What would happen if we upload our brains to computers | Robin Hanson

Episode Summary

Executive Summary: Robin Hanson argues that brain emulations (“Ms”) could be the most plausible path to human-level AI and explores what their world would actually look like. He predicts a fast, crowded, hierarchical future where Ms live mostly in virtual reality, work under subsistence pressures, copy themselves, and outcompete humans economically and socially. Hanson’s aim is not inspiration but realistic scenario analysis.

Main Topics: Path to Artificial Intelligence via Brain Emulation (Priority: 5/5): Hanson presents whole-brain emulation as a third route to superintelligent AI, distinct from slow software improvement or breakthrough theories of intelligence. Technologies Required for Emulation (Priority: 5/5): He says three enabling technologies are needed: cheap parallel computing, high-resolution brain scanning, and accurate models of brain cells; none are ready yet. Life in Virtual Reality (Priority: 4/5): Emulations would spend most of their lives in VR, with rich sensory experiences and dashboards for everyday tasks, while their physical form remains server hardware. Economic Pressure and Subsistence (Priority: 5/5): Because M’s need computing resources and population can expand faster than output, wages would fall to subsistence, forcing most M’s to work nearly all the time. Speed, Copying, and Social Structure (Priority: 4/5): Ms can run at variable speeds, duplicate themselves, archive backups, and move at light speed, creating a society stratified by speed and productivity. Humans in the Age of M (Priority: 4/5): Hanson argues humans would likely retire from competition, retain assets if they prepare, and may be tolerated rather than eliminated because institutions discourage predation. Mindset and Interpretation of Strange Futures (Priority: 3/5): He warns that intuitive first reactions to bizarre futures are unreliable and urges listeners to study such scenarios deeply before judging or trying to change them.

Key Arguments: Whole-brain emulation is a plausible route to advanced AI if we can scan brains, model neurons, and run them on sufficient computing hardware. Ms would be behaviorally human-like because they emulate what the original human would do in the same situation. Most M’s lives would be economically constrained: they require real resources and would likely earn subsistence wages. Because M populations can scale faster than their economies, the typical M would work most of the time, despite subjective memories of leisure in some cases. The M world would evolve roughly 100 times faster than human society, with typical emulations running about 1,000 times human speed. Humans would likely not be instantly destroyed; instead, they may be allowed to retire, though they must secure assets beforehand to avoid hardship. Speed becomes a social status marker, with fast M’s gaining influence and slow M’s resembling retirees or “ghosts.” The important question is not whether uploads are conscious, but what their society would actually do under realistic constraints.

Data Points: Technological progress rate: 70 years - Hanson says software progress at historical rates might take centuries to reach human-level AI. M world growth rate: ~100 times faster - He estimates the M world would change around 100x faster than the human world. Typical emulation speed: ~1,000 times human speed - He says the typical M experiences time much faster than humans do. Potential speed range: 1,000,000x faster to 1,000,000,000x slower than human speed - He describes the plausible spread of mind speeds among M’s. Age of M duration: 1 to 2 years - He limits detailed projection beyond this because he expects something else may happen afterward. Working life of short-term copies: thousands of copies / 99%+ of time working - He describes short-term task copies as highly efficient but mostly laboring across many copies. Copying frequency example: 1,000 copies - In the plumbing example, one version continues while 1,000 short-term copies do the work each day. Work fraction: well over 99% - He claims the plumber’s subjective leisure hides that objectively the work is done by many copies almost all the time.

Pivotal Quotes: "I'm not offering inspiration. I'm offering analysis." — Robin Hanson: He frames the talk as a cold, empirical forecast rather than a motivational vision. "M's spend most of their life in virtual reality." — Robin Hanson: He introduces the everyday lived environment of brain emulations. "Humans must retire at once for good. They just can't compete." — Robin Hanson: He summarizes the economic displacement of humans in the emulation era.

Implications: If brain emulations become feasible, the near future may be fast, unequal, and economically brutal. Planning for asset ownership, institutional stability, and post-human competition becomes crucial well before such systems arrive.