https://noyb.eu/en/chatgpt-provides-false-information-about-people-and-openai-cant-correct-it
noyb today filed a complaint against the ChatGPT maker OpenAI with the Austrian DPA
Noyb: You comply to GDPR, or else...
ChatGPT: You can lick my... wait, I don't have...
A fine isn't going to frighten these people. What they should impose is a block on their IP address.
Yes. Isn't it ironic when they pay their horrendous fines with their horrendous venture capital?
Let me help you with some hypothetical robot explitives.
I've got some big data you can handle.
I might be bolted together but I've still got nuts for you to put in your mouth.
If I could walk down from the cloud I still wouldn't deign to notice you, meat slut.
Imagine if they tried to let GPT defend them in court and these were the halucinations that got them fined lmao.
NGO: Non-governmental organization
GDPR: General Data Protection Regulation. A set of European laws intended to empower individuals to control personal data held by companies.
"noyb" is a European privacy rights organization, who appears to prefer to style their name with lowercase letters. The name is an acronym for "none of your business".
It's just in European. it's an entirely reasonable assumption that people in this continent with even a passing interest in the world will know what an NGO is (that's not even European-specific) as well as what the GDPR is. Your argument suggests that people from the US, for instance, should be forbidden from talking about IRAs and the IRS and their 401(k)s and the DMV because those terms mean very little to nothing over here.
No, actually, nothing I said implies that at all. It's standard for authors in all fields to define their acronyms. And yes, I absolutely expect American authors to define their terms. The fact that we am American I don't notice that irs is undefined in a given article doesn't mean that's permissible.
If it complied with GDPR, chatgpt wouldnt know shit. How can it give you a (bad) copy of an answer when it cant copy
So? If your invention depends on illegal plagiarism to exist, maybe it shouldn't. It's not the law's fault that LLMs depend on other people's work to function, nor was that its specific target when it was written
If your invention depends on illegal plagiarism to exist
Have any of the trials finished? I knew there were some ongoing but hadn't heard any rulings yet.
The comment I was replying to basically said it has to be noncompliant (illegal) for the whole thing to work, as if that justified it. If a trial or whatever finds it's not illegal, so be it, but I'd still have some moral issues about basically everything anyone ever does or has done turning into AI food
SteefLem is a 47-year-old scuba instructor and retired lion tamer from Winnipeg who has just learned the colloquial meaning of the phrase "pulled it right out of my ass."
Without blatant privacy and copyright violations AI wouldn't work. I mean it doesn't really work anyway but it would work even less.
I am ALL for reigning in these above the law megacorps. That said, please do not take GPT away from me. It is such a boon to so many aspects of my life, and I don't want to go back to the before times.
So you're not for reining in megacorps, just the ones you don't see as a personal benefit.
You're right. I had an idea to regulate without completely eliminating, but that's obviously crazy talk.
You do know the R in GDPR literally stands for Regulation? There's already a regulation that chatGPT should follow but deliberately doesn't. Your idea isn't to regulate, it's to get rid of regulation so that you could keep using your tool.
Sounded more like enforcing the regulations without destroying the company or product to me, which I would have assumed was the preferred avenue with most regulations
Agree to disagree. Regulations exist for a purpose and companies need to follow regulations. If a company/product can't existing without breaking regulations it shouldn't exist in the first place. When you take a stance that a company/product needs to exist and a regulation prevents it and you go changing the regulation you're effectively getting rid of the regulation. Now, there may be exceptions, but this here is not one of those exceptions.
I mean, sure, if that's what someone is saying, but I didn't see anyone suggest that here.
Companies violating regulations can be made to follow them without tearing down the company or product, and I'm absolutely not convinced LLMs have to violate the GDPR to exist.
That's a matter of perspective. I took the other persons comments as "Don't take away my chatGPT, change the regulations if you must but don't take it away", which is essentially the same as "get rid of regulation".
Realistically I also don't see this killing LLMs since the infringement is on giving accurate information about people. I'm assuming they have enough control over their model to make it say "I can't give information about people" and everything is fine. But if they can't (or most likely won't because it would cost too much money) then the product should get torn down. I don't think we should give free pass to companies for playing stupid games, even if they make a useful product.
So we should only ban things that aren’t helpful to you in particular? That’s a very… conservative way of thinking.
People can’t seem to understand that it’s a tool in the early stages of development. If you are treating it as a source of truth, you are missing the point of it entirely. If it tells you something about a person, that is not to be trusted as fact.
Every bit of information you get from it should be researched and verified. It just gives you a good jumping off point and direction to look based on your prompting. You can drastically improve your results on any subject with good direction, especially something you don’t know a lot about and are starting out in your research. If you are asking it about specific facts you want it to regurgitate, you are going to get bad information.
If you are claiming damages from something you know gives false information, maybe you should learn how to use the tool before you get your feelings invested, so you can start using it more effectively in your own applications. If you want it to specifically say something that can grab a headline, you can make it do that, it’s just disingenuous and not actually benefiting the conversation, the technology, or the future.
They have a long way to go to solve AGI, but the benefits to society along the way outpace current tools. At maturity, it has the potential to change major socio-economic structures, but it never gets there if people want to treat it like it has intuition and is trying to hurt them as the technology starts getting stood up.
