Critique of Anthropic's AI Model Naming Convention: 'Mythos' vs. 'Logos'

The Naming of Anthropic's AI Model

Anthropic, an AI safety company, named its most powerful AI model 'Mythos.' The author questions this choice, given the company's stated mission to be truthful about AI development. The term 'mythos' originates from ancient Greek, meaning a story that conveys meaning beyond the literal, explaining origins and significance rather than providing verifiable facts.

Mythos vs. Logos

In ancient Greek thought, 'mythos' was often contrasted with 'logos.' 'Logos' represents reasoned argument, evidence, and systematic explanation, making claims that can be interrogated and verified. 'Mythos,' conversely, involves claims that are received and inhabited by a community, providing a framework for understanding without requiring verification. The author argues that by choosing 'Mythos,' Anthropic is implicitly instructing users to engage with the model in a 'mythos-mode' rather than a 'logos-mode,' discouraging critical examination of its safety evaluations or benchmarks.

Leadership's Language

The author points to statements made by Dario, a leader at Anthropic, as further evidence of this 'mythos-mode' approach. Dario's quote, "I believe we are entering a rite of passage, both turbulent and inevitable, which will test who we are as a species," is characterized as the language of mythos, not that of a scientist or engineer. This reinforces the author's concern that the company is leaning into narrative and belief rather than empirical evidence and verification.

Historical Precedent

The article draws a parallel to historical figures like Augustus, who cultivated myths to consolidate power. Augustus, while not declaring himself a god, allowed eastern provinces to worship him and commissioned works like the Aeneid to establish a myth of Roman greatness. This historical context is used to suggest that the use of 'mythos' can be a tool of manipulation by those in power.