Hacker News Reader

Comment by delusional | Hacker News Reader

> The current state of AI is a perfect mass surveillance technology.

How? Are you just going to ask the LLM about who is doing the crime? OpenAI is not an "AI company", it's an "LLM company".

View on HN

You ask it to do the things that human analysts already currently do with the vast amounts of text, image, and video data they collect but at an unprecedented scale. You can extract what people are talking about, identify people, weapons, or objects in images, extract addresses, license plate numbers, or other identifying references. It can probably even detect when people are using coded language that won’t trigger explicit keywords.

OpenAI is an AI company, it’s literally in the name. Even currently they use more than LLMs, they use other transformers and related technology in the field of AI.

There's also the option of running sentiment analysis on social media posts, easily allowing placing individuals on a left-to-right scale of political leaning.

Yes, you can do all those things, and critically, if the bottom falls out of the market, you can make that case to the US government, complete with a fear-based narrative (China!) and get bailed out by the American taxpayer.

It is even worse. The AI can build a profile of you and then feed you stuff that works on you.

Not even fear, more like a personal bubble that feeds you stuff that works on you

I'm gonna go out on a limb and say no it can't. It's not a super intelligence. If an ai tried to market to me right now it would be roughly equivalent to an enthusiastic and very naive intern trying to market to me based only on two random pieces of my biographical information. I'm much more concerned about the surveillance

You do what they're already doing - ask the LLM to summarise someone's entire facebook feed (including all the private stuff they can access) into a few bullet points. Whether it works reliably is a different matter.

Reply on news.ycombinator.com