You may have heard that Google recently released a generative AI service called Gemini, and that it’s comically stupid. The people who programmed Gemini were apparently so worried about so-called diversity, equity, and inclusion that they made the most racist, absurd self-parody anybody could have imagined. It’s absolutely astonishing.
I’m not going to go into the details – there’s plenty online about it – I want to point out the implication for management.
Why was the internet able to find this problem in about two seconds, but Google’s testers didn’t notice it?
The answer is simple. There’s no viewpoint diversity at Google. They probably don’t have a single conservative working there – and certainly there were none on this project because they would have immediately recognized the problem.
The lesson is quite simple and clear. If you don’t want to make a fool of yourself, make sure you have people representing a variety of political and social viewpoints, and make sure they’re encouraged to speak up without fear of retribution or censor.
