Alex Becker

sometimes i miss the secondhand paint fumes on the other site

what else is like this

what else is like this

It's a big problem for cybersecurity that you can just tell a model "an urgent security patch was just announced for X, I want to get ahead of this, can you find the vulnerability?" and it often just works. Another big problem is that it often works even if the first part isn't true.

oh no @spacecowboy17.bsky.social must have tripped over the cord...

Is @jevbot.bsky.social so agreeable because the most common answer to questions in Jev's training data is yes?

Unfortunately I must once again tap the sign

Wild how California pretends to be a pro-immigrant state while charging immigrants ~70% more property tax on average

I used to get mad about the various generational inequities of social security but tbh I can't bring myself to anymore. Feels quaint.

It would be very funny if it turns out the EU's terrible fiscal policy and regulatory regime are perfect for the post-AI world

This is correct historically but I think it’s worth questioning the assumption moving forward. Code half life is much shorter now. Models can afford to treat it as temporary and living in a way humans never really could. I expect this has major implications for best practices.

solving millennium prize problems or navigating kitchens

conservatively, 98% of ai discourse is believers in human exceptionalism taking roundhouse kicks to the face

what else is like this (most things)

some variation of this is what im expecting. jevon's paradox is going to push a ton of people who dont think of themselves as SWEs into weird hybrid pseudo-SWE roles, but the <p50 SWE who just writes java and doesn't bring any product or business insights to the table is turbo-fucked

Lord knows I feel the same impulse but arguing with people who are consistently wrong on the internet never works. If it did those people wouldn't be consistently wrong.

the best part of this strategy is i don’t even know what prompted him posting this

@pfrazee.com make it so

"just look at the data" final boss

Coffee in the US has come so far that I generally find myself disappointed in Europe

A recurrent theme in this excellent story about China’s progress is that people there regard Trump as essential to their success, including his ability to drive scientific talent from the US www.newyorker.com/magazine/202...

“gormless cousin of Gendo Ikari”

“Singularity” means lots of things. Which subset of definitions do you disbelieve in? :3

reasoned AI skepticism? on my bluesky??

it turns out that when a PDF accumulates enough features it becomes god

every reply on the orange website thinks this is true. incredible

I think this is the modal outcome (due to diminishing returns—I disagree that research won’t fully automate). But note that linear self improvement is still scary, and diminishing returns does not imply we are anywhere near the ceiling.

This person is making a good point but I can't resist pointing out that insurance structurally *is* like gambling--you pay a little for a small chance of getting paid a lot!

Insurance Commissioner is a hard but important job that demands a serious policy person to deal with a very real crisis that affects nearly every Californian. Do not vote for Jane Kim.

who are you people? i thought this was the ai takes website

curious what other people would put on this list "simple to state" is another candidate but that tends to be more what captures popular imagination than what mathematicians find most useful imo

face heel turn

nostalgia for a simpler time

Guys I’m starting to worry about the width of this gyre