Microsoft boss Satya Nadella doesn’t really like the tendency of AI to move towards anthropomorphization. The recent trajectory taken by OpenAI, a very close partner, is indirectly criticized.
It’s a little sentence that could sound like a call to order for OpenAI. During an interview with Bloomberg on May 21, 2024, the boss of Microsoft displayed a firm stance with regard to a trend that we are seeing increasingly appear with progress in artificial intelligence (AI). : anthropomorphization.
This phenomenon consists of giving a human appearance or behavior to a thing (or an animal). For example, this translates into AI simulating a voice as natural as possible and offering reactions that resemble those of real people — laughter, breathing, intonations, interjections and so on.
This trend displeases Satya Nadella. “ I don’t like the anthropomorphization of artificial intelligence. I think it’s a tool “, underlined the American leader. This seemingly innocuous remark came a few days after the presentation of GPT-4o, a new language model designed by OpenAI.
During the demonstrations, OpenAI showed an evolution of the tool called ChatGPT Voice. It is possible to ask him to sing, take on a robotic voice or be dramatic. He responds straight away, makes jokes and has astonishingly realistic intonations. ChatGPT Voice even imitates human breathing.
Controversy over Scarlett Johansson’s voice and OpenAI’s attitude
Added to this was a controversy involving Scarlett Johansson. A voice from OpenAI’s voice assistant resembles that of the actress, so much so that one could confuse them. The company assured that it did not take his vote without his consent. Subsequently, it was discovered that there had indeed been an attempt to dealwhich was unsuccessful.
This orientation was influenced by the film Her by Spike Jonze, in which Scarlett Johansson never appears except in the form of a voice. She plays an artificial intelligence who is given a first name, Samantha, and who is so convincing that the main character (Joaquin Phoenix) ends up falling in love.
Beyond the anger of Scarlett Johansson, the recent trajectory of OpenAI is part of this trend of anthropomorphization (the artificial voice that the American company uses is called Sky, which also serves as a first name in the United States ). And this is obviously not to the liking of Microsoft, which is OpenAI’s closest partner.
Therefore, Satya Nadella’s words must resonate strongly in the ears of the OpenAI teams – and also in those of Sam Altman. The founder and boss of the firm appears to be the trigger for this affair. He contacted the actress and posted a cryptic tweet (“her”) on May 13, the day of GPT-4o.
Satya Nadella, however, recognizes that there is an original sin with AI: we should not say artificial intelligence. “ I think one of the most unfortunate names is artificial intelligence — I would have preferred it to be called different intelligence. Because I have my intelligence. I don’t need artificial intelligence. »
If he did not address the subject, this anthropomorphization could also in the longer term cause legal problems if AI becomes even more efficient and “human”. In recent years, we have seen reflections on the relevance of creating a new legal personality for the benefit of robots/AI.
Such a perspective would pose major challenges (for example, will there be copyright for machines creating something? Today, only humans have the right to copyright) and undoubtedly also new responsibilities for companies who design these systems. Microsoft, which is developing it, would have reason to slow down.
The controversy with Scarlett Johansson does not appear likely to call into question the link established between OpenAI and Microsoft. This, especially since the software giant has not been completely exemplary when it comes to anthropomorphism. Its bots and assistants also received first names, like Murphy, Tay or Cortana.
Well-ordered charity begins with oneself.
Subscribe to Numerama on Google News so you don’t miss any news!