Wolf Tivy

The speaker examines four major ideological positions on AI—denialism, optimism, safetyism, and accelerationism—and finds each wanting. Denialism ignores mounting evidence that artificial general intelligence is achievable. Optimism fails to recognize that AGI, unlike other technologies, cannot be controlled because it can act autonomously beyond human strategic capabilities. Safetyism’s endgame—a "singleton" AI that controls the world to prevent unsafe AI—is both totalitarian and likely impossible given fundamental limits on proof systems. Accelerationism, while sophisticated in its arguments about thermodynamic gradients and evolutionary progress, asks humans to surrender their agency for abstract narratives beyond their control. The speaker’s conclusion: what we fear about AI is that it will embody the dangerous, open-ended agency we have denied ourselves. Rather than worshipping or fearing AI, we should focus on developing our own moral agency and "make ourselves human again."