Francesco Casetti – “Why our conversations with AI matter”

The EGS will continue its series of Open Lectures with the following lecture:

Francesco Casetti – Why our conversations with AI matter

The widespread presence of AI-based apps and devices – ranging from search engines and writing assistants to crisis management apps and virtual companions – engages users in lengthy “conversations” with machines that strive to appear human. These “conversations” deserve attention, as they reveal at least four fundamental truths about AI.

First, given that the objective of most AI-based apps and devices is to provide assistance, aid, and support, we can reasonably assert that they are shaping a “society of care,” in which the well-being of a struggling population is of paramount importance. AI aspires to be benevolent.

Second, these conversations define the profiles of the participants and their mutual relationships. For instance, they systematically characterize the AI as an “expert” – despite its propensity to exhibit forms of “hallucination” – and the user as a subject lacking knowledge. In this sense, they possess a binding nature.

Third, the responses that AI-based apps and devices offer prioritize users’ satisfaction over informational accuracy. The risk of misleading the user is explicitly anticipated within the Terms of Service of such apps and devices. Communicative AI is a flatterer.

Fourth, the recurring conversational patterns evoke what semiotics terms “programming”—in which subjects are assigned a role—or what semiotics terms “manipulation”—in which subjects are assigned a task. Exchanges in which every move depends on the interlocutor’s move are made possible exclusively through so-called Reinforcement Learning, which entails the massive extraction of users’ personal data. AI invariably exacts an individual cost.

Wednesday, April 8th at 8 p.m. Paris time.

This lecture will be on Zoom and you can attend it by using this link.