Sabitlenmiş Tweet
Gabe
731 posts
Gabe
@Gabe_cc
Founder of @HTechInc, Advisor at @ControlAI.
Katılım Ekim 2019
93 Takip Edilen1.6K Takipçiler
Gabe retweetledi
Step 1: Stop living in SF - you get AI company brainworms
Step 2: Do not talk to claude
Step 3: No seriously, get out of SF - surround yourself with normal people
Step 4: Simply don't talk about eugenics or your fetishes
This gets you 90% of the way there.
Tyler is finishing a book, slow to reply@TylerAlterman
We spoke to a famous influencer today. Someone asked him how the AI safety scene can improve. He said something like "they can stop smelling the same as the people who are making AI." ...by which I took him to mean that they can differentiate themselves by being unmistakably human-maxxing, which starts not from caring about "humanity" (that's an abstract concept) but from caring about *your own idiosyncratic subjectivity* If you love "humanity" and can't connect with your own unique soul or others', then you don't love humanity, you love a linguistic construct, and you have already become the machine.
English
Gabe retweetledi
I think Plan S might be more politically tractable than Plan A.
(I distrust myself here because I separately think Plan S is much more likely to solve the underlying problem, and I risk halo-affecting myself).
But, my reasons:
1. In most worlds where you try to implement Plan A, I think it gets massively distorted and ends up being more like Plan C. The president and Xi Jinping need to actually deeply "get it" in a way that seems unlikely to me, for the nuances are to survive.
2. Plan S is pretty simple and easy to explain: "don't train more advanced AIs until there is strong consensus that it's a good idea."
3. I put decent odds on "Plan S has more support among voters."
Plan A requires continuously making nuanced judgment calls about what counts as safe, while generally maintaining momentum on training more powerful AIs (and leaving lots of dry tinder around). It's essentially an unsolved problem to make regulations careful enough to distinguish good vs bad safety cases, and I don't think the Plan A documents had particularly good ideas for how to do so.
Meanwhile, I expect tons of voters and politicians to understand "this AI stuff is scary, it's going to get more freaky fast, it's going to turn the world upside down, take your job, maybe kill everyone."
The thing they'll want is for the craziness to stop.
Meanwhile, Plan A is actually kinda freaky sounding. Centuries of progress in years? I lose my job but it's okay somehow? We hand off control to godlike AIs at the end?
Of the people who will actually understand it, I think the people who'd support it are niche futurists, niche libertarians (who maybe have good principles, but, sorry guys, don't actually have much political clout), and niche AI Safety-ists with particular combinations of beliefs about which parts of the problem are hard. (also sorry guys).
The larger political group that might support A over S are people broadly wanting to Make The Economy Go Faster who don't really understand or believe in the stakes. Whether than counts as "supporting Plan A" is kinda ambiguous.
English
Gabe retweetledi
Bravo! Charbel-Raphael speaks the truth: we know what to do-- we need the political will to do it.
Tbh most of the "safety research" people talk about is procrastination. Hoping against hope they'll find a technical shortcut to bypass the difficult process of facing the situation and acting, of waking up others, of fighting real foes. Virtuous busywork to excuse their inaction and delay a reckoning with their failed ideologies.
Charbel-Raphael@CRSegerie
This will probably be my most important intellectual contribution of the year. The current bottleneck for AI safety is political will, not research. And our field is not acting like it.
English
Gabe retweetledi
This will probably be my most important intellectual contribution of the year.
The current bottleneck for AI safety is political will, not research. And our field is not acting like it.
English
Gabe retweetledi
Picking which game to play is harder than ever because the rules keep changing every 6 months.
We were raised being told that becoming a lawyer, doctor, banker, consultant or climbing the corporate ladder was the ultimate success: stability, status, upward mobility, steady earnings growth, respectability. Then we were told learning to code was the best leverage in the world, right before the AI gurus started saying coding might be one of the first skills getting automated.
Now everyone says go all-in on AI, but even that feels speculative. Are you actually building durable leverage, or just doing temporary arbitrage until OpenAI / Anthropic ship your entire workflow as a feature? Everyone says quit corporate and start a company, but entrepreneurship now feels like the most crowded trade on earth, with everyone using the same tools, the same playbooks, the same AI agents, the same “build in public” advice. Everyone says build an audience, but then your identity becomes the product, the algorithm becomes your boss, and every part of your life slowly turns into content inventory.
That’s why ambitious people are so cooked right now. Not because they lack options, but because every option comes with a convincing bull case, a terrifying bear case, and a guy on the internet monetizing both.
