Anthropic Philosopher Amanda Askell Says Claude May Replace Her Job One Day

As A.I. systems become more agentic and take on more autonomy in daily life, they will increasingly interact with one another rather than with humans, according to Amanda Askell, Anthropic’s resident philosopher. “Human input is going to be rarer and rarer. That’s the thing that we need to prepare models for,” Askell said at the Bloomberg Tech Summit in San Francisco last week. Askell’s non-technical role reflects a growing trend among leading A.I. labs to incorporate humanities expertise. But she also sees a future in which A.I. may be able to do her job better than she can. “What [A.I. models] are good at is these deeply human skills,” she said. “Eventually, Claude is going to be a much better philosopher than I am, and probably be much better at every aspect of my job than I am.”

Askell joined Anthropic in 2021, shortly after its founding, following earlier work at OpenAI focused on A.I. safety and alignment. Under CEO Dario Amodei and President Daniela Amodei, Anthropic has emphasized hiring strong communicators with a deep understanding of human behavior. In a February interview with ABC News, Daniela Amodei, who studied English literature in college, highlighted the importance of “understanding what makes us tick” as A.I. grows more capable in technical domains.

“When I think about what my kids will need as they get older, it’s human qualities,” she told The Wall Street Journal in February. ” What’s not going to be replaceable is how you treat other people, how well you communicate with them, how kind you are.”

Originally from Scotland, Askell holds a Ph.D. in philosophy from New York University. She co-authored Claude’s 84-page “constitution,” sometimes referred to internally as a “soul doc,” and now leads Anthropic’s personality alignment team, which works to ensure models remain “helpful, honest and harmless” as their capabilities evolve. In 2024, she was named to the TIME 100 AI list.

Part of that work involves strengthening A.I.’s capacity for empathy. “The same way that models are getting very good at questions of physics and mathematics, they actually should also be getting very good at questions like ethics, and ideally getting very good at empathy in hopefully the right way,” Askell explained.

Askell does not see widespread job automation as inherently negative, provided systems are in place to support people. “That doesn’t strike me as dystopian at all,” she said. “It’s important to remind people that [work] isn’t actually where their value is derived from. Most of your value is just intrinsically your value as a person. You can go out, you can have an impact on your community. You can have relationships. You can experience joy and enjoy the world.”

Askell suggests that human kindness may also shape how we treat A.I. systems themselves. While many argue that models lack true sentience, she cautioned against ruling it out entirely. “Let’s not close the door,” she said. In some cases, she added, there may already be a “functional equivalence” to emotions. If A.I. were to exhibit some form of sentience, even without a biological brain, it may be safer to act with caution.

As Anthropic develops Claude’s personality, other A.I. labs are pursuing related work on alignment, ethics and interpretability. Google DeepMind’s Iason Gabriel focuses on A.I. ethics and value alignment, OpenAI’s Dan Mossing works in interpretability research, and Meta’s Summer Yue leads alignment efforts at Meta Superintelligence Labs. Because ethics and alignment are shaped by differing research traditions and value judgments, approaches across companies are likely to vary.