Funny algorithms make a serious point.

NewScientist reports on an amusing collection of algorithms that have come up with unexpected results. There’s the robot with six damaged legs that figured out how to walk on its knees instead. And there’s the AI tasked with writing a programme to limit the usage of a computer’s processor. It wrote a programme that simply put the computer to sleep. The serious point is that, if you knew the answer, then you would not bother getting an AI to figure it out. So, by definition, AIs will come up with the unexpected. Even if AI research rolls out a ubiquitous programme to interrogate all black-box AI solutions, it might be too late to stop any individual AI from going off the ranch.

Link to article:

You might like to browse other AI articles: