Microsoft’s 18 AI design commandments are here, recommending AI doesn’t over-promise or over-step

Microsoft has come forward with a list of 18 guidelines for better AI, and the list is a must-read for fans of artificial intelligence and the rapid progress being made in the sector and those adjacent to it.

You can see Microsoft’s full list of AI commandments here, and there’s not a ‘A robot may not injure a human being’ in sight. Instead, primarily the goal here is to create 18 bullet points that should help developers and designers work to create AI systems that are more “human-centered”

Related: Best Android Phones

The guidelines, which Microsoft claim are the result of more than 20 years of work, are important for designers because the standard design principles won’t hold for AI components as they themselves can learn and adapt over time.

Primarily, these commandments seem to be about being polite to humans, initially suggesting that a service powered by AI makes clear what it can do and how well it can do it upfront, in addition mitigating social bias, supporting efficient correction and dismissal, and being clear about the reasons it has taken a certain action.

Related: Best PC Games

There are actually only 17 points at the time of writing, no doubt due to an error, but please don’t blame me if number 7 turns out to be “teach your AI to torment the world”.

AI, as a broad catch-all term, has become a bit of a technology bogeyman in the last few years, but it’s interesting to see a lot of good, common sense, suggestions that warn people off of over-promising or over-stepping when it comes to AI, with the aim of creating a healthy industry moving forwards.

Does AI give you the heebie jeebies? Do you think we need to Feel free to reach out on Twitter at @TrustedReviews