I predict Shannon's compression limit would be mathematically achievable by 2028.
I also predict how the technology for testing this limit would bottom out at being available to less than 3% of 3% of empirical researchers. I predict Mahoney would derive an analog solution for the future Shannon Entropic Damping testing problem. This would be a very-big deal. Last, I predict how Homo Sapiens would increase its inverse-scaled development of collective intelligence. On Wed, 27 May 2026, 05:41 Matt Mahoney, <[email protected]> wrote: > When I was 13 (in 1968) I was reading about how neurons represent concepts > and can stimulate or inhibit other neurons, but that memory and learning > was a mystery. I had the idea that synapses could change their state, so > that if both fire together then the synapse changes to excitory, and if the > output doesn't fire then it changes to inhibitory. I was so excited I told > everyone that would listen that I knew how the brain works. Years later > (about 1980) I wrote about my theory for how classical conditioning worked > in my freshman psychology class and got a B because it was off topic. It > wasn't until over a decade later that I learned about Hebb's rule. Not > fair! I wasn't even born in 1949. > > Of course all the LLMs praised your papers for your precience that you > wrote as a teenager criticizing the Turing test. That's what they do. > People reward AI for flattery. But lots of other people criticized it too > for the same reasons you did. It's not like Turing wasn't aware of the > problems in his 1950 paper. He gave an example of the computer taking 30 > seconds to solve an arithmetic problem and give the wrong answer. > > Anyway, I was asking about your predictions for the future, not 200 pages > explaining all the past predictions you got right. I mean, I could point to > my 1999 paper "text compression as a test for artificial intelligence" or > my 2013 paper estimating the cost of AGI at $1 quadrillion. The Turing test > is certainly not the right test for universal intelligence in the sense of > achieving arbitrary goals. But it is the right test for modeling human > behavior, which is the actual reason for building AGI for profit. We don't > want machines that are smart enough to kill us. We just want machines to do > everything we would otherwise need people to do for us. Just predict what I > want by running a copy of my brain and then give it to me. > > I predict that we are a long way from the limits of computation. That in > the far future, nanotechnology will replace DNA based life, build Dyson > spheres, and spread to other stars and galaxies. In the near future, humans > will become socially isolated and stop having children in a world where > machines grant all our wishes except happiness. > > What are your predictions? > > -- Matt Mahoney, [email protected] > > On Tue, May 26, 2026, 9:12 PM twenkid <[email protected]> wrote: > >> There's another form of Singularity, called *"Singularity of Tosh". *It >> has already happened many times in the scale of hundreds or thousands or >> millions and decades in time, but the expected "great singularity" will >> happen when a sufficient implementation of Вседържец/Vsy=Jack of All Trades >> = Master of All Arts = Allmighty/EMIL/VLESI/Vseboravitel/ACS/"the AGI >> infrastructure" are published and get out of stealth. >> (...) >> >> The picture below is an example of ST - from the introduction of *"Stack >> Theory is yet Another Fork of Theory of Universe and Mind"**, 2025 >> >> *Artificial General Intelligence List <https://agi.topicbox.com/latest>* > / AGI / see discussions <https://agi.topicbox.com/groups/agi> + > participants <https://agi.topicbox.com/groups/agi/members> + > delivery options <https://agi.topicbox.com/groups/agi/subscription> > Permalink > <https://agi.topicbox.com/groups/agi/T7daa29d46d037f94-Mbe5216647b34a6db7c121759> > ------------------------------------------ Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/T7daa29d46d037f94-M785c65ab1c472e3db83d427f Delivery options: https://agi.topicbox.com/groups/agi/subscription
