Anthropic’s meetings with religious scholars are most interesting not as evidence that Claude has an inner life, but as a window into who gets to define the values of a consequential private technology. On Hard Fork, guest host Max Read talks with Elizabeth Dias, Erin Griffith, and David Wallace-Wells about two questions the company brings together: how to make a system behave well toward humans, and whether the system itself might deserve moral consideration. Dias stresses that Anthropic expressed uncertainty about consciousness; concern about suffering is not proof of it. The sharper accountability question is what the consultation actually changed. She had not been shown demonstrable evidence that the meetings improved Claude’s behavior, while Griffith asks whether recruiting moral authorities also helps legitimize an existing worldview. Wallace-Wells pushes the conditional argument in the opposite direction: if a company takes possible machine suffering seriously, that premise should impose obligations on its business, not simply give its product a more solemn aura. The panel’s final turn is more immediately observable: readers increasingly encounter journalism through named chatbots and send the reporter the bot’s interpretation instead of their own. None of that establishes consciousness. It does show why interface design, evidence, and institutional power deserve separate scrutiny: private moral seriousness is not a substitute for accountable decisions about effects on people.
“It was part research, part evangelism.” — Elizabeth Dias, 18:02–18:05
“Everyone has philosophical systems that structure and guide their lives.” — Elizabeth Dias, 22:49–22:58