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.
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.
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.
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 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.
Don’t you realize everyone exists to serve me?
No, they think everyone exists to serve them.
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.
is the perfect description
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.
Thanks!
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.