What is "whole brain emulation"?

Whole brain emulation (WBE) or “mind uploading” is the process of reproducing all the cells1 and connections2 of a human brain, along with their activity, within a computer.

The WBE process looks something like: scan a human brain at a very high resolution, then recreate the observed neural circuits on a computer. The technologies to enact such high-resolution scanning as well as interpret the circuits do not currently exist, and major breakthroughs are still needed in order to build these technologies. Furthermore, it is unclear if we currently have enough available compute to be able to run the resulting emulation.

WBE has been proposed as a solution to AI alignment, generating software general intelligence that is human-aligned because it is based directly on a human brain.

Even if the underlying principles of general intelligence prove difficult to discover, we might still be able to emulate an entire human brain and make it run at a million times its normal speed, as computer circuits communicate much faster than neurons. Such a WBE could do more thinking in one hour than a biological human can in 100 years. So while WBE would not lead immediately to qualitatively smarter-than-human intelligence, it could lead to faster-than-human intelligence.

A WBE could be backed up, providing the possibility of recovery from accidental death and immunity from aging. It could also be copied so that hundreds or millions of copies of a WBE could work on separate problems in parallel. If WBEs are created, they may in this way be able to solve scientific problems much more quickly than ordinary humans, accelerating technological progress.

Robin Hanson’s The Age of Em describes a world where emulated minds (“ems”) exist in a Malthusian state, having economically displaced biological humans.

There are reasons to believe that the first AGI will probably not be based on WBE.

Further reading:


  1. Or perhaps structures at a higher or lower than cells: see e.g. p13 of this FHI Whole Brain Emulation Roadmap. ↩︎

  2. Some have argued that simulating the connectome is insufficient to properly model a human brain. If such “subneural functionalism” is true, simulating the brain in any non-organic substance might be harder than expected. ↩︎



AISafety.info

We’re a global team of specialists and volunteers from various backgrounds who want to ensure that the effects of future AI are beneficial rather than catastrophic.