LLMs will miss a lot of edge cases. This is critical for a library that parses PDFs. Ask it to review the code for issues and by the 20th time you ask, it still will find issues. How's that faster (or safer)?
I use AI heavily but it's not this omnipresent deity that knows everything at all times.
It's high-risk in that it's proven to not work anymore. Scale isn't a differentiator or won't be in the short term.
Meanwhile, everyone else is working on making these models more useful through a ton of ecosystem around them.
I don't see how the frontier labs will have any control over that to deliver on what investors expect. As mentioned in another post here, if they could, they would already be spawning hundreds of companies to diversify their reach but they aren't, because this era of AI is awesome but still doesn't allow a few individuals to replace millions of others that easily (or efficiently).
I for one flag everything I see him posting because his extreme positions hurts our country. It is crazy HN doesn't rate-limit how much he can post and people posting technical stuff can post like 4 comments then get blocked.
Exactly. And that will be the lasting thing here wrt maths. These things have value, just not commercial value, and that is why long term, it won’t be the interest of private companies.
A huge portion of these trillion dollar businesses will be government and higher education contracts. Private companies already sell millions/billions of equipment and software to education.
Without novel proofs. If AI can churn out proofs, then anybody's work can get "scooped." I've seen this happen to students in other fields, where it has reached the point that students refrain from giving conference talks in order to avoid revealing their research topic.
My thesis project, 30+ years ago, was protected by a unique assemblage of expensive equipment, some of it purpose-built. The people who wanted my result were perfectly happy to wait for me to finish it. I even shared my shop drawings, circuit diagrams, and source code with them as I progressed.
I think at the very least students need to know what to expect. Maybe it will settle down and we'll be OK.
Speaking of github in the age of AI, I heard a rumor from another partner of their parent company that fixing those things should be easy with AI, that the AI that they have now is so terrifyingly good that people shouldn't be allowed to use some of it without a license, and that the stuff they're holding back is far better than even that.
Seeing as they have infinity tokens to spend on godlike secret AI, isn't it weird that those same bugs are still sitting on the bugtracker? Shouldn't Microsoft just pull up the secret OpenAI panel and prompt "fix all github bugs, triage out primitive public-level AI slop first"?
the cognitive dissonance between “software engineering is a solved problem” and the state of software everywhere (still pretty crappy, maybe even worse than a couple of years ago) is staggering
reply