[SystemSafety] Safety of AI trained to deceive

[Derek M Jones]

All,

A happy New Year to everybody.

Figuring out whether software is working correctly is
hard enough when it is expected to be telling the 'truth'.

A safety case for software trained to deceive sounds
like an oxymoron
https://shape-of-code.com/2022/12/28/the-commercial-incentive-to-intentionally-train-ai-to-deceive-us/

-- 
Derek M. Jones           Evidence-based software engineering
blog:https://shape-of-code.com