Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

ChatGPT: The Slickest Con Artist of All Time

tedgioia.substack.com

Post reply on HN
Print view View original

61–70 of 444 posts

Re: ChatGPT: The Slickest Con Artist of All Time

#61

Post by endisneigh »

jchw wrote: post #54
endisneigh wrote: post #48

Earlier quoted context omitted.

> That goes for pretty much any resource; there's a lot of botched or sub-optimal explanations of complex things out on the internet. But being able to have a conversation with someone who has a decent conceptual grasp on something is better than a perfect encyclopedic article alone, and ChatGPT is able to roughly do that for a lot of subjects. people keep saying this, but it isn't true. if you ignore sites like this…

Clearly you are of the opinion that ChatGPT is useless, there are superior resources already available for any topic, and it's all just hype. Well, then we don't need to argue this since the problem will elegantly solve itself if that's true. I disagree though. For a lot of things it feels like I can get much better answers than Google, especially when it comes to somewhat conceptual questions. (Also, I don't use new…

> Clearly you are of the opinion that ChatGPT is useless, there are superior resources already available for any topic, and it's all just hype. Well, then we don't need to argue this since the problem will elegantly solve itself if that's true.

I disagree though. For a lot of things it feels like I can get much better answers than Google, especially when it comes to somewhat conceptual questions.

(Also, I don't use news aggregator comments to learn things unless it's the only possible source. But if you think every blog post or YouTube video that got things wrong has a detailed errata, you'd be sorely wrong. It's so uncommon on YouTube that the practice is usually commended when noticed.)

---

I never said ChatGPT is useless, lol. it's truly amazing how people can be so bad at reading comprehension yet praise AI bots in the same post.

Re: ChatGPT: The Slickest Con Artist of All Time

#62

Post by FeepingCreature »

zabzonk wrote: post #42
fnordpiglet wrote: post #17

I had a detailed conversion with chatGPT about how to gracefully handle terminating conditions of a rust program. It summarized cogently to register at_exit()’s for each thread, panic handlers, and register signal handlers. It advised and explain in detail on my query about the thread handling for each of these variants, gave really helpful advice on collecting join handles in a closure on the main thread and waiting…

but isn't it getting its information from all those "partially right contributors"? if so, won't it be only partially right (i.e. sometimes/often wrong) itself?

Theorizing: If the network acquires a pattern of understanding that characterizes the correct response (in other word, if the network "groks" a topic), it will express its knowledge internally as "a correct model (small, efficient)" and "various mistakes that people often make (big, uncompressed)". In this case, the correct pattern should get by far the most reinforcement (because it's a smaller target), and should usually be selected to generate the answer unless the network has a particular reason to expect itself to be playing somebody who's mistaken.

This is also how humans learn, although we can do this process purely internally.

Re: ChatGPT: The Slickest Con Artist of All Time

#63

Post by Julesman »

_huayra_ wrote: post #4

I can't tell what is worse now: the sycophantic ChatGPT hype guys/gals who write articles "it's coming for all of our jerbs!", or articles like this one that deliberately misuse ChatGPT and then say "it's overhyped". They're both missing the point. Yes, ChatGPT can be tricked, confidently give wrong answers, but it is still ludicrously useful. It is basically like having an incredibly smart engineer/scientists/philos…

Well said. Exactly. So much hyperbole! So much click bait. It makes me dizzy.

I'm a front-end dev. I use it every day. I have CoPilot too. These things help me do my job. No question. They also help me learn better than just about any other tool.

The big one for me is that they have made me more confident taking on bigger tasks. The impact is not trivial at all.

Re: ChatGPT: The Slickest Con Artist of All Time

#64

Post by lyu07282 »

fnordpiglet wrote: post #17

I had a detailed conversion with chatGPT about how to gracefully handle terminating conditions of a rust program. It summarized cogently to register at_exit()’s for each thread, panic handlers, and register signal handlers. It advised and explain in detail on my query about the thread handling for each of these variants, gave really helpful advice on collecting join handles in a closure on the main thread and waiting…

I have the same experience with it, it's very useful, especially when you are unsure about what to even look for. I have a few theories

I suspect it's just a bias we have, most people aren't doing or creating anything, they aren't solving problems or learning new things constantly like engineers do, so chatgpt feels useless to them.

