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

ChatGPT Enterprise

openai.com

Post reply on HN
Print view View original

421–430 of 532 posts

Re: ChatGPT Enterprise

#421

Post by stormfather »

ofermend wrote: post #385
mediaman wrote: post #82

Earlier quoted context omitted.

Pretty unlikely. Generally you don't use fine-tuning for bare documents. You use retrieval augmented generation, which usually involves vector similarity search. Fine-tuning isn't great at learning knowledge. It's good at adopting tone or format. For example, a chirpy helper bot, or a bot that outputs specifically formatted JSON. I also doubt they're going to have a great system for fine-tuning. Successful fine-tunin…

Totally agree. retrieval augmented generation is still the preferred way to give the LLM more knowledge. Fine-tuning is mostly useful for adapting the base model for another task. I wrote about this in a recent blog post: https://vectara.com/fine-tuning-vs-grounded-generation/ . Anyone knows how this new capability works in terms of where the model inference be done? Would it still be at the OpenAI side or is this go…

In your opinion, is it an either or scenario? Or would fine-tuning on docs + RAG be even more powerful?

Re: ChatGPT Enterprise

#422

Post by peoplefromibiza »

sinfulprogeny wrote: post #392
kortilla wrote: post #387

Earlier quoted context omitted.

Using chatgpt to code is not a skill. It’s a crutch. Any employees that feel held back by not being able to access it aren’t great in the first place.

Using $technological_aide \in {chatgpt,ide,stackoverflow,google,debuggers,compilers,optimizers,high-mem VMs}$ to code is not a skill. It’s a crutch. Any employees that feel held back by not being able to access it aren’t great in the first place.

> It’s a crutch

problem is most of the code chatgpt spouts is wrong, in so many subtle ways, that sometimes you just have to run it to prove it.

so basically you have to be better than chatgpt at that particular task to spot its mistakes.

using it blindly it's similar to the Gell-Mann amnesia effect

https://theportal.wiki/wiki/The_Gell-Mann_Amnesia_Effect

said by someone who uses chatgpt extensively, it is good for the structure, to get an idea, but as a code generator it kinda sucks.

Re: ChatGPT Enterprise

#423

Post by sgt101 »

arghwhat wrote: post #415
wodenokoto wrote: post #407

Earlier quoted context omitted.

Isn’t that a benefit of using a provider? Microsoft/OpenAI are selling a service. They’re both reputable companies. If it turns out that they are reselling stolen data, are you really liable for purchasing it? If you buy something that fell of a truck, then you are liable for purchasing stolen goods. But if it turns out that all the bananas in wall mart were stolen from cosco you’re not as a customer liable for theft…

Even if you find a way to successfully forward liability and damages to Microsoft and OpenAI - which I doubt you will be able to as the damages are determined by your use of the IP - you do not gain the right to use the affected IP and will have a cease and desist for whatever is built upon it. How legitimate the IP concern is and whether it holds up in court is one thing, but finger pointing will probably not be suf…

Also I thing that MS / OpenAI cannot and will not indemnify you. I think that their CEO and CFO have a duty not to...

Re: ChatGPT Enterprise

#424

Post by YetAnotherNick »

Gentil wrote: post #383

The ChatGPT model has violated pretty much all open source licenses (including MIT license which needs attribution. Show me one single OSS project's license attribution before arguing please.) and is standing still. With the backing of microsoft, I am confused. What will happen if they violate their promise and train data selectively from competitors or potential small companies? What is actually stopping them? Most…

> What is actually stopping them? Most companies won't have the fire power to go against microsoft backed openai.

Microsoft/Amazon/Google already have competitor's data in their cloud. They could even fake encryption to get all the customer's disk access. Also most employees use google workspace or office 365 cloud to store and share confidential files. How is different with OpenAI that makes it any more worrying?

Re: ChatGPT Enterprise

#425

Post by sensanaty »

fxnn wrote: post #417
irrational wrote: post #325

Earlier quoted context omitted.

My company (Fortune 500 with 80,000 full time employees) has a policy that forbids the use of any AI or LLM tool. The big concern listed in the policy is that we may inadvertently use someone else’s IP from training data. So, our data going into the tool is one concern, but the other is our using something we are not authorized to use because the tool has it already in its data. How do you prove that that could never…

For me that discussion is always hard to grasp. When a human would learn coding autodidacticly by reading source code, and later they would write new code — then they could only do so because they read licensed code. No one would ask for the license, right? So why do we care from where LLMs learn?

