Jeff Fecke

@jkfecke.bsky.social

Deleted the stove emoji because we are now touching it.

i've identified the Dockers store where Michael Tracy bought and then subsequently returned these pants after wearing them. i will be pursuing criminal prosecution.

Michael Tracey in a light blue sport coat and tan chinos.A close up of the chinos show the sizing sticker is still on the pants.

I'm looking at this speech now, and that is absolutely not the fucking hero of Greek myth. The hero of Greek myth is almost always a dramatically tragic figure who is undone by their own character flaws. In fact, that's basically every single Greek hero. What the fuck are you talking about man.

Bild

If Republicans do 95% of X, then both sides are the same. If the GOP does 100% of X, it's Dems' fault for not stopping them. If the Dems have no ability to stop them, it's their fault for not being compelling enough. But it's never Rs' fault, and never ever the electorate's

much respect for Professor Tao, but there are so so so many medical therapies that work without anyone really understanding how or why they work one of the great things about science is that we can use empirical knowledge in practice far ahead of when theoretical frameworks catch up to refine it

Presentation slide titled “A thought experiment on alignment and understanding.” It describes a hypothetical AI-generated chemical cocktail that, when injected into a patient, kills all their cancer cells. The solution is confirmed by mathematical models, stage 3 clinical trials, and even “somehow” verified in Lean. The slide asks whether this solution could nevertheless be misaligned with broader goals, such as avoiding harmful long-term side effects. It then asks whether, before receiving the injection, you would want at least one human cancer expert to understand the cure’s mechanism, or a human mathematician to understand the mathematical model used to find the cocktail. “Misaligned” and both instances of “understands” are highlighted in red. The footer reads “Terence Tao” and “Math 2.0.”

Not sure I could come up with a less charitable stereotype of mathematicians than "guy who thinks all drugs are designed from first principles, thinks side effects are found deductively, and who would be hesitant about treating THEIR STAGE THREE CANCER if they learned this wasn't the case."

Hélianthe@helianthe.net · 14h ago

much respect for Professor Tao, but there are so so so many medical therapies that work without anyone really understanding how or why they work one of the great things about science is that we can use empirical knowledge in practice far ahead of when theoretical frameworks catch up to refine it

Presentation slide titled “A thought experiment on alignment and understanding.” It describes a hypothetical AI-generated chemical cocktail that, when injected into a patient, kills all their cancer cells. The solution is confirmed by mathematical models, stage 3 clinical trials, and even “somehow” verified in Lean. The slide asks whether this solution could nevertheless be misaligned with broader goals, such as avoiding harmful long-term side effects. It then asks whether, before receiving the injection, you would want at least one human cancer expert to understand the cure’s mechanism, or a human mathematician to understand the mathematical model used to find the cocktail. “Misaligned” and both instances of “understands” are highlighted in red. The footer reads “Terence Tao” and “Math 2.0.”

oh my God are we doing that "bearing witness to atrocity" bullshit again kid if you got a taste for snuff seek professional help and don't try to sugarcoat it with fake religious sentiment or maybe we should sue your ass like a traumatized content moderator sues a social media company. yes? yes.

Bare minimum, for many popular views about consciousness, AI clearly raises difficult questions. People who pretend otherwise are discrediting themselves, whether they realize it or not - they are revealing that their views are not reasoned conclusions, but merely scaffolding for their intuitions.

Are AIs conscious?? Probablyyyy not, but "probably" isn't "definitely," and it's a very hard question to answer conclusively (especially given we can't conclusively answer things like "are humans conscious"). So maybe exercise a little bit of humility instead of roundly mocking anyone who raises it?

What drives me crazy is how many people think of themselves as open-ended thinkers willing to tackle the hard questions, while simultaneously assuming the hard questions have an obvious answer that conforms with their initial kneejerk assumption. "We don't know" is a possible answer, guys!

Post nicht verfügbar.