Generative AIs excel at organizing information thanks to their logical reasoning capabilities. By submitting our ideas to them, even in a scattered manner, they can reveal coherent structures and unexpected connections, often very useful.
Obvious to some, new to others
What I am developing here may be absolutely obvious to some readers because they already use it daily, and perhaps for others, this is not yet the case. Generative artificial intelligences have logical reasoning faculties linked to their design. Thus, they are effective, among other things, in finding structure in disparate things, in developing a plan, in finding common points. They can be of great help in organizing ideas, thoughts, and narratives because they have this greater capacity than ours to embrace a large corpus in detail and weave relationships within it. Artificial intelligences are all the more efficient when they have a lot to work with.
So, if you have noted, whether by hand on paper (as long as the writing is legible) or in a text document, scattered ideas, where you have written intuitions, desires, fears, questions, a kind of collection of raw materials, currently formless, and you give these elements to the AI in the form of a large PDF file, for example, or a series of images, you can then ask it questions based on your corpus. The answers will generally be very inspiring. Often, they will not be complete. You should not expect it to do the work for you, but rather to provide very useful inspiration.
It doesn’t do the work for us, it inspires us
For example, if from all these notes and ideas, you ask ChatGPT to write, without forgetting any of these ideas, a cinematic narrative that respects the spirit of your notes, or if you ask it to create, from these ideas, a list of characters that would illustrate them, or to link your ideas to the political context of 18th-century England, you will get suggestions that will need to be reworked, but which can bring you significant time savings as well as a distanced vision of what is understood of your intentions and the potential of your project, which you may not have identified yourself so explicitly and concretely.
This is quite unsettling, and even a bit “chilling,” because in just a few moments, you get a structured result. It may seem disembodied at first, but this structure has more to do with you than with ChatGPT. The AI has only been a machine of logical reasoning based on what you have given it and in connection with its learning model, that is, written language and all the literary works and reasoning it contains.
And ecology, in all this?
Yes, generative artificial intelligences are very large consumers of natural resources, they should be used sparingly and consciously. For example, the Swiss ecologist host Infomaniak offers generative artificial intelligence models that run on computer servers whose heat is used to heat 10,000 homes in Geneva, and in which they still use old computers that they do not discard, for tasks that require less power.
But I believe that these tools are absolutely fascinating, and the question lies in how to be in tune with our time, in the most ethical way possible, that is, in alignment with what we wish for the world. I believe that when it comes to generative artificial intelligences, we should not reject them outright nor use them without conscience. We can use them consciously. We can also download language models onto our own computers and run them “locally,” with our own electricity. It may be slower than on remote servers, but we know exactly what we are consuming.
