Should we let AI respond to messages for us?

” I’ll call you later ? » The message is too laconic, vaguely strange with this question mark which has nothing to do there, and you recognize it immediately, without ever being really certain of its origin either. With their wording sometimes too neutral, sometimes too cheerful, their slight discrepancy with a message written by a human, automatic responses have discreetly insinuated themselves into our conversations. In 2016, Google launched Smart Reply, an automatic email response suggestion service, designed to serve as many people as possible (” well received “, ” With pleasure “, ” heard “etc.), from which we only have to choose. No need to waste time typing long words like “LOL” or “thank you”, the machine does it for us.

Read the survey | Article reserved for our subscribers Artificial intelligences, the thousand and one ways to derail them

At a time when ChatGPT, OpenAI’s conversational robot, can write personalized cover, love or break-up letters, the concession seems trivial: what do we put of ourselves when we send ” I’m in a meeting “ ? There are those who embrace technology, and only see in these little crutches the saving of time, or who consider it a consideration to respond quickly to their correspondent. In 2017, already 10% of emails exchanged were automated.

And there are the resistance fighters. Those who feel they are losing something of themselves by delegating their expression, even in its most basic form. “It is the eternal debate which already existed in the 19the centuryasks François Jarrige, specialist in the history of technology and industrialization at the University of Burgundy. Yes, it’s more convenient, it saves time. But, at the same time, we are aware that we are part of a long-term dynamic of alienation, and that this micropractice has much broader effects. »

“Last Frontier”

“I would be ashamed to do it, it’s contemptuous of others. And I am not a machine”testifies a reader before adding : “Even if I often fail the test to prove that “I am not a robot”. » The degree of disturbance is in fact comparable to that which we feel when a computer orders us to prove to it that we are indeed human by clicking on images of trucks. When I receive a ” I’ll call you back ? », who is really talking to me? When I validate a suggested response, am I really expressing myself? “We almost reach the end of logiccontinues François Jarrige. We are in the process of automating a whole series of practices that were considered the last frontier of what is impossible to automate. »

You have 66.31% of this article left to read. The rest is reserved for subscribers.

source site-23