121–130 of 284 posts
ChatGPT is hilariously buggy - I asked “it” how to use an open source library i made. The output was wrong ranging from a broken github url to outright broken or nonexistent code. I suspect it may even have used private code from other libs - couldnt find some of the output it generated anywhere public.
Earlier quoted context omitted.
What would you call it instead?
A glorified Markov chain generator. Now, humans could very well also be statistical inference machines. But they have way more tricks up their semantic-level understanding sleeves than ChatGPT circa 2023.
This is not a service we provide. It is not a service we have ever provided, nor a service we have any plans to provide. Indeed, it is a not a service we are technically capable of providing. I'm curious: why not? It seems like a lot of people would be interested in this if you could figure out how to provide it.
How would this work? If a phone number is for a mobile phone then looking up the location doesn't make sense at all: mobile phones are mobile. I guess you could try and crawl an index of business phone numbers and associate those with the listed address for businesses, but that's a completely different business from running a geocoder. You could provide a bit of geographical information about the first three digits o…
I expect there are also patterns in other countries?
The biggest takeaway for me was that it was getting info from YouTube videos. Is it actually watching and learning from the videos or where links to GitHub just included in the comments?
ChatGPT very convincingly recommends us for a service we don't provide. Dozens of people are signing up to our site every day, then getting frustrated when "it doesn't work". Please do NOT trust the nonsense ChatGPT spits out.
have you been able to contact OpenAI about this? It sounds like they're actively adding load to your CS ops with this
> All suggestions are welcome. Monetize it! Evil answer: Partner with an advertiser and sell https://api.opencagedata.com/geocode/v1/json as an ad space. This may be the first opportunity for an application/json-encoded advertisement. Nice answer: Partner with an actual phone lookup platform and respond with a 301 Moved Permanently at the endpoint.
there's no "actual phone lookup platform" you can't get a person's location by knowing their phone number, that's a huge privacy violation. You can get the location of your own phone via icloud or google's system for android. You could also install an app on your phone to track your phone's location. You cannot find people based on knowing their phone number, that would be a serious safety issue for you know people t…
I think white pages are still a thing, no?
For the young 'uns - the white pages were part of the physical phone book in every city. You got a new phone book delivered to your doorstep each year for free. Yellow pages listed the phone numbers of every business, white pages listed the phone numbers of the residents.
The crazy part is: almost everyone added their numbers voluntarily to the white pages, because you wanted people to be able to easily find and reach you.
Earlier quoted context omitted.
I get frustrated at the number of things ChatGPT gets blamed for that aren't its fault. It is completely understandable that if there are repos out on GitHub like the one for Phomber[1] thant ChatGPT would find that code and have no idea that it was phoney. Suggesting that ChatGPT just made this up out of thin air when you know it didn't is not very responsible. 1. https://github.com/s41r4j/phomber
You are blaming the victim. OpenAI is to be blamed. They know what they are doing. They provide something that sounds over-confident for anything it says, knowing full well that it can't actually know if what it generated is accurate because it is designed to generate plausible sentences using statistics and probabilities, not verified facts from a database. On top of it, they trained it on an uncontrolled set of tex…
The key thing is that it isn't a knowledge base. It doesn't claim to have correct information about the world. It has the ability to translate a question in natural language into what would be an answer in a natural language - but it isn't necessarily correct because its about the language rather than the knowledge.
People misusing the LLM as a knowledge base are at fault just as a person misusing a CD tray as a cup holder is at fault if it doesn't work correctly as a cup holder.
I'm curious -- does anyone know of ML directions that could add any kind of factual confidence level to ChatGPT and similar? We all know now that ChatGPT is just autocomplete on steroids. It produces plausibly convincing patterns of speech. But from the way it's built and trained, it's not like there's even any kind of factual confidence level you could threshold, or anything. The concept of factuality doesn't exist…
There's research being done on this: https://arxiv.org/abs/2302.04761 At its core using an LM alone to solve factual problems seems silly: It's not unlike asking Dall-E to draw DOT compliant road signs. I've gone at length at how unfortunate it would be if LMs start to get a bad rap because they're being shoehorned into being "Ask Jeeves 2.0" when they could be so much more.
I love that. That's going to be my new explanation for people around ChatGPT.
For some reason it seems so much more obvious when Dall-E does something close but still totally wrong (e.g. 3 or 6 fingers, 3 arms, etc.), but it's not immediately obvious with text. But it's still the same underlying principles.
I'm curious -- does anyone know of ML directions that could add any kind of factual confidence level to ChatGPT and similar? We all know now that ChatGPT is just autocomplete on steroids. It produces plausibly convincing patterns of speech. But from the way it's built and trained, it's not like there's even any kind of factual confidence level you could threshold, or anything. The concept of factuality doesn't exist…
> does anyone know of ML directions that could add any kind of factual confidence level to ChatGPT and similar? Yes. It's a very active area of research. For example: Discovering Latent Knowledge in Language Models Without Supervision ( https://arxiv.org/abs/2212.03827 ) shows an unsupervised approach for probing a LLM to discover things it thinks are facts Locating and Editing Factual Associations in GPT ( https://a…