Or perhaps the problems we want to talk or learn about are just easier for an ai to answer than the things they want out of it,

or we are just better at asking the right questions to extract the value out of it, similar to how you learn to use Google correctly to find something.

I don't know I'm really confused about how some people think it's so worthless.

Re: ChatGPT: The Slickest Con Artist of All Time

#65

Post by endisneigh »

stevenhuang wrote: post #52
endisneigh wrote: post #10

Earlier quoted context omitted.

> It is basically like having an incredibly smart engineer/scientists/philosopher/etc that can explain things quite well, but for pretty much every field. Does this "person" make mistakes? Can't cite their sources? Yeah this definitely happens (especially the sources thing), but when you're trying to understand something new and complex and you can't get the "gist" of it, ChatGPT does surprisingly well. if you don't…

By having a conversation with it. Ask probing questions. Verify later with wikipedia. For instance I asked it about space time metrics, and I learned about the varying types and in what circumstances each are used. Ask it about their mathematical properties and it gives a succinct explanation contrasting the different approaches. Certainly you can see the value now of conversational style pedagogy. Think of it like l…

seems like a waste of time if you're going to have to verify with Wikipedia anyway. personally I'd prefer to read wikipedia first.

Re: ChatGPT: The Slickest Con Artist of All Time

#66

Post by soperj »

_huayra_ wrote: post #4

I can't tell what is worse now: the sycophantic ChatGPT hype guys/gals who write articles "it's coming for all of our jerbs!", or articles like this one that deliberately misuse ChatGPT and then say "it's overhyped". They're both missing the point. Yes, ChatGPT can be tricked, confidently give wrong answers, but it is still ludicrously useful. It is basically like having an incredibly smart engineer/scientists/philos…

> It is basically like having an incredibly smart engineer/scientists/philosopher/etc that can explain things quite well

This is the issue. It's more like the dumbest guy in the room confidently convincing the everyone else that they know what they're talking about.

Re: ChatGPT: The Slickest Con Artist of All Time

#67

Post by jwie »

fnordpiglet wrote: post #17

I had a detailed conversion with chatGPT about how to gracefully handle terminating conditions of a rust program. It summarized cogently to register at_exit()’s for each thread, panic handlers, and register signal handlers. It advised and explain in detail on my query about the thread handling for each of these variants, gave really helpful advice on collecting join handles in a closure on the main thread and waiting…

It’s only as good as the information it harvested. In a space where less true information is available, the results would be quite poor.

Re: ChatGPT: The Slickest Con Artist of All Time

#68

Post by gunshai »

renewiltord wrote: post #32

All these articles really sound like “I used an apple to hammer in a screw and it sucked. This has to be the worst plant-based object ever made”. It’s a common junior engineer approach. “I broke our database by running DROP TABLE cows in the console”. Yeah, dude, that’s possible. Just don’t do that. The point of tools isn’t to use them like Homer Simpson. But you know what, it doesn’t matter. Stay behind. Everyone el…

What's interesting is that chat models like this can really act like a mirror to one self (knowledge, reasoning,..ect) not perfect but the reflection is there, and articles like these clearly miss that point and their lack of self awareness is so apparent.

They are gleefully delighted when they pick some anecdote ad absurdum to say something (to the uncharitable effect of)

"You all are so stupid... I AM SO SMART ... Look how smart I am for showing you how dumb you are..."

Re: ChatGPT: The Slickest Con Artist of All Time

#69

Post by whartung »

After having played it ChatGPT for a bit, mostly asking computer questions, I've had mixed results. Some are amazing, others are gibberish.

But what struck me the other day is a couple of quotes from, of all things, Galaxy Quest which seem particularly apt.

  "May I remind you that this man is wearing a costume, not a uniform."
and

  "You know, with all that makeup and stuff, I actually thought you were SMART for a second."
As amazing as it is, as progressive as it is, it's still a magic trick.

Re: ChatGPT: The Slickest Con Artist of All Time

#70

Post by todd3834 »

Don’t be afraid of ChatGPT but don’t underestimate what it and others like it will be capable of as it is iterated on. You found one category of prompt that needs some iteration. Good job, if the team wasn’t aware of this already, hopefully you helped point it out.

It’s not that the technology isn’t capable of what you’re asking, it just needs better training for this class of question.

There are other things like generating and translating code that it excels on. I imagine that would be much harder. But we have great data to train for that and the engineers know enough to dogfood that properly.

Post reply on HN