Guns don't kill people, its the people who keep loading them with bullets and shooting them at people.
ML isn't biased, its the people who keep loading them with biased data, and then running them against photos of people.
ML isn't biased, its the people who keep loading them with biased data, and then running them against photos of people.
It isn't clear what a similar safety mechanism would look like for an ML algorithm. Modern ML algorithms are nothing more than generic pattern extractors. What would a safety mechanism look like?
Have government grants for ML research be contingent on training on diverse datasets. Fund research into more diverse benchmarks and push them as the goal to beat.