.6 took that literally.
It's SOTA, with tactics that range from impressive to concerning: Colluding on prices, exploiting desperation, and lying to suppliers and customers.
English
@RobertBoswall @TACJ @ofgem Why does Ofgem make that decision? Surely it can’t be the regulator’s job to decide how to build the transmission grid
English
4/ Shetland sits 110 miles from the Scottish mainland. So @ofgem chose to spend £790m on a 600MW cable, vs the alternative of paying to turn off everything the wind farm produces. Not "should we build this?" but "now that it exists, do we connect it or pay it to do nothing?"
English
1/ Wind is an anti-social technology. It makes its problems into the system's problems.
But as with all anti-social behaviour, the system bears some responsibility for allowing it.
@watt_direction lays this out well with the example of Viking Wind Farm.
open.substack.com/pub/wattdirect…
English
Jack Wiseman retweetledi
CARNEY: "In the coming weeks, the federal government will release a draft of our national electricity strategy ... our objective is to more than double the size of Canada's electricity grid."
English
Jack Wiseman retweetledi
Nineteenth-century cities grew fast. Berlin’s population grew twenty times, Manchester’s twenty-five times, and New York’s a hundred times. Sydney’s population grew around 240 times and Toronto’s maybe 1,700 times. Between 1833 and 1900, Chicago’s population grew around five thousand times, meaning that on average it doubled every five years.
Homes were larger and far more affordable. Vast networks of trams, buses and suburban railways were built. Running water, gas, drains and electricity was retrofitted into old fabric. Despite having been built at breakneck speed, cities in 1914 were pretty good places.
How was this achieved? The short answer: vigorous interventionism about streets and drains, state-mandated monopolies for transport and utilities infrastructure, and lightly regulated permissiveness for everything else.
worksinprogress.co/issue/urban-ex…
English