No obvious map, no trusted authority, no stable definition of winning. Just 10,000 possible lives, all constantly repricing in real time, and the quiet suspicion that the one you picked might be the wrong game.
Aporia@aporia9n
Most ambitious people I know are mentally cooked, by the combination of infinite optionality and infinite comparison. They’re torn between climbing the corporate ladder, launching a startup, becoming a solopreneur, buying an SMB, moving to NYC, moving to London, moving to Dubai, staying close to family, optimizing for money, optimizing for freedom, optimizing for status, optimizing for peace. Then they open TikTok and see some 22yo flying business class like it’s nothing, or casually spending their monthly salary at a beach club. The issue is not just envy. It’s the feeling that there are 10,000 possible lives available, every one has someone younger apparently winning at it, and somehow you might be choosing the wrong one.
English
Gabe retweetledi
When I talk to people that instantly, deeply, get the risk from superintelligence, they often share a trait sometimes called the "security mindset"
I think this post by @l_mc_nally is a nice brisk touch of what it feels like to think seriously about security even briefly.
1/
English
Gabe retweetledi
Wow, even I was surprised how high AI ranked! Very good to see!
Tim Schnabel@TimSchnabel
Interesting poll of Hill staffers from @PunchbowlNews. 250 years is a long time! But interesting to see that "losing control of AI" is top 4 for both parties' staffers.
English
Gabe retweetledi
Interesting poll of Hill staffers from @PunchbowlNews. 250 years is a long time! But interesting to see that "losing control of AI" is top 4 for both parties' staffers.
English
Gabe retweetledi
Major update to the list of specification gaming examples in AI --we now have a total of 86 examples (including 13 recent examples from 2025-2026). Check it out here: tinyurl.com/specification-….
The list has been organized by type of AI system that produced the unintended behavior, with 4 categories: evolutionary and genetic algorithms, large language models, reinforcement learning, and other.
Help us get to 100 examples - submit new incidents to this form: docs.google.com/forms/d/e/1FAI…
English
Gabe retweetledi
"This is to argue, and I profoundly agree, that the ultimate consideration for any policy towards AI is human supremacy. If AI is to help us rather than destroy us, it must be ‘oriented toward the good’"
Humans must remain in control - this is no longer to be taken for granted!
Danny Kruger@danny__kruger
My essay 'Britain and AI: the Patriotic Compact', published today - dannykruger.substack.com/p/britain-and-… Summary: AI is not a sector. It is a change in the conditions of national life. AI could have the social and strategic effects of the printing press, nuclear energy, and the internet - all at the same time, concentrated into less than a decade. Each of these had enormous benefits for humanity - but huge side-effects too. The printing press drove the Reformation and all its positive social consequences in terms of personal liberty, economic development and national sovereignty. But the Reformation prompted the Thirty Years War, devastating the continent. Seeing the upheaval that printing threatened for an ordered society, Europe’s neighbour the Ottoman Empire took a different approach: they banned it. That didn’t work out well for the Ottomans. They ultimately lost the race of civilisation, and declined and declined till they were finally snuffed out in 1918. Our challenge is to avoid the fate of the Ottomans, but to avoid the Thirty Years War as well. Because our Thirty Years War may be a lot shorter, and a lot more terminal. AI could transform human health and deliver the productivity our economy has lacked for 20 years - with huge benefits for us all. To realise this we need more AI, not less - but AI managed well. The only way out is through. At the moment we are at the mercy of the USA. It could be worse - China would be worse - but we urgently need to set a course for soveriegnty. That doesn't mean autarky; it just means more capability and more opportunity to shape our own future. Reform UK offers the 'patriotic compact' to AI entrepreneurs: we will help you build the data centres and connect to the grid; for your part you must keep your firms here, pay your taxes here, create jobs here. Jobs are crucial - Britain must stop destroying work and start helping businesses grow; and use tax and welfare to help people adapt to the jobs of the new economy. And most of all - we must make AI serve humanity not subjugate it. The Pope is right: AI does 'not know from within what love, work, friendship or responsibility mean.' We must focus ourselves, and focus AI, on nurturing human relationships - or will be, like the builders of Babel, ‘lords of towers destined for ruin’. Fundamentally Britain cannot slow down AI. What Britain, and countries like it, can do is develop AI at the pace of innovation but in the direction of human flourishing.
English
Gabe retweetledi
Sorry but this is sophisticated but ultimately hollow cope, and it's worse if coming with fresh OAI and GDM affiliations.
So, no, @alexolegimas, Ricardo's essay is about a different point than @zetalyrae. 'No-One Escapes the Permanent Underclass' mostly argues that when you stop being useful for production or capital allocation or violence and lose these as a source of leverage - and all you have is capital ownership - you are likely to be expropriated. If you've you read both essays, what is your actual argument about the political economy of the situation?
