Live data from Hacker News

Don't believe ChatGPT – we do not offer a "phone lookup" service

blog.opencagedata.com

221–230 of 284 posts

Re: Don't believe ChatGPT – we do not offer a "phone lookup" service

#221

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…

You're describing a problem as old as academia, on which very little progress has ever been made. Before "add a factual confidence level evaluator to a bot that doesn't understand the concept of fact" you must first figure out how to calculate a "factual confidence level" at all, in general.

There was a model that could set up a simulation to estimate the answer for you. So it won't use pure language, but it only works in a few cases.

Re: Don't believe ChatGPT – we do not offer a "phone lookup" service

#222

Earlier quoted context omitted.

By definition, an LLM doesn't have a semantic world model or ontology. Even the most "dumb" (and I use that in quotes because they really aren't) animal is able to reason about uncertain concepts and understands risk and uncertainty. Yann Lecun has posted a lot recently about this but basically LLMs are a "useful offramp on the road to AGI".

There's nothing "by definition" that says so. In fact many propose that when you train an LLM, in order to be able to predict the next word with enough accuracy, it must internally build a world model. Yann Lecun is very salty about chatgpt, I wouldn't take his word seriously.

LLMs are models of language, and language is a model of the world. So we have a model of a model of the world, but a LLM does not get much grounding in the real world.

There is a phase of training called multi-task instruction tuning where the LLMs solve problems and thus are grounded in exact answers. That makes the difference between the difficult to handle GPT-3 from 2020 and the better behaved GPT-3 of 2022. But that dataset is small by comparison to the raw text used in pre-training, it won't do the grounding perfectly.

Real grounding comes from real feedback, even humans need the feedback or we are just going on wild tangents.

Re: Don't believe ChatGPT – we do not offer a "phone lookup" service

#223
Related: One reason I just started using Rainforest API is because Github Copilot recommended it.

But also last night I tried for 30 minutes to get it to write me some fairly simple HTML parsing code. The tricky part was I couldn't use DOMParser since it was running on Cloudflare Workers and it could never produce any working implementation using HTMLRewriter or regex no matter how many examples I gave it

Re: Don't believe ChatGPT – we do not offer a "phone lookup" service

#224

Earlier quoted context omitted.

Great idea! Governments should start writing laws using LLMs.

If an AI is just a large language model without any ethical reasoning, what is a lawyer but the same thing with a smaller language model?

What do you mean? Lawyers have extensive ethical obligations.

Re: Don't believe ChatGPT – we do not offer a "phone lookup" service

#225

> 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.

Obvious suggestion: don't keep writing blog posts that mention your company name and the phrase "phone lookup service," which ChatGPT may get trained on in the future.

Re: Don't believe ChatGPT – we do not offer a "phone lookup" service

#226
post #208

> 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.

Maybe they gave a simplified explanation of their service, but if all they do is parse the country code of a phone number to return the geocoordinates for the center of that country then maybe just deprecate phone number inputs. I can’t think of why that’d be actual useful (a function that accepts a phone number country code and returns the center of the country’s geocoordinates) but if they have customers who use it…

They don't do anything with phone numbers. You can give them lat/lng coordinates and get the address, or an address and get the coordinates.

So "7 Carmine St, New York, NY 10014" will return "(40.7305290, -74.0020706)" and vice versa.

There are youtube tutorials claiming you can do phone lookups using their service. What these youtube tutorials really do is use some other library to determine the country name from the phone number. Then they call the OpenCage geolocation API with the country name as the address input.

Re: Don't believe ChatGPT – we do not offer a "phone lookup" service

#227
post #126
post #5

Earlier quoted context omitted.

have you been able to contact OpenAI about this? It sounds like they're actively adding load to your CS ops with this

I think the key thing is for the AI company to actually let the user know that this is a language model, and the information it spits out should not be trusted. Obviously, Microsoft is not going to do that as they are trying to market the new bing as a information search engine.

OpenAI does its best to make it clear that it is just a language model, but what can you do with you have users that just instantly click "Agree, Agree, Next, Next, Agree"

Re: Don't believe ChatGPT – we do not offer a "phone lookup" service

#228

Earlier quoted context omitted.

It’s not incorrect like the rest of us. It’s incorrect in a very different way. Providing detailed information on the usage of a service that has never existed is a brand new kind of incorrect that is carelessly causing the rest of us grief.

Every technology devolves to TV. The fact that you have to not only read, but write to interact with ChatGPT means 99.99% of people will not use it. I trust Alexa & Siri completely though.

Hook it up to fMRI

Re: Don't believe ChatGPT – we do not offer a "phone lookup" service

#229

I'm an attorney. I've typed legal questions into ChatGPT and it has spit out answers that are grievously, 100%, libelously wrong. It has named individuals and said they committed crimes, when it is unquestionable they did no such thing. I'm waiting for people to start calling me to ask questions about something ChatGPT said, and I'll tell them it's wrong. Then they'll start arguing with me and saying if ChatGPT said…

You're trying to use a language model as an information reference. A translator can explain what a diplomat is saying but they can't perform their whole job.

Re: Don't believe ChatGPT – we do not offer a "phone lookup" service

#230
post #123
post #30

Earlier quoted context omitted.

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…

Phone numbers have geographic structure. For mobile phones it's just the area code, but for landlines there is also information in the exchange portion. For example, I grew up in Medford MA which is 781-39x-xxxx. I expect there are also patterns in other countries?

This is what the inaccurate youtube tutorials do that they complain about in the linked blog post. It only gets you country level lookups for most phone numbers, county level at the very best. And it turns out users don't want that.

There's also the issue of moving. I have a Florida phone number. I haven't lived in Florida in a long time.

Post reply on HN