“AI’s ability to invent a joke has not improved much”

All summer long, we interview employees, freelancers, and business leaders about their relationship with artificial intelligence. How do they use it, how do it change their professional practices? Today, cartoonist Kak.

Patrick Lamassoure, aka KAK, wears two hats. He is the president of the association Cartooning for Peace, which defends freedom of expression through press cartoons. Every day, he also draws the front page of the daily The OpinionThe first time he used generative artificial intelligence was a little over a year ago.

“I put the AI ​​in the shoes of the press cartoonist. I told it: I’m at a party where people have just received a trophy because they have done things that have changed society, what funny drawing would you do? And there, no answer, panic on board. I had to give it the idea: I would like a drawing where Vladimir Putin arrives in front of the office where the trophies are awarded, he has a nuclear bomb under his arm and asks to apply. The AI ​​told me it couldn’t because it didn’t comply with its conditions of use. By removing words, I understood what was bothering it. First, it didn’t want to draw Vladimir Putin, but it did want to draw a Russian. Then, it didn’t want to draw a nuclear bomb, but just a bomb. I ended up getting a drawing that was still not in my style, but which looked more or less like what I wanted to do. I had to touch it up. So I wasted time on this affair.”

While cartoonist Kak doesn’t use generative AI in his daily practice, he continues to test them regularly.

“I found that AI’s ability to invent a gag, starting from almost nothing, had not improved. Where I was quite astounded was when I suggested to it how I wanted to illustrate a drawing on the debate between two rounds between Gabriel Attal and Jordan Bardella. I wanted to stage two knights, who faced each other, in the style of a medieval joust. Above them, the sovereigns Emmanuel Macron and Marine Le Pen threw their sashes at them. ChatGPT, to whom I first asked the question, offered me a description of the drawing that went well beyond what I had asked him, totally in line with my logic. He had extrapolated the idea to make it almost a painting, with composition ideas. Everything he suggested to me was interesting from a graphic point of view.”

Kak believes that AIs are incapable of generating a good press cartoon today, in a specific cultural context and for a given audience. On the other hand, he already sees the tasks on which he could be supported.

“I’m not very good at colors, it’s a laborious exercise, I generally consider that the result is not pretty. I can see that I could ask the AI ​​to create an atmosphere for me by taking inspiration from a painting by Turner, Van Gogh, a watercolor, etc. Then, there is everything that relates to graphic research. If I need a garbage truck seen from three-quarters back in my drawing, today, I go to a search engine to find images, which I have to go through one by one. The AI ​​could find them for me in two seconds.”

For Kak, generative artificial intelligence represents the biggest technological leap since the arrival of the Internet. What intrigues him most is to see how humans will collaborate with them, and above all, how they will maintain control over them.