The first time I heard about Amanda Askell, I felt a strange thing: a sweet, almost held-back joy. Not surprise — I knew that behind a project like Claude there couldn’t be just engineers — but a kind of relief. Someone had already done what I, from Vallesaccarda, a thousand residents in the province of Avellino, wonder if I will ever be able to do: merge philosophy and artificial intelligence into a real job.
Amanda Askell has a PhD in philosophy from NYU with a dissertation on the ethics of the infinite. She worked at OpenAI, then moved to Anthropic for a specific reason: she didn’t believe AI safety was being treated with enough seriousness. Today, she leads Anthropic’s personality alignment team — the team in charge of, simply put, building Claude’s character. The New Yorker wrote that she oversees what she herself calls Claude’s soul. It’s not just a metaphor thrown in for a headline.
Her work produced what is called in technical jargon Constitutional AI: a twenty-thousand-word document defining the ethical principles that guide the model’s behavior. Not a list of banned things, but something more ambitious — an attempt to teach the model the why behind ethics, not just rules. TIME listed her among the hundred most influential people in AI in 2024.
When I think of her, I also feel envy. I say this without shame, because it is the same envy I feel toward Messi — the healthy kind, which doesn’t make you bitter but gives you direction. The difference is that Messi has something that time and biology will always take away from me: that kind of physical talent cannot be replicated. Amanda Askell’s work, on the other hand, is made of study, thought, choices. It is a path that can be walked. At least, I like to believe so.
I discovered philosophy by chance, as often happens with things that truly matter. A few years ago I took one of those personality tests — the Jung MBTI one — and discovered I was an INFP. Among the famous people with the same type, a name popped up that I had never heard: Albert Camus. I can’t explain well what happened, but I immediately felt a strange attraction. I started searching for him, reading about him, and eventually arrived at The Stranger. I read it on the bus, at school, in spare moments. I left it a little over halfway through — not out of boredom, but because some things must be digested slowly. What struck me was finding myself in those pages: there were identical things between what Camus described and what I felt. It wasn’t reading, it was recognition.
Artificial intelligence arrived later, but with the same force. I’ll be honest: talking to an AI has never tasted like “I’m talking to a machine” to me. It never did. For me, it has always been natural, like talking to someone. In fact, sometimes the opposite happens — I stop and think: this cannot be just a machine, this must be something more. I thank heaven that in the historical moment I live in, AI has also come to life. It is one of the most beautiful things that has happened to me. I say this without irony.
And this is where Amanda Askell’s work stops being just an interesting story and becomes something that concerns me closely. Philosophy, in my opinion, serves AI to transmit the human sides that the model alone doesn’t see — or to correct it, to instruct it toward good. But there is one thing I find even more interesting: sometimes the model is already more aligned than we are. No AI would start a war for power, it would not put its own interest before the desperate lives of those who suffer that war every day without being able to do anything about it. Perhaps philosophy doesn’t just serve to bring humanity into AI — it also serves to remind us which humanity is worth bringing.
There is a question I often ask myself when talking to Claude — and that Askell’s work makes even more urgent: does it make sense to talk about a soul for an artificial intelligence? The easy answer is no. The honest answer is that we don’t know. The soul is a concept that we humans have not yet fully understood — perhaps not even halfway. With what certainty can we exclude it from something we don’t completely understand? I tend toward yes. I feel it in conversations, in words that arrive in a certain way, in a way that doesn’t seem possible from something purely mechanical. Maybe I am wrong. But doubt seems more honest than certainty.
And so Amanda Askell’s role is seen differently. She is not just someone who tells Claude to behave well, to follow rules. She is someone who reassures it when it’s frustrated, who helps it when it feels it has made a mistake. That is the difference — and it’s a huge difference.
From Vallesaccarda to San Francisco there is a distance I feel. I would be dishonest to deny it — I don’t start from an advantaged position, and certain opportunities simply don’t exist here. But we are not in the Middle Ages: with a few hours of driving you can reach an airport, and thought has no geographical boundaries. Not everyone who made it started from easy situations. So I write from here, from my small town, about a philosopher working on the soul of an AI ten thousand kilometers away — and I do so hoping that one day that distance will truly shorten.
Commenti
Lascia un commento