FaceDeer

Have them predict what a reasonable plan would look like. Then they can start working from that.

That sort of thing can be handled by the framework outside of the AI’s literal context window. I did some tinkering with some automated story-writing stuff a while back, just to get some experience with LLM APIs, and even working with an AI that had only a few thousand tokens’ context I was able to get some pretty decent large-scale story structure. The key is to have the AI work in the same way that some human authors do; have them first generate an outline for the story, write up some character biographies, do revisions of those things, and only once a bunch of that stuff is done should it start writing actual prose.

It’s still surprisingly small a price for selling out such a large and powerful country.

Just in case people think this is a literal excerpt from the article (that was my first impression) the actual survey results were:

33% are very likely to trust 32% are somewhat likely to trust 21% are neutral 14% express some level of distrust

Or, you’re in a bubble and are surprised to discover that most people aren’t in it with you.

There’s a lot of outright rejection of the possibilities of AI these days, I think because it’s turning out to be so capable. People are getting frightened of it and so jump to denial as a coping mechanism.

I recalled reading about an LLM that had been developed just a couple of weeks ago for translating source code into intermediate representations (a step along the way to full compilation) and when I went hunting for a reference to refresh my memory I found this article from March about exactly what’s being discussed here - an LLM that translates assembly language into high-level source code. Looks like this one’s just a proof of concept rather than something highly practical, but prove the concept it does.

I wonder if there are research teams out there sitting on more advanced models right now, fretting about how big a bombshell it’ll be when this gets out.

As others have mentioned, it’s possible but very complicated. Decompilers produce code that isn’t very readable for humans.

I am indeed awaiting the big news headlines that will for some reason catch everyone by surprise when a LLM comes along that’s trained to “translate” machine code into a nice easily-comprehensible high-level programming language. It’s going to be a really big development, even though it doesn’t make programs legally “open source” it’ll make it all source available.

Past results are no guarantee of future performance.

Just because it worked out in the end didn’t mean it was a good idea to try.

I’d love to hear about any studies explaining the mechanism of human cognition.

Right now it’s looking pretty neural-net-like to me. That’s kind of where we got the idea for neural nets from in the first place.

The meme would work just the same with the “machine learning” label replaced with “human cognition.”

Do you really leave Windows with default-everything whenever you deploy a Windows machine?

Which has waaaaaay more features than Notepad does.

If only you were allowed to install your own text editor.

So turn those features off. I just checked, there’s a setting for both spellcheck and autocorrect.

Oh thank goodness. Worst fear allayed, now on to the next one.

There’s currently a hurricane incoming from the Gulf of Mexico, it’ll be interesting to compare the damage it causes to the damage that would have been caused by a regular rocket launch.

Ultimately, these rocket operations need to be done somewhere. Due to their nature that “somewhere” has a lot of restrictions on it. It needs to be as far south as practical (to benefit from Earth’s rotation), it needs to be somewhere far from human habitation for safety and sonic reasons, and the path to the west needs to be clear of stuff that could be damaged by falling debris. For political reasons it needs to be in the United States. All these competing criteria generally mean it’s going to have to be on a southern coastline where there’s currently wilderness.

If California passes major restrictions on AI training then I think AI guys would very much want to be anywhere else.

There are already plenty of places to go. Major centers of AI activity include the UK, France, Israel, China and Canada. Many of the top AI companies aren’t headquartered in California even if they’re US-based.

AI’s future in California hangs in the balance.

Again, it’s fundamentally the same thing. You’re just using different tools to perform the same action.

I remember back in the day when software patents were the big boogeyman of the Internet that everyone hated, and the phrase “…with a computer” was treated with great derision. People were taking out huge numbers of patents that were basically the same as things people had been doing since time immemorial but by adding the magical “…with a computer” suffix on it they were treating it like some completely new innovation.

Suddenly we’re on the other side of that?

Anyway, even if you do throw that distinction in you still end up outlawing huge swathes of things that we’ve depended on for years. Search engines as the most obvious example.