A Stark Warning From a Former AI Doomer

What It Covers

Cal Newport interviews former AI doomer Emile Torres, a researcher who spent a decade inside transhumanist and effective altruist communities before defecting. Torres decodes the TESCREAL ideology bundle driving Anthropic and OpenAI, explaining how these labs pursue a posthuman digital utopia that explicitly excludes biological humans from the future they are building.

Key Questions Answered

• •TESCREAL Decoded: The acronym represents seven overlapping ideologies — Transhumanism, Extropianism, Singularitarianism, Cosmism, Rationalism, Effective Altruism, and Long-termism — that form Silicon Valley's dominant worldview. Leaders including Sam Altman, Dario Amodei, Elon Musk, and Mark Zuckerberg operate within this framework. Understanding transhumanism as the foundation makes all other ideologies easier to interpret, since each is essentially a variant emphasizing different aspects of the same posthuman cosmic vision.
• •Pro-Extinctionist Core: Every major TESCREAL faction shares a goal where biological humans disappear. In the "good" aligned-AI scenario, humans become obsolete because posthuman successors outperform them in every domain, then consume resources needed for cosmic expansion. Torres distinguishes terminal extinction (species disappears) from final extinction (no successors remain). TESCREAL advocates only oppose final extinction — many explicitly endorse terminal extinction as the desirable transition to a posthuman digital civilization.
• •The Internal Schism Is Misleading: Public feuding between AI safety advocates and accelerationists creates a false impression of broad opposition to frontier labs. Both camps share identical end goals — posthuman cosmic utopia via superintelligence — and only dispute timing and alignment methodology. Congressional staffers and policymakers risk misreading this as external criticism when it is entirely an internal family dispute within Team Posthuman, leaving Team Human effectively unrepresented in the policy debate.
• •Terminology Conceals Radical Positions: Words like "safety," "human extinction," and "human values" carry definitions inside TESCREAL that differ sharply from common usage. Safety means aligning AI to TESCREAL's worldview specifically. Human values means TESCREAL values. An AI aligned to any non-TESCREAL framework is classified as an existential catastrophe. When Anthropic employees cite a 10% human extinction probability, they are expressing concern about losing their utopian trajectory, not straightforward concern for preserving biological humanity.
• •Self-Fulfilling Prophecy Dynamic: TESCREAL advocates did not predict the AI race — they created it. DeepMind received early funding after cofounders Shane Legg and Demis Hassabis spoke at Yudkowsky's Singularity Summit, then secured Peter Thiel investment. OpenAI formed specifically in response to DeepMind's perceived dominance. The network of funding, conferences, and personnel is entirely internal to TESCREAL. Crediting these figures with prescience about AI risk is equivalent to praising someone for accurately predicting they would make their own coffee.
• •Utilitarianism Enables Extreme Conclusions: Long-termism's total utilitarian foundation — maximizing value across all future digital beings estimated at 10^58 — mathematically justifies present harm at enormous scale. Yudkowsky has publicly stated he would accept global thermonuclear war killing billions if it prevented misaligned superintelligence, since minimum viable human population estimates run as low as 40,000 survivors. This reasoning mirrors twentieth-century utopian movements — Nazi and Soviet — where pursuit of an imagined future justified mass death, and none reached their stated utopia.