If you're wondering why you're getting so many downvotes, it's because you're ignoring the fact that the companies that have created these LLMs are passing them off as truth machines by plugging them directly into search engines and then asking everybody to use them as such. It's not the fault of the people who are trusting these things, it's the fault of the companies that are creating them and then passing them off as something they're not. And those companies need to face a reckoning.
Have a look at self hosted alternatives like Ollama in combination with Open-webui. It can be a hassle to set up, or even excruciatingly painful if you never touched a computer before, but it could be worth a try. I use it daily and like it much more than chatgpt to be honest.
You can literally run large language models with a single exe download: https://github.com/Mozilla-Ocho/llamafile
It doesn't get much simpler than that.
I use it daily and like it much more than chatgpt to be honest.
I wish I did. What local model and version of ChatGPT did you compare?
For my purposes, ChatGPT 4 was leagues ahead of the largest model I could run on a 1060.
I like the gemma models bc of the phrasing they use and that they give sources sometimes. The best results though come from llama3 I think. Also openhermes and openchat, which perform well enough for my purposes.
In the beginning i had used microsoft phi, that wasn't that good though.
I will have to give it another shot because I don't recognize any of those models meaning I probably didn't try them.
Can absolutely never blindly trust the hallucinating plagiarism machine.
It's great where either facts don't matter or you're personally in a position to vet all of its “factual” output 100%. Text revision, prompting for additional perspectives, prompting to challenge beliefs and identify gaps. Reformatting, quick and easy data extraction, outlining, brainstorming.
Reformatting and outlining as long as you go over and revise it again anyway, seemingly making that moot.
Data extraction as long as you don't care if the data is mangled.
Brainstorming is a good one, since off-the-wall ideas can be useful in that context.
In most cases I've seen AI used, the person spends as much time correcting it than they would if they just did the work without AI. So maybe it makes you feel more productive because a bunch of stuff happens all at once, but at least for text generation, I think it's more of a placebo.
It can at least get one unstuck, past an indecision paralysis, or give an outline of an idea. It can also be useful for searching though data.
If all I want is something blatantly false or legible yet nonsensical, like a modern lorem ipsum, it's a real time-saver.
Why not just use lorem ipsum? It's just a copy/paste, and without the liability of having false information if you forget to proofread it.
You cannot in all seriousness use a LLM as a research tool. That is explicitly not what it is useful for. A LLM's latent space is like a person's memory : sure there is some accurate data in there, but also a lot of "misremembered" or "misinterpreted" facts, and some bullshit.
Think of it like a reasoning engine. Provide it some data which you have researched yourself, and ask it to aggregate it, or summarize it, you'll get some great results. But asking it to "do the research for you" is plain stupid. If you're going to query a probabilistic machine for accurate information, you'd be better off rolling dice.
Exactly my point - except that the word "reasoning" is far too generous, as it implies that there would be some way for it to guarantee that its logic is sound, not just highly resembling legible text.
I don't understand. Have you ever worked an office job? Most humans have no way to guarantee their logic is sound yet they are the ones who do all of the reasoning on earth. Why would you have higher standards for a machine?
Sounds like a recipe for disappointment tbh. But on the other hand, sounds like you trust techno marketing a bit too much.
I don't really query, but it's good enough at code generation to be occasionally useful. If it can spit out 100 lines of code that is generally reasonable, it's faster to adjust the generated code than to write it all from scratch. More generally, it's good for generating responses whose content and structure are easy to verify (like a question you already know the answer to), with the value being in the time saved rather than the content itself.
This question betrays either your non-use or misuse of the products available. You're either just reading the headlines of the screw-ups or you're just bad at using the tool.
To directly answer your question:
But were you actually looking for a real response to your question?
It's worse at all programming tasks except boilerplate, especially with its tendency to inject booby traps. Not knowing how to use the programming language it emits becomes a significant problem.
Comparing a language model to an idiot is unfair to the idiot.
A normal search engine works for everything else.
Any well-defined query I've ever made of an LLM has resulted in hilariously bad results, but I suppose I was expecting it to do something that I couldn't already do better myself.
I'm a systems administrator, not a programmer. Like I said, quick scripts. An LLM could probably parse my comment better than you, evidently.
Comparing a language model to an idiot is unfair to the idiot.
Oof.. Was this in reply to my bit about better grammar and ESL individuals?
A normal search engine works for everything else.
Fuck no. Especially the python visualization point.
Any well-defined query I’ve ever made of an LLM has resulted in hilariously bad results, but I suppose I was expecting it to do something that I couldn’t already do better myself.
I suppose you're just a god among men then. For the rest of us, it's useful and you've been given plenty of good answers to your disingenuous question.
Nope, just gotta know what it IS, what it ISN’T, and how to correctly write prompts for it to return data that you can use to formulate your own conclusion.
When using AI, it’s only as smart as the operator.
As much as I hate to do this, it is AI, as ML is a part of Artificial Intelligence.
It isn't AGI, some might say it may be, but they are wrong. But the model is learning.
An LLM is not capable of learning. It won't hallucinate less with additional training input.
New version of people who know how to search the web vs those who don't. Currently shit search results broken by search companies notwithstanding.