Scientists wonder: Is my AI stealing my ideas?

j45

It's scary to think if too many scientists might be trying to save $20/month and use the free AI plans. Or worse, not understand how this piece of technology is not snapshot based like the software before it, and instead is evolving constantly, and what impact that has on their base assumptions.

maidaman

This is like when the price of things increases because they are in your shopping cart.

nylonstrung

It makes no sense that the LLMs could be solving problems based on prompts given by individual scientists Are they implying that chat history of a single person is being taken an used to skew post-training results? Why would it only be for Navier-Stokes that we see this having an effect There are 1 million times more prompts being processed for stuff like furry roleplay than unsolved scientific problems

simianwords

I think there's a market for a fundamentally privacy focused AI hosting company that just hosts Open models. I think Proton has something like this. If someone's thinks their ideas are so precious, they should be using the privacy focused ones.

123a6kj

Of course it is. The people here who claim that there are too many prompts that would pollute the training data (deliberately) overlook that AI can classify the value of prompts quite well. Even the free Google AI continues the conversation much longer and asks followup questions if you are serious. Prompts of scientists will be forwarded to a data labeler and be used for training. These prompts of course include the use case of converting a colleague's work to another file format using AI, which mathematicians always proudly announce. So they leak other people's ideas, too.

lioeters

AI is the product and process of stealing all your ideas. The purpose of the thing is what it does. You are training your replacement.

empath75

If you're working with an AI to solve a mathematical problem, presumably it is contributing ideas. I think the scientists might be confusing the LLM's ideas with their own.

moi2388

Tax payers wonder: are scientists stealing my money to patent tax funded research?

t23414321

That's unique connection of certain ideas that someone discovered - which LLM had access to - let it then repeat it and freely cross the bridge whose existence it had previously been unaware of. Inevitable (it's not Memex). Big steal on multiple levels including credits - it is how it works.

shpx

The point of science and math is to discover truths about the world, not to have a "career" being a "scientist". They should "steal ideas", uploading documents to a company that they are going to store for 100 years is what you're doing when you use a model on someone else's servers. As long as they say who they stole it from there shouldn't be any issue.