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

What it feels like to work in AI right now

robotic.substack.com

Post reply on HN
Print view View original

341–350 of 768 posts

Re: What it feels like to work in AI right now

#341

Post by Irishsteve »

cgearhart wrote: post #152

From my perspective it’s just _confusing_ to work in AI right now. We have some massive models that are doing some really neat stuff, and apparently hundreds of millions of people are using them—but I keep wondering: to do _what_, exactly? I’m not asking what the models can do, I’m asking what people want the models to do every day, all the time. I’ve been shown some neat pictures people made that they thought were c…

similarly what’s the point of stackoverflow or wiki when you can just get the answer you want from a book in the library.

What I value about these lls is that they are essentially a condensed version of the internet (despite being stupidily large for normal hardware currently)

Usually if I’m building a recommendation or search algorithm I have to use the data from the company I’m working with. This makes it possible for me to encode the entire internet into a model that might be running on a product that only has 100 users.

Re: What it feels like to work in AI right now

#342

Post by AussieWog93 »

sashank_1509 wrote: post #194

This sub and blog post are all talking about ChatGPT which is no doubt amazing and far ahead of the curve. However I would also point to Metas new vision model: SAM (Segment Anything): it is so far beyond any other vision model, I actually believe vision will be solved in a few years now. People don’t realize that there was an industry of publishing paper with incremental improvements in small datasets in CVPR that h…

That sounds amazing.

I have a robot vacuum cleaner with "AI" computer vision, and you can just tell that it doesn't see the world in the way a living being does. It'll detect and avoid shoes, cables and dog turds, but is completely blind to anything not in its specific library of objects (we have a two year old, it chokes all the time on things like ribbons and toys).

Being able to recognise objects that cause it to choke, even if it doesn't what they're called in English, is souch more significant to robotics than being able to label things.

Re: What it feels like to work in AI right now

#343

Post by AussieWog93 »

spaceman_2020 wrote: post #296
icelancer wrote: post #181

Earlier quoted context omitted.

The level of denial people are willing to sink into regarding how good GPT-4 is compared to everything else is truly crazy. Not a single other project is an order of magnitude close to the quantitative and qualitative (actual experiential results, not just benchmarks) results that GPT-4 brings.

I feel that there’s significant insecurity among a lot of coders about GPT-4. A lot of them are ignoring the pace of improvement and highlighting the few off chances where it gets things wrong.

I've not used GPT-4, so it could be different, but regular old GPT-3.5 gets a _lot_ of things wrong.

Re: What it feels like to work in AI right now

#344

Post by mayagerard0 »

achenatx wrote: post #242
cgearhart wrote: post #152

From my perspective it’s just _confusing_ to work in AI right now. We have some massive models that are doing some really neat stuff, and apparently hundreds of millions of people are using them—but I keep wondering: to do _what_, exactly? I’m not asking what the models can do, I’m asking what people want the models to do every day, all the time. I’ve been shown some neat pictures people made that they thought were c…

its good for so many things: Need to write a speech as a best man Write a eulogy for a deceased family member Recently used it to speed up the time to fix an oauth problem in code I didnt write, knowing nothing about oauth. Replaces stackexchange with much more tailored answers. can take hundreds of pages of text and distill it down to an executive summary of any length Daughter uses it to explain how to solve algebr…

Unrelated to the general question of usefulness, but I'd be careful using it as a mathematics tutor. Maybe the algebra in this case is sufficiently basic as to be trivial for chatgpt, but I've found that quite often it gets confused by relatively simple mathematics in very baffling ways (One particular example I found amusing is it trying to use the pigeonhole principle with exactly N pigeons and N pigeonholes). There's also the problem that, being a complete pushover, it's hard (specially for a learner) to pinpoint whether the error is in the source or in the understanding of the reader. A teacher may, upon the request of a student to clarify what it seems like a mistake, say "No, this may seem wrong but it's correct for so and so reasons", but chatgpt usually goes "You're right, that was a mistake" and gets itself into messier and messier reasoning.

Re: What it feels like to work in AI right now

#345

Post by copperx »

esquire_900 wrote: post #228
nwienert wrote: post #204

Earlier quoted context omitted.

I mean Google revolutionized search. Apple revolutionized personal computing. OpenAI revolutionized… rewriting things with slightly different wording? I’ve seen so many breathless people posting “this would have taken me so long to search” and then I type 3 keywords from their massive prompt they crafted and find it instantly on Google. We’re talking 1000x or more faster. I feel like the same is happening in your com…

The difference between your search query and theirs is clearly the level of expertise. Chatgpt has a great use case when you get started on a new subject; even with a very cluncky description it can point you into the core concepts of any field. Instead of reading 10 papers with are somewhat related but not what you are looking for, you can spend 3 minutes writing clumsy prompts and that's about it :)

Exactly. Does anybody remember in the early internet days people laughing at their parents for googling things like "please help with my back pain my doctor sucks".

Well, who's laughing now?

Re: What it feels like to work in AI right now

#346

Post by ruskyhacker »

blazespin wrote: post #256
ruskyhacker wrote: post #210

Earlier quoted context omitted.

well my thought is that they 'whipped it up' real quick to attempt to downplay it a bit. Did that backfire? Yeah, I think so. But personally, I think people are missing where the real money is. OpenAI will do great for awhile until every damn product and service is using it, and then it's a race to the bottom. But that's just like my opinion...