> So why do we care from where LLMs learn?

Because humans aren't computers and the similarities between the two, other than the overuse of the word "learning" in the computer's case, are nonexistant?

Re: ChatGPT Enterprise

#426

Post by blackhaz »

peoplefromibiza wrote: post #420
fxnn wrote: post #417

Earlier quoted context omitted.

For me that discussion is always hard to grasp. When a human would learn coding autodidacticly by reading source code, and later they would write new code — then they could only do so because they read licensed code. No one would ask for the license, right? So why do we care from where LLMs learn?

> So why do we care from where LLMs learn? same difference there is between painting your own fake Caravaggio and buying a fake Caravaggio (or selling the one you made). the second one is forgery, the first one is not.

The way I see it is that with AI you have really painted your own Caravaggio, but instead of an electrochemical circuit of a human brain you've employed a virtual network.

Re: ChatGPT Enterprise

#427

Post by tomjen3 »

kortilla wrote: post #387
fragmede wrote: post #382

Earlier quoted context omitted.

> How do you prove that that could never occur? Realistically you can prove that just as well as you can prove that employees aren't using ChatGPT via their cellphones. There are also organizations that forbid the use of Stack overflow. As long as employees don't feel like you're holding back their career and skills by prohibiting them from using modern tools, and keep working there, hey. As long as you pay them enou…

Using chatgpt to code is not a skill. It’s a crutch. Any employees that feel held back by not being able to access it aren’t great in the first place.

Here is a bunch of JSON. Output the c# classes that can deserialise it.

An intern in college could do that, but it isn’t worth our time to do.

For this function, write the unit tests. Now you do not have anything that you can blindly commit, but you are at the stage where you are reviewing code.

Could you do all of this by hand? Sure but you never would, you would use an IDE. Chatgpt is better than an IDE when you know how to use it.

Re: ChatGPT Enterprise

#428

Post by peoplefromibiza »

blackhaz wrote: post #426
peoplefromibiza wrote: post #420

Earlier quoted context omitted.

> So why do we care from where LLMs learn? same difference there is between painting your own fake Caravaggio and buying a fake Caravaggio (or selling the one you made). the second one is forgery, the first one is not.

The way I see it is that with AI you have really painted your own Caravaggio, but instead of an electrochemical circuit of a human brain you've employed a virtual network.

> but instead of an electrochemical circuit of a human brain you've employed a virtual network.

technically it is still a tool you are using, differently from doing it on your own, with your hands, using your own brain cells, that you trained over the decades, instead of using a virtual electronic brain pre-trained in hours/days by someone else on who knows what.

Re: ChatGPT Enterprise

#429

Post by rhdunn »

superfrank wrote: post #396
Gentil wrote: post #383

The ChatGPT model has violated pretty much all open source licenses (including MIT license which needs attribution. Show me one single OSS project's license attribution before arguing please.) and is standing still. With the backing of microsoft, I am confused. What will happen if they violate their promise and train data selectively from competitors or potential small companies? What is actually stopping them? Most…

> The ChatGPT model has violated pretty much all open source licenses Are you claiming this because they used copyrighted material as training data? If so, I think you're starting from the wrong point. Please correct me if I'm wrong, but last I heard using copyrighted data is pretty murky waters legally and they're operating in a gray area. Additionally, I don't think many open source licenses explicitly forbid using…

Taking this from a different perspective, let's say that ChatGPT, CodePilot, or similar service gets trained on Windows source code. Then a WINE developer uses ChatGPT or CodePilot to implement one of the methods. Is WINE then liable for including Windows proprietary source code in their codebase even if they have never seen that code.

The same would apply to any other application. What if company A uses code from company B via ChatGPT/CodePilot because company B's code was used as training data? Imagine a startup database company using Oracle's database code through use of this technology.

And if a proprietary company accidentally uses GPL code through these tools, and the GPL project can prove that use, then the proprietary company will be forced to open source their entire application.

Re: ChatGPT Enterprise

#430

Post by fdeage »

jrpt wrote: post #229
fdeage wrote: post #147

Earlier quoted context omitted.

Yes, this would definitely be a game changer for almost all companies. Considering how huge the market is, I guess it's pretty difficult to do, or it would be done already. I certainly don't expect a nice drag-and-drop interface to put my Office files and then ask questions about it coming in 2023. Maybe 2024?

You can use https://Docalysis.com for that. Disclosure: I am the founder of Docalysis.

Thanks, I'll have a look!
Post reply on HN