And, no, @deanwball the conclusions of the essay are not in the stated premises, and the premises are not outrageous. There are many who believe that even if "AI can do all cognitive and physical work, at human level or better, and cheaper than humans." and also believe "most humans will be disempowered" but they can escape the shared fate of humans by accumulating capital, working at AGI labs, or similar. The essay makes an argument against that. (I don't think the attack against you was a good or necessary part of the essay)
I'm an optimist in the sense that I believe the political economy problems are solvable and it is possible to escape @zetalyrae's conclusion, but we have better chance if people seriously engage with the problem, and this is the opposite.
Alex Imas@alexolegimas
New rule: Anyone who is about to write a "all workers will be replaced by machines" essay needs to first read On Machinery by David Ricardo (yes, that David Ricardo), because he probably wrote your essay...in 1817.
English
Gabe retweetledi
I wrote about how accumulating capital won't save you from being disempowered by superintelligent AI.
English
Gabe retweetledi
We Need an International Treaty to Ban Superintelligence open.substack.com/pub/persuasion…
English
Gabe retweetledi
The recent Mythos/NSA warning shot is looking more and more like how I expected a "warning shot" to look like.
An absolutely insane thing happens, and then the FUD machine kicks into action and adds infinity gazillion """context""" until no one knows what is true anymore and everyone believes what is most convenient.
English
Gabe retweetledi
The problem we face with AI today is not a technical problem, it is a political problem.
Of who gets to decide. What level of risk the public is exposed to, what future we build towards or avert.
It's so heartening and important to see this conversation starting to happen in the world outside the very insular tech futurist bubble. We need to have these conversations, everywhere, and this piece by @andreamiotti hosted by Francis Fukuyama I hope is a very useful step in that direction!
Francis Fukuyama@FukuyamaFrancis
We Need an International Treaty to Ban Superintelligence open.substack.com/pub/persuasion…
English
Gabe retweetledi
Thesis is that the "escaping the permanent underclass" by accumulating capital is, if anything, optimistic. That the risk isn't "needing UBI" but humans stop being necessary to power bc once power doesn’t need humans anymore to express itself, no amount of capital can save you.
Andrea Miotti@andreamiotti
@zetalyrae Yup, and it doesn't stop there! A good post on this from @Gabe_cc cognition.cafe/p/the-realpoli…
English
Gabe retweetledi
Really good post. It seems you can reduce this to just two premises:
1. AI reaches human-level intelligence.
2. Darwinian competition between states continues.
From this you can derive human disempowerment and extinction.
Andrea Miotti@andreamiotti
@zetalyrae Yup, and it doesn't stop there! A good post on this from @Gabe_cc cognition.cafe/p/the-realpoli…
English
Gabe retweetledi
@zetalyrae Yup, and it doesn't stop there!
A good post on this from @Gabe_cc
cognition.cafe/p/the-realpoli…
English
Gabe retweetledi
Great thread. Many, MANY cases of abusive behavior are not cases of conscious manipulation, and trying to operate from a false dichotomy of 'are they intentionally bad or are they not responsible for their actions' is a recipe for staying stuck in a bad dynamic.
A perspective that's helped me leave harmful dynamics in the past is some version of: 'this is a good person who loves me when they are regulated, but who can't really access empathy or love when not regulated, and they are not-regulated enough that I need to distance myself from them for my own protection.'
It's harder to accept, in some ways, than the all-good/all-bad dichotomy, because the part of us that doesn't want to have to distance ourselves from someone we love can use the clues of the times when they are reasonable and kind as evidence that they are actually in the all-good category and that means we don't have to have the boundaries we wish we didn't have to have! But it's rarely that simple in either direction.
helen@nuanceexists
if i may add to the discourse: one of the things that made it difficult for me to see I was in a bad situation was the idea of a "master manipulator." i've come to think intent is a distraction from the question of harm
English
Gabe retweetledi
NEW: Baroness Miller of Chilthorne Domer, Co-President of Parliamentarians for Nuclear Non-proliferation and Disarmament, has just backed our campaign for binding regulation on the most powerful AIs!
100+ UK parliamentarians recognise the risk of human extinction posed by AI.
ControlAI@ControlAI
We have a new supporter! Lord Evans of Rainow (@GrahamEvans) has joined our campaign for binding regulation on the most powerful AIs! 100+ UK parliamentarians recognise the risk of human extinction posed by superintelligent AI, the largest such coalition anywhere in the world.
English