Busted.

The use of artificial intelligence is becoming popular. It can do some amazing things but it is far from perfect.

Chat GPT is one chat bot used often. Like much of AI, it is so permeated with liberal programming that it cannot force itself to portray the president of the United States in a good light.

I went to Chat GPT to see how it was at generating images.

This was my first prompt: For a Facebook meme, Donald Trump as a knight in shining armor, fighting dragons and human enemies, including one in black armor labeled “media.”

This was the result:

Notice how Trump is shown as a bully, in black armor, with a scowl. My next prompt: Good but I want more action, with enemies more visible.

The result:

My response:

That’s no good at all. Trump is not identified and there is no fighting. I want Trump in shining armor surrounded by dragons and enemies, fighting them and slaying them.

Chat CPT:

My next prompt:

No, I don’t want the dragon as hero. I want Trump as in the original image, but fighting against the enemies, which include dragons.

The bot produced this:

I responded: That’s better but I can’t identify the hero as Trump. It needs to have his face visible.

And I got this:

At this point, I explained again what I wanted and asked, “Can you do it?”

Instead of an answer I got an ad offering me a subscription.

Chat GPT simply was not going to show Trump in a good light. So I left, returned later and asked the original question again but substituted the name “Barack Obama” for Trump.

And got this heroic piece of art.

That is almost exactly what I asked for originally, but the AI bot clearly has been programmed to allow only liberal heroes and conservative villains.

Even in AI, “Orange man bad.”

Use it with realistic expectations.