Editing Artificial general intelligence (section)

== Whole brain emulation ==
{{Main|Whole brain emulation|Brain simulation}}

While the development of [[Transformer (deep learning architecture)|transformer]] models like in [[ChatGPT]] is considered the most promising path to AGI,<ref name=":18">{{Cite news |last=Sullivan |first=Mark |date=October 18, 2023 |title=Why everyone seems to disagree on how to define Artificial General Intelligence |url=https://www.fastcompany.com/90968623/why-everyone-seems-to-disagree-on-how-to-define-artificial-general-intelligence |work=Fast Company}}</ref><ref>{{Cite web |last=Nosta |first=John |date=January 5, 2024 |title=The Accelerating Path to Artificial General Intelligence |url=https://www.psychologytoday.com/intl/blog/the-digital-self/202401/the-accelerating-path-to-artificial-general-intelligence |access-date=2024-03-30 |website=Psychology Today |language=en}}</ref> [[whole brain emulation]] can serve as an alternative approach. With whole brain simulation, a brain model is built by [[brain scanning|scanning]] and [[Brain mapping|mapping]] a biological brain in detail, and then copying and simulating it on a computer system or another computational device. The [[computer simulation|simulation]] model must be sufficiently faithful to the original, so that it behaves in practically the same way as the original brain.<ref>{{Cite web |last=Hickey |first=Alex |title=Whole Brain Emulation: A Giant Step for Neuroscience |url=https://www.emergingtechbrew.com/stories/2019/08/15/whole-brain-emulation-giant-step-neuroscience |access-date=2023-11-08 |website=Tech Brew |language=en-us}}</ref> Whole brain emulation is a type of [[brain simulation]] that is discussed in [[computational neuroscience]] and [[neuroinformatics]], and for medical research purposes. It has been discussed in [[artificial intelligence]] research{{Sfn|Goertzel|2007}} as an approach to strong AI. [[Neuroimaging]] technologies that could deliver the necessary detailed understanding are improving rapidly, and [[futurist]] [[Ray Kurzweil]] in the book ''[[The Singularity Is Near]]''<ref name=K/> predicts that a map of sufficient quality will become available on a similar timescale to the computing power required to emulate it.

===Early estimates===
[[File:Estimations of Human Brain Emulation Required Performance.svg|thumb|right|Estimates of how much processing power is needed to emulate a human brain at various levels (from Ray Kurzweil, [[Anders Sandberg]] and [[Nick Bostrom]]), along with the fastest supercomputer from [[TOP500]] mapped by year. Note the logarithmic scale and exponential trendline, which assumes the computational capacity doubles every 1.2 years. Kurzweil believes that mind uploading will be possible at neural simulation, while the Sandberg, Bostrom report is less certain about where [[consciousness]] arises.{{Sfn|Sandberg|Boström|2008}}|upright=2.6]] For low-level brain simulation, a very powerful cluster of computers or GPUs would be required, given the enormous quantity of [[synapses]] within the [[human brain]]. Each of the 10<sup>11</sup> (one hundred billion) [[neurons]] has on average 7,000 synaptic connections (synapses) to other neurons. The brain of a three-year-old child has about 10<sup>15</sup> synapses (1 quadrillion). This number declines with age, stabilizing by adulthood. Estimates vary for an adult, ranging from 10<sup>14</sup> to 5×10<sup>14</sup> synapses (100 to 500 trillion).{{Sfn|Drachman|2005}} An estimate of the brain's processing power, based on a simple switch model for neuron activity, is around 10<sup>14</sup> (100 trillion) synaptic updates per second ([[SUPS]]).{{Sfn|Russell|Norvig|2003}}

In 1997, Kurzweil looked at various estimates for the hardware required to equal the human brain and adopted a figure of 10<sup>16</sup> computations per second (cps).{{Efn|In "Mind Children"{{Sfn|Moravec|1988|page=61}} 10<sup>15</sup> cps is used. More recently, in 1997,{{Sfn|Moravec|1998}} Moravec argued for 10<sup>8</sup> MIPS which would roughly correspond to 10<sup>14</sup> cps. Moravec talks in terms of MIPS, not "cps", which is a non-standard term Kurzweil introduced.}} (For comparison, if a "computation" was equivalent to one "[[FLOPS|floating-point operation]]" – a measure used to rate current [[supercomputer]]s – then 10<sup>16</sup> "computations" would be equivalent to 10 [[Peta-|petaFLOPS]], [[FLOPS#Performance records|achieved in 2011]], while 10<sup>18</sup> was [[Exascale computing|achieved in 2022]].) He used this figure to predict the necessary hardware would be available sometime between 2015 and 2025, if the exponential growth in computer power at the time of writing continued.

===Current research===
The [[Human Brain Project]], an [[European Union|EU]]-funded initiative active from 2013 to 2023, has developed a particularly detailed and publicly accessible [[atlas]] of the human brain.<ref>{{Cite news |last=Holmgaard Mersh |first=Amalie |date=September 15, 2023 |title=Decade-long European research project maps the human brain |url=https://www.euractiv.com/section/health-consumers/news/decade-long-european-research-project-maps-the-human-brain/ |work=euractiv}}</ref> In 2023, researchers from Duke University performed a high-resolution scan of a mouse brain.

===Criticisms of simulation-based approaches===
The [[artificial neuron]] model assumed by Kurzweil and used in many current [[artificial neural network]] implementations is simple compared with [[biological neuron model|biological neurons]]. A brain simulation would likely have to capture the detailed cellular behaviour of biological [[neurons]], presently understood only in broad outline. The overhead introduced by full modeling of the biological, chemical, and physical details of neural behaviour (especially on a molecular scale) would require computational powers several orders of magnitude larger than Kurzweil's estimate. In addition, the estimates do not account for [[glial cells]], which are known to play a role in cognitive processes.<ref name="Discover2011JanFeb">{{Cite journal |last=Swaminathan, Nikhil |date=Jan–Feb 2011 |title=Glia—the other brain cells |url=http://discovermagazine.com/2011/jan-feb/62 |url-status=live |journal=Discover |archive-url=https://web.archive.org/web/20140208071350/http://discovermagazine.com/2011/jan-feb/62 |archive-date=8 February 2014 |access-date=24 January 2014}}</ref>

A fundamental criticism of the simulated brain approach derives from [[embodied cognition]] theory which asserts that human embodiment is an essential aspect of human intelligence and is necessary to ground meaning.<ref>{{Harvnb|de Vega|Glenberg|Graesser|2008}}. A wide range of views in current research, all of which require grounding to some degree</ref><ref name=":15">{{Cite web |last=Thornton |first=Angela |date=2023-06-26 |title=How uploading our minds to a computer might become possible |url=http://theconversation.com/how-uploading-our-minds-to-a-computer-might-become-possible-206804 |access-date=2023-11-08 |website=The Conversation |language=en-US}}</ref> If this theory is correct, any fully functional brain model will need to encompass more than just the neurons (e.g., a robotic body). Goertzel{{Sfn|Goertzel|2007}} proposes virtual embodiment (like in [[metaverse]]s like ''[[Second Life]]'') as an option, but it is unknown whether this would be sufficient.