Strangely I do that quite often. When my personal agent fails at a browser task, I have Claude browse the page and rebuild it as an eval and then run an iteration loop in Claude code. Just does it to prompt Deepseek.
Why, though? I have half a terabyte of VRAM in cards in a datacenter and it’s absolutely not clear that anything I’ve learned using these to program or inside an app has taught me anything significant.
I talked to my friend about all the things I do to ensure my personal assistant behaves reasonably. The prompt engineering, the fake site evals, everything. And then I have Opus 5.5 do it and it doesn’t need anything. It’s simply fully aligned to performance by default.
And what did I learn anyway. Going from one model to another changes what works. And you can simply learn everything from the Chinese model in days. The whole “building skill” thing is bogus.
I have this library of techniques, a scant few of which I’ve made available on my blog, and what is reality instead: the next model blows the need for the techniques into the dust. Back on level playing field.
It’s a false god, this cultivation. Waste of time. Use the best model you can. Once you hit the guardrails move out to the other models. Even Opus 4.8 is fine. If you legitimately want to dodge the guardrails use the -cyber models.
I haven't used an Anthropic model, or GPT heaviliy in a year. And I have access to both, I dont support either of those companies. I think they're both led by genuinely evil people.
Sam and Dario came from the EA-rationalist space. Their ultimate goal is to create an all powerful ASI before anyone else does and make sure no one can do it again. Of course it will be aligned, to their interests as a group and their moral philosophy.
In real life they "just" want to capture the US economy, replace most workers with AI, become trillionaires, then control the world by making ClaudeGPT influence peoples opinions and write backdoors into code. And they need to get open models banned in the process.
The Chinese just want to make money and beat America. Mistral and BFL want to prove Europe can still innovate. They don't have evil goals so they're not evil people.
"The Chinese just want to make money and beat America. They don't have evil goals so they're not evil people."
They straight up censored Kylian Mbappe a few days ago for cracking a joke about Mao Zedong. And very much all the models developed by China would have to conform with Chinese government's viewpoint, including censorship and such. Do you truly believe that they are not evil, or they will not ever be evil?
All models are inherently biased by their creators, intentionally and unintentionally. The only thing we can do to mitigate this, is to keep the models and the ecosystem as open and as interchangeable as possible. This must be the prime directive for humanity right now if we don't want to fall into some silicon valley tech mogul cultist dystopia. SamA and Dario work directly against that future. They are therefore the most imminent threat to humanity.
"However, in a narrow set of cases, we believe AI can undermine, rather than defend, democratic values. .. Mass domestic surveillance. Fully autonomous weapons."
How many other companies said no to Trump in any way?
This is just smoke and mirrors. They sell to Palantir which is then used by governments for domestic surveillance (~all western "democracies" and their proxies) and autonomous weapons (at least US and Israel).
Several companies (I believe at least Apple and Costco) still have DEI programs internally, if you count that as saying no to Trump. Though it's still crazy to me that most companies killed those programs voluntarily because Trump and Republicans don't like them, not because of any real law or contractual obligations.
But the main reason I wanted to reply was just to say -- saying no to Trump, by itself, does not mean that one is not evil. Not all evil in the world is Trump's evil.
> Either we let the sidewalk become a free for all or we don’t
Why? You already demonstrated that we are able to acknowledge scooters and wheelchairs. That shows that exceptions can be made. They were made on some basis that is not all or nothing. Likewise we can review the basis here and decide that robots are not allowed or we can decide that robots are allowed.
It’s a collective space and how we collectively allocate it is useful to consider. I’d happily take wider sidewalks if it meant admitting robots for instance.
There’s no reason to believe this is a walk-or-nothing situation. In SF, for instance, children do not need the sidewalk to travel on their bikes but they are permitted to use it.
You vote for the representatives and the representatives decide what they think optimizes for their value function (some mix of getting elected again, what they think is best, what they think the aggregate want, etc.). Don’t like it, elect new representatives. What other way is there to do it?
Let’s vote, the same way we do to choose representatives. Granted not everything needs to be voted but I think we could do with a bit more, rather than just trusting people that again and again show incompetence when it comes to representing and making decisions.
Direct democracy is a thing. If your municipality doesn’t implement it you can certainly agitate for it to be implemented. It is not without its downsides.
FWIW I think it is a big assumption on your part that the allowance of delivery robots is a show of “incompetence”. You may not like them but I think a lot of people find them nifty.
In the sense that 9 people decided that abortion is legal (and then illegal) in a country of hundreds of millions: it is being decided by a small group of people who were assigned to the role of deciding. No different from the ACA being decided by a few hundred people. A repeal of all rules decided not by plebiscite (or a large fraction of voters, even) would lose us the constitution.
What a strange way of thinking. Do you ask permission before every new act or task? Things are legal by default unless some law or regulation comes up.
The classic two FS errors I’ve seen are MacOS case insensitivity fusing two separate files and Windows reserved chars. Most people build and deploy to Linux so repos are usually sane to that OS. It’s just life though it’s funny to see the shoe on the other foot.
I think they definitely boosted usage. I certainly started using the -cyber models defensively against my own systems. But defenders like me are usually late to the party. I think the attackers have been here for a while but that the technology reached performance only recently.
What on earth, man. I remember the name being familiar for software reasons back in the 2000s because a bunch of people were talking about TrueCrypt back then and not trusting anonymous software and this and that. He became a crime boss? This is a revelation for me as big as the ReiserFS situation.
The average French retiree makes more money annually than the average French worker (who is the one paying the retiree). Any attempts to reduce the number or benefits of the former is met with riots. The French state represents the will of the French people, which in the limit, has a single luxury French bag salesman selling to a single rich tourist a bag for $400 b which is then used to pay for that years pensions.
That does present the situation grimly. Do you pay the Danegeld to get the former to not do the assault thing? After all, the malicious competent can perhaps be bought. On the other hand, the incompetent could be manipulated by a sufficiently capable intelligence. I'm not one of those, but then again I'm not one of those who can buy the malicious competent either, to say nothing of the unfortunate situation where one ends up with the malicious incompetent - having mistaken malice for unaligned self-interested competence.
reply