Oh this entire post was also AI generated, I don’t know if a human was even involved beyond having an idea that’s not actually worth pursuing

Yeah sure, you found the one notorious TypeScript feature that actually emits code, but a) this feature is recommended against and not used much to my knowledge and, more importantly, b) you cannot tell me that you genuinely believe the use of TypeScript enums – which generate extra function calls for a very limited number of operations – will 5x the energy consumption of the entire program.

Only if you choose a lower language level as the target. Given these results I suspect the researchers had it output JS for something like ES5, meaning a bunch of polyfills for old browsers that they didn’t include in the JS-native implementation…

I’m an empirical researcher in software engineering and all of the points you’re making are being supported by recent papers on SE and/or education. We are also seeing a strong shift in behavior of our students and a lack of ability to explain or justify their “own” work

more like MCETAPB (most cops enable, tolerate and protect bastards)

At least on a Mac keyboard, the en dash is also alt+hyphen and the em dash is shift+alt+hyphen.

Even worse, the pilots and the airlines didn’t even know the sensor or associated software control existed and could do that.

let’s see if we can find supporting information on this answer elsewhere or, maybe ask the same question a different way to see if the new answer(s) seem to line up

Yeah, that’s probably the best way to go about it, but still requires some foundational knowledge on your part. For example, in a recent study I worked on we found that programming students struggle hard when the LLM output is wrong and they don’t know enough to understand why. They then tend to trust the LLM anyways and end up prompting variations of the same thing over and over again to no avail. Other studies similarly found that while good students can work faster with AI, many others are actually worse off due to being misled.

I still see them largely as black boxes

The crazy part is that they are, even for the researchers that came up with them. Sure we can understand how the data flows from input to output, but realistically not a single person in the world could look at all of the weights in an LLM and tell you what it has learned. Basically everything we know about their capabilities on tasks is based on just trying it out and seeing how well it works. Hell, even “prompt engineers” are making a lot of their decisions based on vibes only.

I don’t know if it’s just my age/experience or some kind of innate “horse sense” But I tend to do alright with detecting shit responses, whether they be human trolls or an LLM that is lying through its virtual teeth

I’m not sure how you would do that if you are asking about something you don’t have expertise in yet, as it takes the exact same authoritative tone no matter whether the information is real.

Perhaps with a reasonable prompt an LLM can be more honest about when it’s hallucinating?

So far, research suggests this is not possible (unsurprisingly, given the nature of LLMs). Introspective outputs, such as certainty or justifications for decisions, do not map closely to the LLM’s actual internal state.

kind’ve

Ok not to be nitpicky but this is the first time I’ve ever seen the opposite (complementary?) mistake to “could of”. That’s actually kinda fun :D

Oh shoot my bad haha

… why did you have ChatGPT write this? Clearly you have your own thoughts on this no need to ask a machine lol

Well, I’m generally very anti-LLM but as a library author in Java it has been very helpful to create lots of similar overloads/methods for different types and filling in the corresponding documentation comments. I’ve already done all the thinking and I just need to check that the overload makes the right call or does the same thing that the other ones do – in that particular case, it’s faster. But if I myself don’t know yet how I’m going to do something, I would never trust an AI to tell me.

(structuring inheritance) before the Jesus Club took over

and then it took humanity another 2000 years to move away from inheritance in favor of composition. you’d think someone would’ve realized sooner that it’s not always the right abstraction…

Language designer for a widely used programming language. Basically I want to be Brian Goetz

I have a compsci background and I’ve been following language models since the days of the original GPT and BERT. Still, the weird and distinct behavior of LLMs hasn’t really clicked for me until recently when I really thought about what “model” meant, as you described. It’s simulating what a conversation with another person might look like structurally, and it can do so with impressive detail. But there is no depth to it, so logic and fact-checking are completely foreign concepts in this realm.

When looking at it this way, it also suddenly becomes very clear why people frustratedly telling LLMs things such as “that didn’t work, fix it” is so unproductive and meaningless: what would follow that kind of prompt in a human-to-human conversation? Structurally, an answer that looks very similar! Therefore the LLM will once more produce a structurally similar answer, but there is literally no reason why it would be any more “correct” than the prior output.

Ah, thanks so much for reaching out again! Downloading now :D

What do you think the point of this post is, then? Comedic hyperbole only works if there is still some truth to it

is everyone here a complete beginner? how do so many people relate to this? as soon as you need to do anything halfway interesting the thing just confidently spews nonsense.

I’ve literally done my own study on this with CS students and found a similar result. Students who reported using AI regularly couldn’t recognize when it wasn’t giving them any useful output