I'm not a fan of expanding surveillance but the alternative of unmonitored transit seems worse.

Going to shoot up a place? Take your guns on a waymo. It won't complain. Guy wants to take an obviously distressed and co-erced girl to an abandoned lot? No complaints. Person with obvious head injury and slurring speech wants to go home instead of the hospital? Sure. Person has heart attack and passes out in back? We'll just drive your limp body 45 minutes in traffic to your destination.

These are just a few situations any human driver would obviously react to. And a modicum of detection and human review could solve in an autonomous vehicle.

Goodness thanks for taking a terrible article and getting to the point.

That's the pickle isn't it? AI solved it with human help. But it's standing on their shoulders. Neither AI nor the humans got it alone.

Humanity is better having solved this issue. But which humans were credited and benefited is the issue.

I agree (and so does Buckmaster based on his written statement) that we are better having solved this.

But I disagree that which humans were credited is the heart of the issue in this particular controversy. The question is what do you need to bring to the table for a result like this. A pre-release frontier model trained on the open literature and $15 million of inference? Or all that plus a year of the experts finding the path to the solution for the model to run with?

I think it makes a huge difference in terms of what we think the future of mathematical research will be like, and whether we should still encourage students to go into this field, which was the original topic of this thread

The world is pretty pissed at the u.s. and seeking sovereign models and ai. Richest or not if the companies are wise they will try a bit not to piss off the entire rest of the world. At this point many countries are willing to cut off the u.s. even if they take a short or medium term hit.

This is a succinct summary of most the the profession of software engineering. Just replace AI with "programmers".

Anyone remember the gpt-3 era where adding "this is important to my job" and "if this is wrong someone dies" to prompts made a measurable improvement in accuracy and reduced hallucinations?

The answer is genuinely yours

And that changes the game. Again.

Worth sitting with.

a caveat worth mentioning:

Crispr gene editing is an amazing and terrifying technology. We have the capability to edit people in permenant ways which could be passed to all future generations with unknown side effects and consequences.

I believe now we only have a shaky agreement by researchers and countries to "promise" not to do this due to the unforeseen dangers to humanity.

The hardcore ai folks still assemble their instructions in prompts for best speed and control of the underlying model.

I don't consider meta too big to fail. If they go out of business the world can move on with minor disruption and inconvenience.

> If they go out of business the world can move on with

substantial improvement to quality of life. Goals: build a company such that people don't think the world would be better off if you don't exist.