This isn’t the first time I’ve shared my relationship with artificial intelligence with you, and I’d like to do it again. Yes, I’ve activated the ChatGPT Edu license (thanks to an agreement between the CRUI ICT group and OpenAI, since last September universities have been able to access one of the leading generative AI platforms at favorable conditions). So, while writing this issue’s online learning dossier, I couldn’t resist the temptation to ask the program to create images of certain concepts. Indeed, what many have been telling me for some time is true: it can seem amazing how it manages to provide attractive images (even if the style is always the same, immediately identifiable, devoid of personality) that manage to capture the essential elements of even complex concepts. After spending time reading documents, texts, and processing previous experiences, she (or he, I’d prefer an esso or essa, which are rarely used anymore; I envy the English it) gives me back in a few seconds precisely the key elements I wanted. Of course, sometimes I’ve found repetitions or errors (in fact, I read in small print at the bottom of the screen that ChatGPT can make mistakes) that I can correct, and I’m happy to correct them (because, again at the bottom of the screen, I read that OpenAI will not use data from our university’s workspace to train its models). I’m not sure which feeling prevails: amazement (but it’s a software with millions of data points and information, the amazement and admiration are more for whoever was able to develop it and for whoever generated the information that feeds it), satisfaction (I had figured it out without its processing power), or anger (instead of spending all that time reading and studying, I could have done something else and asked the system directly). But the last feeling quickly fades and brings to mind the famous shorthand that were all the rage in my day, super summaries of entire subjects, lifesavers for so many students in preparing for high school exams and beyond. Yet, needless to say, the difference was huge between those who used them at most as a final aid to a long and dedicated preparation and those who used them instead as the sole source of preparation. And it will still be the same now, only everything is faster. But returning to my beautiful images, even though I only requested one image of a specific phrase, I find myself receiving a barrage of information and explanations I hadn’t asked for. Why? Is he doing me a favor? Does he take me for someone who, knowing nothing, asks the software directly so as not to waste time and effort? This irritates me, and I close the window without reading. I move on to the request for the next image, and here I have to make a few more attempts to get what I wanted. I think of the dozens of people, now expert users of the tool, who explained to me how important it is to ask questions in the right way. I tell myself that, after all, it’s an evolution of Internet search engines, where we’ve learned that depending on the search settings, the results change. But, I don’t know, I don’t feel like trying to ask the question better now because the AI ​​doesn’t understand. I’ll close it and go back to reading a fiction book on my comfy couch. Here, I’m the one who has to understand the text, which is easy if the writer is good and you’re used to reading. Exactly, some might say, you’re obviously not used to asking questions well. Exactly, I’ll retort, if you’re not used to reading and understanding, what use are you going to make of the AI’s summary?