- Researchers develop programmes that ask computers to do something and then deliberately thwart them
Researchers are developing programmes which will ask computers to try and do something only for them to be deliberately thwarted.
By understanding the difference between the desired outcome and the reality, the machines will learn a sense of ‘regret’ and how to minimise it.
Computers which experience this will be less likely to make the same mistake in the future and will run more efficiently.