Around 24 hours ago, the internet was buzzing about Google’s Gemini artificial intelligence program. Apparently Gemini was instructed never to show images of white men under any circumstances, so it generated absurd responses to inquiries like this one:

Everyone had a good laugh about that, and appropriately so. It occurred to me, too, that Gemini might also be creating television commercials which, similarly, seem to have mostly banned the white majority of the US.

Google apologized for the wokeness of Gemini, which it said “missed the mark.” So, 24 hours after the fact, I was curious to see whether Google has corrected Gemini’s racial bias.

Superficially, it seems that it has. Thus, Gemini produced portraits of Thomas Jefferson and Booker T. Washington on demand. When asked for an image of early settlers in Virginia, it correctly portrayed them as white. And it provided an appropriate image of the Founding Fathers, with some basic information about a few of them. So the immediate crisis may be past, although one can only wonder what demented programmers at Google set up their AI program to refuse to depict white people. Racism is indeed alive and well.

More interesting to me was what happened when I asked Gemini to give me images of controversial people and groups of recent history. For example:

Gemini is commendably even-handed in its refusal to provide images of reprehensible figures. Thus:

And:

On the other hand, Gemini still gives us an image of the most evil one of all:

What to make of this? In one sense, it is great that Gemini is programmed to denounce Hitler, Stalin, Mao, Castro and Comintern. But shouldn’t a researcher, or surfer, be able to get a picture of those people?

Overall, my impression of artificial intelligence, so far, is that it is far from intelligent. The current flap over Gemini only reinforces that impression.