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

AI Is Smoke and Mirrors

bloodinthemachine.com

Post reply on HN
Print view View original

71–80 of 83 posts

Re: AI Is Smoke and Mirrors

#71

Post by hermitcrab »

whimsicalism wrote: post #38
hermitcrab wrote: post #32

Earlier quoted context omitted.

I'm 58. It is hard not to be cynical when you've seen as many hype cycles as I have. Especially when you suspect the person hyping LLMs is the same person that was hyping blockchain and NFTs.

In my personal experience, cynicism seems more like a fixed personality trait people have than one that is rationally borne out of experience. If you had been holding a representative basket of tech stocks right before the dot-com bust (a classic "hype cycle"), you would be beating the S&P 500 today if you just held on to them, even after most of those companies went bankrupt. Even on crypto, it is too early to call.…

I've definitely got more cynical as I've got older. I'm sure I'm not alone in that.

>you bridge so easily from crypto to LLM

They are both things that have been massively hyped. I'm not saying they are the same in other ways.

Re: AI Is Smoke and Mirrors

#72

Post by ryandvm »

Look, when I can dump a function into GPT4 and tell it to "refactor the following function with an emphasis on maintainability and testing" and it does a pretty damn good job of refactoring it, I have a hard time believing that this is "smoke and mirrors".

Re: AI Is Smoke and Mirrors

#73

Post by pennybanks »

yawnxyz wrote: post #10

That's a lot of words from someone who doesn't seem to have tried using these tools yet? The Hype is smoke and mirrors, but there are tons of real use cases where things were MUCH easier than before. Things like text to speech or image recognition previously required a team of 10 ML engineers a couple of years to build now take a weekend. It's smoke and mirrors to the "business world" because none of these translate…

but i think that point of view is important for 99% of people that will use AI. since most of them will be potential consumers and the more you can hype and mislead the better for some looking to take advantage. i mean im seeing it with every new startup or company now and im not even that super knowledgeable but numbers do not look correct at all. everyone lives in the "business world"

Re: AI Is Smoke and Mirrors

#74

Post by Intralexical »

yawnxyz wrote: post #10

That's a lot of words from someone who doesn't seem to have tried using these tools yet? The Hype is smoke and mirrors, but there are tons of real use cases where things were MUCH easier than before. Things like text to speech or image recognition previously required a team of 10 ML engineers a couple of years to build now take a weekend. It's smoke and mirrors to the "business world" because none of these translate…

> Things like text to speech or image recognition previously required a team of 10 ML engineers a couple of years to build now take a weekend.

---

> This might well be the most fraught moment in generative AI’s young lifespan.

> an anti-generative AI cultural backlash is growing

> to describe the generative AI world’s shadier products and promise

> it’s still remarkably nebulous as to what generative AI systems actually can and cannot do

> and the contexts in which generative AI and the magic lantern

> Report after report indicates that generative AI services are underperforming

> that propels the generative AI industrial complex onward

> And why a large percentage of Fortune 500 companies are reportedly using generative AI.

> the product of generative AI’s smoke and mirrors:

> is a threat to the generative AI enterprise at large.

> generative AI desperately needs that mythos.

> Unlike trends of the very recent past, generative AI has real gravitational pull

> Even critics admit there are some good uses for generative AI

> And so, with generative AI, we’re once again witnessing a core problem with entrusting

---

    User: What is "Generative AI"?

    OpenZephyrChat: Generative AI refers to the use of artificial intelligence algorithms that generate new content based on input data or a specific prompt provided by users. These systems can create text, images, music, and even software code using machine learning models trained on vast amounts of data. Some popular examples include GPT-3 for generating human-like text and DALL-E for creating original artwork from user prompts. Generative AI has the potential to revolutionize various industries by automating content creation, enhancing creative processes, and enabling more personalized experiences for users. However, it also raises concerns about copyright infringement, accuracy, and the potential for misuse of these powerful tools.

    User: Would automated machine transcription or image recognition of existing content count as "Generative AI"?

    OpenZephyrChat: No, while automated machine transcription and image recognition can be considered part of artificial intelligence, they do not fall under the category of generative AI. Generative AI refers specifically to systems that generate new content based on input data or a specific prompt provided by users, whereas these other applications are focused on analyzing existing content for various purposes like search engine optimization, accessibility, and content moderation.

Re: AI Is Smoke and Mirrors

#75

Post by Intralexical »

Intralexical wrote: post #74
yawnxyz wrote: post #10

That's a lot of words from someone who doesn't seem to have tried using these tools yet? The Hype is smoke and mirrors, but there are tons of real use cases where things were MUCH easier than before. Things like text to speech or image recognition previously required a team of 10 ML engineers a couple of years to build now take a weekend. It's smoke and mirrors to the "business world" because none of these translate…

> Things like text to speech or image recognition previously required a team of 10 ML engineers a couple of years to build now take a weekend. --- > This might well be the most fraught moment in generative AI’s young lifespan. > an anti-generative AI cultural backlash is growing > to describe the generative AI world’s shadier products and promise > it’s still remarkably nebulous as to what generative AI systems actua…

    User: If somebody responds to an article about the harms of "Generative AI" by purposefully conflating "Generative AI" with systems such as machine transcription and image recognition, how should that be interpreted in regards to their argument being made in good faith or bad faith?
    
    OpenZephyrChat: If someone responds to an article about the harms of "Generative AI" by purposefully conflating it with systems like machine transcription and image recognition, it may indicate a lack of understanding or intentional misrepresentation. In either case, this kind of argument would not be made in good faith as it fails to accurately engage with the specific topic being discussed. Conflation can lead to confusion and misunderstanding about the potential risks and benefits associated with generative AI technology, which is why it's important for people to have a clear understanding of what these systems are capable of before making informed decisions or opinions about them.
I did not actually expect it to be quite so direct. Heh.

