a57721

> suggestions for ingredient substitutions (if they’re good, which is the crucial part!) would be a really useful feature for me

Can't they implement this feature without any gen AI? OK, the answer is that they don't want to pay someone for making a list of substitutions, but at least they can generate it and make a human cook revise the results. There is no need for calling gen AI APIs when the app is running.

People are shoving gen AI everywhere out of the hype, including tasks that have better, more reliable solutions with traditional approaches.

While there are professors who use students to ghostwrite books or force them to submit papers in co-autorship, that's definitely not common, and practically impossible outside of graduate school. Your average professor/TA is underpaid and struggles to teach basic stuff to the undergrads, that is totally unrelated to their scientific interests and "hobbies".

That's what I am doing for my classes, just provide useful feedback and mark the homework as submitted. The university wants points, but for me any submission is 100% points, even if there are errors and omissions.

However, now there are students that just copy the output of LLMs without even caring to understand terminology and notation. Should that count as a completed assignment? In some egregious and too obvious cases I ask them simple questions about what they wrote and don't give them marks/points if they have no clue.

Using LLMs as assistants for homework is one thing, and there's no way to prevent it. But students who just copy characters from the screen without caring to understand what is going on are on another level. Above all, they are wasting their time.

Regarding giving or not giving scores, that's a complicated issue. Scores provide a "negative incentive" for learning. A positive incentive would be some pat on the back for doing the assignments. However, our brains are not wired by the evolution to do things only for positive incentives when there are no grades/deadlines/etc.

In my experience, being lenient works great with graduate students, and undergrads won't do and won't understand anything in the course if there are no grades. I learned that the hard way.

It only remains to explain this to the management... That does not work with code, engages with AI only in the form of chatbots, but demands everything to be "AI native" and won't be there to deal with the consequences.

When there are 100 explanations of some topic, maybe it's not so trivial, after all, even when the bare axioms seem trivial.

I majored in mathematics and I learned category theory before touching any serious Haskell, and I still found most of the "monad tutorials" kind of opaque and handwavy, and I think there are inherently non-trivial details unrelated to the notion of functor etc.

I see increasing amounts of code merged to the codebase that is sometimes wrong, sometimes too brittle, sometimes fixes the symptom without addressing the bug. Two years ago the same code would be criticized harshly and reworked before being accepted.

Also, now people who used to manage the development but not write any code also feel their superpowers to merge LLM slop in humongous amounts. It feels like a ticking time bomb, but I am afraid of calling it out and rain on their parade of 10x productivity (or often AI psychosis). Leaving the team doesn't feel like a good decision because what I hear from many colleagues is much worse. Ah what a time to be alive...

Mathematicians care a lot about the exposition of their ideas, invest a lot of time in giving talks, writing, and don't publish too frequently, compared to other sciences.

They normally don't feel like they are in some kind of race to publish the results ASAP and claim priority. Cases like that are very rare (but they get media coverage because they are so unusual).

OpenAI did a publicity stunt, their motivation is not to make a good contribution to the field, which has very different standards and culture, compared to the AI labs.

> journals are not the arbiter of truth and getting published in them is something only academics have an incentive to do

Good, I just wanted to point out that peer review isn't primarily an arbitrage of truth, it is also to make sure the exposition is nice to read. When you get a reviewer who actually cares, you receive lots of feedback that isn't related to the correctness of Lemma 3.14.15 and stuff like that.

What kind of prestige? Peer review is anonymous unpaid work.

A good review does not merely check the correctness of logical arguments, it gives suggestions for the exposition, citing the correct references, putting everything in the right context, etc.

> the most non-trivial thing that technology can accomplish is a pseudo-form of automation for mathematics and coding

It's irrelevant that this is a pseudo-automation, what's important is that techbros can convince people who make the decisions and concentrate wealth that this is a full automation. So expect reverse centaurs in increasingly more professions in the future.

In a sense, because now journals will be flooded by LLM slop.

GraphQL uses POST for queries.

I heard the following nice metaphor from a professor: learning mathematics is like learning to ride a bike, at some point you just need to get on the bike, you don't just study some "theory of bicycles".

I don't think math is that different from other subjects, but it has this advantage that it is a whole world that you can carry with you and experiment, you don't need that many tools for your exploration.

I recognize many applications of gen AI, but copyright laundering seems the least interesting for me.