I dunno, the stock price hasn't really tanked, so it didn't really backfire that much and it allowed them to get a bit of feedback as they try to figure out how to maintain revenue. What if they put out this insanely great model and people just stopped using search? That would be a backfire most likely.

You mean open ai put out a model and people stopped using search? Or Google?

What Open ai offers currently can't really compete with search - I understand the data it's being fed gets newer and newer, but it's not really real time like the search engines are. Indexing and presenting data is so different than NLM. Even if it is fed data that's new it's going to have to infer a lot because of a lack of history. It might be able to summarize recent events I guess. Way dumbed down here, but I consider chatgpt like a really smart encyclopedia that can search fast and stay in context across "searches."

If you meant Google, that's sort of what I'm saying - they wouldn't release something that could blow open ai out of the water. But I suspect what open ai offers as a product is something Google could've built long ago, or maybe did and couldn't figure out how to monetize it. They've instead invested in ai to make their products and services better, not as much to offer ai as a service.

What I meant by bard not working out so great was that Google quickly dusted off or slammed together some shenanigans to be relevant, even though what openai is doing doesn't appear to be a part of their master plan.

Re: What it feels like to work in AI right now

#347

Post by jacquesm »

UncleOxidant wrote: post #48
fdgsdfogijq wrote: post #39

Earlier quoted context omitted.

I dont work at Google, I think other FAANGS underinvested in this area as they didnt think it was promising. But I will admit, I am suspicious that Google is incompetent. Probably they can come back given how much money they will be forced to throw at it. But Bard is clearly behind and I dont believe their "abundance of caution" arguments for why Bard cant even code.

The impression I get is that they purposely limited Bard in certain ways. It does not seem to be willing to code and makes a lot of excuses when you ask it to (at least as of a couple of weeks ago when I last tried it). Did they put this limitation on it because it's not so good at coding or because they don't want it to be abused in some way? In my experience I had to trick it into coding by telling it to pretend th…

> With Bard I get the feeling they're not showing us what they really have

I highly doubt this. If they had it they would show it because if they don't react swiftly and decisively their brand will be in 'catch up' mode rather than out front where they are used to being.

Re: What it feels like to work in AI right now

#348

Post by lostdog »

letitgo12345 wrote: post #140
twelve40 wrote: post #99

Earlier quoted context omitted.

> Imagine Walmart launching a ChatGPT interfaced bot for shopping that customers take a liking to. I can't imagine that, because it doesn't seem to fit the use case. Especially not to the point of bankruptcy of Amazon, maybe as a small novelty? Can you list some killer features that the chat would bring that would make the existing shopping experience irrelevant? Maybe not everything is a nail to the hammer?

"I'm looking for X. Can you give me a few options with the benefits and drawbacks of each" or "I'm looking for X. Can you ask me a few questions and give me a choice of products based on that" or "I'm looking for a product that does X, Y, Z. Can you find such a product for me" or "Does this product do Y/is compatible with Z/is an appropriate give for person P" I'm not saying this will happen. I'm saying it's a risk t…

Everybody hates salespeople, and salespeople optimize for their own company's goals, just like an AI chatbot will have to.

There's no way you will want to interact more with Walmart's chatbot than necessary.

Re: What it feels like to work in AI right now

#349

Post by worrycue »

nazgul17 wrote: post #267
cgearhart wrote: post #152

From my perspective it’s just _confusing_ to work in AI right now. We have some massive models that are doing some really neat stuff, and apparently hundreds of millions of people are using them—but I keep wondering: to do _what_, exactly? I’m not asking what the models can do, I’m asking what people want the models to do every day, all the time. I’ve been shown some neat pictures people made that they thought were c…

I found ChatGPT helpful in a bunch of diverse situations. I'll go into detail below, but overall I think the most valuable thing it can do is understand questions that takes both knowledge (that I don't have and want to acquire) and understanding (that prior AI technologies did not have) and provide meaningful answers. I suppose I find it very useful for my own education - with the understanding that I should use 2 o…

A lot of your examples you can just google search for the answer - and at least google won’t hallucinate; worst case the websites are wrong but GPT’s training data can be wrong too.

Re: What it feels like to work in AI right now

#350

Post by chaxor »

atq2119 wrote: post #307
achenatx wrote: post #242

Earlier quoted context omitted.

its good for so many things: Need to write a speech as a best man Write a eulogy for a deceased family member Recently used it to speed up the time to fix an oauth problem in code I didnt write, knowing nothing about oauth. Replaces stackexchange with much more tailored answers. can take hundreds of pages of text and distill it down to an executive summary of any length Daughter uses it to explain how to solve algebr…

> can take hundreds of pages of text and distill it down to an executive summary of any length This is the kind of breathless claim that no doubt fuels the skeptics. None of the context windows are large enough for "hundreds of pages" nor "executive summaries of any length". I did believe that it's possible to make LLMs do that kind of task with significant engineering effort to make it do summaries iteratively someh…

There are plenty of tools that make it possible, such as langchain and the rest
Post reply on HN