By Dr Miguel Ángel Serrano, President of EWC and Human writer.
The author shows an example of the ill “thinking” capacities of AI, resulting in a ridiculous solution from the machine.
The day before of our General Assembly in Berlin at the beginning of June was preceded by a welcome reception, in which some German personalities from the world of politics, culture, or music gave speeches and shared their points of view about the importance of written culture and, of course, about the topic of the day, Artificial Intelligence. The subject flooded the whole weekend. The next day we celebrated our traditional Burning Issues Forum, but this year, knowing the preoccupation of our members about AI, we decided to have a monograph on that topic.
When preparing it, the then President, Nina George, proposed a title for the event: “bot or not”. Building on it, our VP Maïa Bensimon came up with the idea for the slogan: “To bot or not to bot”. We liked it, and as I sometimes do, volunteered to design the graphics. We usually use Shakespeare’s image (e.g. our last two surveys on COVID 19 and its impact on writers). And this time the idea was clear: an image “created” by machines and another one created by a human. Machines don’t create anything; we need to be clear on this. They just compile from scrapped images or texts, and they are not able to understand what they have done. They don’t have criteria to elect anything. They just can offer a range of possibilities. The etymology of intelligence, from Latin inter and legere is clear. Inter means between, and legere can be read as select or read. That’s exactly what machines can’t do. That’s why they are not intelligent at all.
Let’s see a practical example. I decided on a design composed by an Artificial “Intelligence” versus another one designed by a human. The human one is from Nicoleta Ionescu and we rented it from a photo bank, 123rf. It’s a simple and nice, clear vector drawing. “Not to bot” was clear, after a deep search of the best image: close to our colors, and funny. Believe it or not, we are usually humorous people, aside from our work as activists… For the “to bot” part, I made a prompt to a software program, CrAIyon: “skull on the hand of a robot”. I imagined that giving background to the machine was useless. It didn’t need to know about the reflection of Hamlet in front of Yorick’s skull and its meaning in western culture. It doesn’t care about those things… So, the machine gave me a lot of possibilities, such as this one:
And the one I selected. It was so strange, and stupid, that it was attractive. The machine was not able to give context to the sentence, so it understood that that the skull was a part of the hand. An impossible malformation. The machine doesn’t cross information with the acquis of knowledge that a human being treasures. The result was only good, if it was at all, because a human put a criterion there. And let me remind you, please, that criterion means to select, also, in classic Greek. It means judgement, choice. It’s the same word as crisis. You need to judge something because it breaks…
I think that proves that machines just vomit puzzles without any practical distinction. An intelligence that bets only on random outcomes to solve a problem (a design one on this occasion) should have disappeared from the planet thousands of years ago. And it seems as if that is exactly our bet as a species these days.
Or have I a silicon brain inside my skull growing in my hand?
Miguel Ángel Serrano is a Spanish poet and novelist. He holds a PhD in Economy and Business Administration.