19,939 karma · joined June 19, 2013
The proof of the pudding.
This (the calculation) is addressed by the author.
And it wasn't that long ago.
I use "frontier" AI models daily at day_job. I can confidently say that anyone who is satisfied with the output of LLM code (enough to commit it straight off) is just an absolutely shit programmer. Sorry but I don't have any other way to put it.
The code is (with rare exceptions) atrocious on every level. It is only not atrocious if you take multiple iterations of "review and correct".
>If we find no flaws in the design whatsoever, will we be less careful about the second one? The third one? What about the ten thousandth one?
Like the saying goes, if my grandmother had wheels she would have been a bike. LLMs can't even produce quality maintainable code for a trivial web service or whatever. Why are we planning for what we will do when they can "design" 10,000 nuclear power plants without any flaw?
But that's not even what happened! They told it to do X and it did X! I swear to god I don't understand the discourse around this.
This sentence alone tells me that it's not a bike lane. A painted line is not a bike lane.
Walter I'm seriously fighting the urge to write a wall of text in response. Why the knee-jerk reaction and the hyperboles? It's not interesting conversation...
I can text my wife when I leave home telling her to meet in the park. I can call for help in an emergency. I can talk to my family, back in my home country, whenever I want and wherever I am. My grandparents emigrated when "seeing your family" meant a month's journey by sea, so I'm quite sensitive to this. It's a miracle we take for granted.
Instant messaging and calling is extremely valuable. Algorithmic social media is extremely noxious. Unfortunately, they both live on the same device.
That's the thing, it very much does NOT show us that. What happened was mathematicians at openAI learned of an imminent development on this problem, and the insight that it entailed, then they were able to prompt a system in the correct direction and spend 20 million dollars to write down the final steps.
Which is rather precisely the point that the article is making!
> If you knew AI could be affordable enough in 3-5 years so that even the most underfunded researchers could use it to solve cancer
As the saying goes, if my grandmother had wheels she would have been a truck.
Wait wait wait. The research mathematicians brought this AI dystopia on us. Not the techbros, not Silicon Valley, not the global financial system and its tentacles, not surveillance capitalism... No, it was the postdocs working on ultrafilters for C(2,3) epimorphisms on symplectic manifolds. That's certainly a take.
Only in a completely stupid, nonsensically A/B tested notion of "prefer". I'll refer you to this ~~excellent piece of satirical writing~~ actual post from Google design team: https://design.google/library/expressive-material-design-goo...