Generative AI is changing pace: what if slowness was progress?
Focus of Letter 12
Three apps for a focus this week!
Admittedly, making this choice on Saturday is not always easy… This week we have an application and a browser oriented towards automation, FlowithOs and Aident, and a language model oriented towards reasoning and complex tasks, Gemini 3 Pro… I group them all three.
They seem to me to bring an important change in the approach we have or have had with generative AIs. The “reasoning” that segments complex tasks into a succession of simpler tasks and the agents that are put in place, brings, in addition to an increased relevance of generations, a different time, a longer time for producing the answers provided.
This relationship to time raises questions, I don’t have a ready-made answer, I will continue to dig into this point and some of you will certainly be able to answer these questions with your expertise. So I give you some of my thoughts, ongoing and moving, which I think touch on our relationship to generative AI:
- Technological limits that are now made visible?
- A design or marketing effect that would result in the display of reasoning steps to make us “wait” after having accustomed us to almost instantaneous answers?
- A gradual abandonment of the “wow” effect brought by the speed of response, synonymous with an increasingly massive adoption that no longer requires “convincing” new users from the first approach? This “wow” effect being displaced on the response provided or its formatting?
- Or the search by publishers for increased relevance, requiring verification phases that take more time? Would quality then take precedence over quantity?
The answer is certainly a mixture of all these points and there will certainly be others to add or who will challenge them… In any case, I think they lead us to question ourselves once again – and each time – about our reception of the answers given and about our relationship with the machine.
Can’t wait to read your reviews on these questions!