Nostr Web Client

Trying to understand how I’d connect a fork of a chatGPT to a non open source code base and have it work through certain tasks/make identifications while maintaining security. Where should I be looking for the different pieces of this? I assume creating/modifying the bot + connecting it to the api are the biggest hurdles and I’m not confident in how to do so well.

I’ve said this in the past but I’m still very new to coding and learning as I go at work. But I am convinced that much of the smaller tasks I do can be replaced. I want to get one the front of these developments and be able to confidently apply tools to the company I work for, within this realm.

async 2y ago

In the case of Gpt needing to understand the semantics of the private info and act based to it I think it is difficult to maintain security without spinning up a private LLM

If data could be “masked” deterministically in a way that the decisional behaviour wouldn’t be changed… I’d say you have some possibilities even using openai

Reply to this note

Please Login to reply.

Discussion

Matthieu 2y ago

Ahh. Got it. So because it (chatGPT) depends on public info/access to make decisions, it’d be more difficult for it to be entirely secure?

Essentially wanting to vet out information - so I think it’s possible but obviously want to keep security at the forefront. So maybe finding ways to obfuscate the data from something serviceable outside of our system?

I imagine a private LLM is a much larger undertaking (to put it mildly)