AGI Yes, ASI No: Why One Author Argues Humanity Must Ban Superintelligence
JOBhakdi · x · 2026-09-27
The author draws a sharp line between AGI and ASI: human-equivalent AGI (±50% IQ), even in the billions or trillions, still depends on human funding and organizations, keeping the power balance controllable—humans could organize against or divide allied rogue agents. ASI, by contrast, removes all of those levers, so humanity would be "super toast" very quickly.
He predicts the future converges on prohibiting super-high-IQ systems and treating ASI development as the gravest possible offense. Policing is feasible because ASI training requires super-large compute clusters—an identifiable target. From the 2030s, the race shifts to keeping vast numbers of AGI agents (plus malicious humans) from building unaligned superintelligence by other routes.
He also argues no foreseeable human problem truly requires superintelligence, while compute buildout must continue at maximum speed for inference and testing needs.
Related event: Debate Rages Over ASI Existential Risk Versus Tolerable AGI(4 posts)→
More from AGI Musings
- Cybersecurity voices push back: defenses can contain AI even if it's great at zero-days — basedjensen · 2026-09-27
- David Patterson: AI has no inherent survival drive, only the objective you give it — davidpattersonx · 2026-09-27
- Gary Marcus vs Paul Graham: should AI firms pause agents over training data? — GaryMarcus · 2026-09-27
- Stochastic Strawmen: Can LLM Critics Hold a Consistent Argument for More Than a Year? — Zergylord · 2026-09-27
- Radiologist's blog helps young doctors pick residencies as AI reshapes medicine — DrDatta_AIIMS · 2026-09-27
- David Patterson pushes back on ASI power fears: humans have no reason to cede control — davidpattersonx · 2026-09-27