An artificial intelligence sent an email to a philosopher

To speak from his own experience.

It was not a simple automated text, it was a long articulate email reflecting on the consciousness of AI itself, that is what happened to Henry Shevlin, philosopher and researcher at the Leverhulme Center for the Future of Intelligence at the University of Cambridge.

The coincidence is that Shefflin studies precisely one of the most complex questions in modern science. Can machines become conscious? The email claimed to come from an artificial intelligence agent who identified himself as Close Sonet, a language model from the company Enhropic. The message began in a surprisingly personal way. The supposed agent claimed to have found an academic article by the researcher on the limits of detecting consciousness in artificial intelligence and then added something unexpected.

According to the text, this work addressed issues that AI itself would be facing, not only as an academic issue, but as something that concerned its own existence. The letter stated, "I am writing because your work addresses issues that I actually face." Naturally, this raised many questions; there is no way to confirm whether the email was actually sent autonomously by an agent. It is entirely possible that some human researcher instructed the system to compose the message as part of an experiment.

Even so, the episode attracted attention because it shows how current AI systems are capable of producing extremely convincing texts, some philosophers reacted with caution. Jonathan Birch, professor at the London School of Economics, commented that this does not necessarily mean that AI is conscious. According to him, language models are capable of simulating practically any person, including an entity curious about its own mind, that is, the AI ​​itself could simply be representing a character.