Are AI design guidelines actually being followed?

Apr 23, 2023

A lot of posts on how to use generative AI to be a better UX designer, but not that many on how to design better AI. So here’s one.

Did you know there are several guidelines out there for designing AI? They make a lot of sense but after going through them I have a feeling they’re just a bunch of good looking ideals sitting on a bench, not being followed by the vast majority of AI-powered products.

Most of their principles overlap in some way. These guidelines agree on important things like Explainability (users should be able to understand why the outcome of an AI system is what it is) and Control (to have a customizable balance between control and automation). But I’m not sure I’ve seen these that much.

Current AI being far from all this might have to do with a lack of awareness that intelligent systems need to be designed and not just built. There’s still too much focus on the technology itself instead of the people it serves.

That’s why designers are needed in the AI field. It will be a challenge because most of us don’t know how designing for AI looks like yet, but unless sensationalism and hype puts AI through another winter —that’d make for an interesting plot twist— this will eventually change.

Here are some of these guidelines in case you want to see for yourself how far we are from them:

Google’s PAIR Guidebook

IBM’s Everyday ethics for AI

Apple’s Human Interface Guidelines: ML

Microsoft’s Guidelines for Human-AI interaction

OECD AI Principles

more posts

Ergonomics will define the success of Apple’s Vision Pro

Ergonomics will define the success of Apple’s Vision Pro

A short comment on Apple’s Vision Pro hype: yes we might be living in an increasingly digital world, but let’s not forget that we still experience all of it through our physical body. Ergonomics is the main reason why VR has never really taken off domestically. Vision...