Focus of Letter 43

If we replay the 43 weeks of this letter in a hurry, the models have evolved, the tools have multiplied and the uses have diversified. I believe that what has changed the most in one season is the tone of the debate around all the subjects that bring us together here.

At the beginning of the one-AI adventure, I was able to talk about it several times, many unknowns, few certainties, a form of wonder mixed with concern that still left a lot of room for questioning. Today, I have the impression that certainties have accelerated faster than uses. Perhaps because the debate around AI is increasingly following the model of other media and societal polarities. Evangelists and alarmists occupy the field and talk to each other less and less. Between the two, the space to not yet know, to look, to change one’s mind seems to me to be shrinking more and more and the readings of all these weeks, more than 370 compiled in our letters since September, seem to me to show this. All realities coexist simultaneously and those who defend each of these positions rarely find themselves in the same conversation. So, where are we talking about when we talk about AI? I can at least answer for myself.

A letter that has been going on for 43 weeks has already said a lot about its author. As you know, if there was only one binary choice and if I had to be classified in one of the polarized camps, it would be more on the side of the “pro”. Being on one side or the other is not a question of naivety. I’ve been interested in the digital fields for a long time now, long before AI became a mainstream subject, and I think I’ll be able to measure both its promises and its limits.
The site has been publishing since February 2023. Forty-three weeks of letters. What it means to me: testing to discover, judging relevance, sorting with assumed subjectivity, and sharing to allow others a first opinion without having to do this skimming work themselves.
On average, one application published for every five observed or tested, certainly without precise statistics on this point, but the order of magnitude seems to me to say something about what “sharing” implies as a silent effort upstream. This daily work forges an opinion and it forged mine. It could have produced the opposite conclusion, an excessive dependence, a low added value, a lot of noise for little innovation. That’s not what I remember from it, even if it could have been a possible trajectory.
For example, I think I underestimated the speed at which the content was going to look the same. I thought that the diversity of tools would produce diversity in the results. It’s not that simple, and it has become one of the issues related to delegation and agency to be observed closely, in my opinion.
I also thought that the automation of daily tasks would happen quickly for everyone. What I observe is that the gap between those who know how to configure the tools and those who do not use them or in an occasional and manual way has widened deeply. Automation, even if very partial, is in fact still very far away for the ordinary user, contrary to what many discourses may suggest.
These daily observations force a discipline: to remain open to what surprises, even when it contradicts what we thought the week before. Being for or against is not wearing blinders, it does not exempt us from observing. In my opinion, it increases the obligation.

I find that we are more and more often confusing conviction with rejection. We can think that AIs open up real possibilities and still find the concerns of those who see them as a threat legitimate. We can refuse these tools or be deeply suspicious of them and still recognize that those who use them are not naïve. We can be concerned about the cognitive effects on its users, and even more so when we talk about education, without condemning the tool as a whole. We can have chosen not to touch them and still recognize that they change something in the world we inhabit. We can use these technologies every day and not dismiss the question of their environmental footprint, which is more than considerable, ignored or misused by the giants of the sector. We can refuse them and not be blinded by this refusal alone: understanding this system, its logics, its geopolitical and economic orientations, is a form of civic thinking that neither enthusiasm nor rejection exempts from pursuing. We can use them on a daily basis and not lose sight of the fact that our own use is only one angle among others in a system that is much broader than what we see of it.
What is often missing in these debates is the big picture: AIs in the world in which they are established, not just through their own use or sector. The nuance is not the absence of conviction: it is the conviction that has survived contact with other points of view. And this survival requires an active effort, not a passive position.

Dialogue, here, with such a global subject, does not only mean talking to someone who thinks differently, even if that is already a lot. It also means reading what one would not read spontaneously, informing oneself outside one’s bubble, taking the time that a continuous flow does not allow. These forty-three letters have tried to go in this direction: educational tools and school bans, studies on the biases encoded in models, reflections and points of view on work, reports and research articles on societal uses and impacts, questions about what information loses when agents read in the place of readers, debates on creativity, memory, delegation, autonomy. Not a single camp. Several realities at the same time, without a verdict rendered in advance.

The letter stops there for the summer. It will come back at the end of August. Until then, I propose that we take away only one question: on what will we have evolved and changed our minds, between now and the start of the school year? Not only what we have confirmed but what we have revised.

Echo:

• Letter 40, “In the rear-view mirror: 39 words for 39 letters“
• Letter 38, “The debate that may not be a debate“