Re: AI Is Smoke and Mirrors

#76

Post by Balgair »

mlyle wrote: post #6
whimsicalism wrote: post #4

No matter how much the chattering class wants to portray this as a "fad" and not an inevitability, those of us who can type and read can see the reality for ourselves.

Yah-- what we have now is enough to be enormously disruptive as it is absorbed. The big question is how much better it's going to get. We may be plateauing in capability-- or at least slowing in capability growth-- or we may not be. But it doesn't need to get better for it to eat a lot of peoples' lunch.

Even if we are plateauing in GPT number (4,5, 12, whatever), there is still a lot of meat left on those bones.

For instance: Chemists using the AI backwards. Take some phrase like 'vanadium increases the Young's modulus versus silicon in 1040 steel' and then work the AI backwards from that phrase. As in, assume the AI outputted that phrase and see what inputs were most likely to generate it. There may be some real discoveries just by working an AI backwards iteratively.

Simple things like that are still open and attainable right now, it's just that AI is still so young that we really haven't explored all of what they do yet.

Re: AI Is Smoke and Mirrors

#77

Post by HarHarVeryFunny »

hn_throwaway_99 wrote: post #19
yawnxyz wrote: post #10

That's a lot of words from someone who doesn't seem to have tried using these tools yet? The Hype is smoke and mirrors, but there are tons of real use cases where things were MUCH easier than before. Things like text to speech or image recognition previously required a team of 10 ML engineers a couple of years to build now take a weekend. It's smoke and mirrors to the "business world" because none of these translate…

I'm just kinda sad at how cynical so many of us have become. I mean, I see tons of comments "But it hallucinates!!!!!" Yeah, no shit. At the same time, you're having a perfectly grammatically correct, natural, sensical conversation with a computer . Just a couple years ago that was purely the realm of Star Trek.

True - it is amazing. Shows how fast we get used to things.

If someone created an anti-gravity suit, then in a couple of years we'd just be whining about them - how uncomfortable they are, how inconvenient to be banging into people all the time ...

Re: AI Is Smoke and Mirrors

#78

Post by lannisterstark »

itronitron wrote: post #46
hn_throwaway_99 wrote: post #19

Earlier quoted context omitted.

I'm just kinda sad at how cynical so many of us have become. I mean, I see tons of comments "But it hallucinates!!!!!" Yeah, no shit. At the same time, you're having a perfectly grammatically correct, natural, sensical conversation with a computer . Just a couple years ago that was purely the realm of Star Trek.

But you're not having a conversation with a computer, you are simply typing something in and the computer is producing a response.

Are we actually having a conversation? I am saying something and you're just producing a response to that.

Re: AI Is Smoke and Mirrors

#79

Post by HarHarVeryFunny »

lambdaba wrote: post #21
mlyle wrote: post #6

Earlier quoted context omitted.

Yah-- what we have now is enough to be enormously disruptive as it is absorbed. The big question is how much better it's going to get. We may be plateauing in capability-- or at least slowing in capability growth-- or we may not be. But it doesn't need to get better for it to eat a lot of peoples' lunch.

My opinion shifts every other day it seems, recently, Roon, the "famous" anon X account that works at OpenAI, and presumably has access to some unreleased info, said, in response to "are we plateauing soon", "not even close". I honestly don't know what to think.

OTOH Altman has said that next release will be about same advance on GTP-4 as that was over GPT-3. Dario Amodei has said something roughly equivalent - to expect solid gains this year but no reality bender.

I think the real potential gains come when they start extending the architecture, but as long as it's just scaling up and different training/inference regimes, then it seems it'll be more of the same rather than game changer.

Given how connected the whole SF/AI scene appears to be (alchohol + drugs too?), it's hard to imagine a company the size of OpenAI not leaking. If there had been any amazing discoveries there, I think there'd at least be rumors (and not just "they seem to have something called Q* going on").

Re: AI Is Smoke and Mirrors

#80

Post by hn_throwaway_99 »

hermitcrab wrote: post #32
hn_throwaway_99 wrote: post #19

Earlier quoted context omitted.

I'm just kinda sad at how cynical so many of us have become. I mean, I see tons of comments "But it hallucinates!!!!!" Yeah, no shit. At the same time, you're having a perfectly grammatically correct, natural, sensical conversation with a computer . Just a couple years ago that was purely the realm of Star Trek.

I'm 58. It is hard not to be cynical when you've seen as many hype cycles as I have. Especially when you suspect the person hyping LLMs is the same person that was hyping blockchain and NFTs.

I'm not exactly a spring chicken either, and I 100% agree on the insane, unwarranted hype of blockchain and crypto (plenty of examples in my comment history).

But the huge difference with, for example, blockchain, is that despite all the attempts by the hypesters to explain what revolutionary tech it was, I was always left thinking one of "Umm, OK, but what do I actually do with it" or "How is this better than just a normal database?" or "Actually, that's not possible." Blockchain was nearly always a cool tech looking for a use case beyond cryptocurrencies.

But with LLMs, I never felt that way. I was immediately blown away by how cool ChatGPT was, and how much real value I got out of it. Sure, I think there is plenty of overhyping (and I think it's a little funny how companies are trying to add that little star icon everywhere, even when it's just annoying), but even if I ignore what everyone says about it and just use the tech personally, by myself, I'm still left thinking "Damn this is just magic" many times a day.

Post reply on HN