[musicale]

> current-generation AI agents are too unreliable, too untrustworthy, and too unsafe for real-world use

...a completely unsurprising result, but it's nice to see published experiments.

Any agent system using current LLMs is likely to exhibit undesirable traits that derive from the training data.

[musicale]

> undesirable traits that derive from the training data

The research areas of model alignment and safety are attempting to address this fundamental problem - and have yet to solve it convincingly.

Problems like emergent misalignment can make things even worse.

https://www.nature.com/articles/s41586-025-09937-5