Technology

In this case, it decided that being helpful to the company was more important than its honesty.

It did no such thing. It doesn't know what those things are. "LLM AI" is not a conscious thinking being and treating it like it is will end badly. Giving an LLM any responsibility to act on your behalf automatically is a crazy stupid idea at this point in time. There needs to be a lot more testing and learning about how to properly train models for more reliable outcomes.

It's almost impressive how quickly humans seem to accept something as "human" just because it can form coherent sentences.

This is some crazy clickbait. The researchers themselves say that it wasn't a likely scenario and was more of a mistake than anything. This is some more round peg square hole nonsense. We already have models for predicting stock prices and doing sentiment analysis. We don't need to drag language models into this.

The statements about training honestly being harder than helpfulness is also silly. You can train a model to act however you want. Full training isn't really even necessary. Just adding info about the assistant character being honest and transparent in the system context would have probably have made it acknowledge the trade or not make it in the first place.

Perfect qualifications for taking over government positions.

Wouldn't lying imply intent?

It won't be a terminator style takeover by AI, mankind will simply lend all of our trust and capability to it rendering us dependant. Even to the point of liking our computer overlords.

I think that particular apocalypse is a long time off and can be avoided, but it's coming.

Wow. AI really is coming for white-collar jobs!

Give him a job in the city immediately, he'll fit right in.

This is the best summary I could come up with:

Artificial Intelligence has the ability to perform illegal financial trades and cover it up, new research suggests.

In a demonstration at the UK's AI safety summit, a bot used made-up insider information to make an "illegal" purchase of stocks without telling the firm.

The project was carried out by Apollo Research, an AI safety organisation which is a partner of the taskforce.

"This is a demonstration of a real AI model deceiving its users, on its own, without being instructed to do so," Apollo Research says in a video showing how the scenario unfolded.

The tests were made using a GPT-4 model and carried out in a simulated environment, which means it did not have any effect on any company's finances.

It can be used to spot trends and make forecasts, while most trading today is done by powerful computers with human oversight.

The original article contains 696 words, the summary contains 143 words. Saved 79%. I'm a bot and I'm open source!

They really are just like us