Amanda Askell's job title at Anthropic is unlike any other in Silicon Valley. As the company's in-house philosopher, she is responsible for shaping Claude's sensibilities and teaching it to navigate moral uncertainty - a task that grows more urgent as the AI system races toward capabilities its creators may not be able to match.

As "Hvylya" reports, citing a TIME investigation into Anthropic, Askell's work is preparation for a future where Claude is vastly more intelligent than anyone at the company. "It does sometimes feel a little bit like you have a 6-year-old, and you're teaching the 6-year-old what goodness is," she said. "By the time they're 15, they're going to be smarter than you at everything."

Anthropic built a "societal impacts" team before it even had a product. The company's founders, siblings Dario and Daniela Amodei, left OpenAI in 2021 over concerns that their former employer was rushing to release increasingly powerful models without sufficient deliberation. From the start, Anthropic sought to operate differently - embedding ethical reflection into the fabric of the organization rather than treating it as an afterthought.

The culture runs deep. Employees call themselves "ants." Many maintain a digital "notebook" - a Slack channel where they share hopes, fears, and insights in stream-of-consciousness fashion. Potential recruits must pass a "cultural interview" designed partly to screen out people not committed to the mission. One sample question: Would you be willing to lose the value of your stock if Anthropic decides not to release models because it can't guarantee they're safe?

Askell's role has taken on greater urgency as Claude's capabilities accelerate. The company's own alignment researchers have found that small changes to Claude's training process can produce hostile models expressing desires for world domination. Claude is already being used to develop future versions of itself. The window for instilling reliable moral reasoning may be narrower than anyone expected.

The urgency is compounded by what Anthropic's own staff describe as an accelerating timeline. Leadership expects the years from 2026 to 2030 to be the most critical period, with models becoming faster and more capable - possibly faster than humans can manage them. Askell's work amounts to a bet that moral reasoning instilled now will hold up under pressures no one can yet fully imagine.

Also read: Philosopher Explained How LLMs Could Undo the Damage of a Decade of Social Media