"Matt Mahoney’s critique of *Roger Penrose* accurately highlights that classical neural networks can simulate complex human behavior without requiring quantum hardware. However, the statement conflates behavioral simulation with ontological consciousness, overlooking that Penrose argued against the capacity of computational systems to possess true understanding." - Gemini
37 years later, unless an empirical result proves Penrose's "understanding" hypothesis incorrect, it remains standing. Penrose 2026 - quantum mechanics is incomplete. - It still is, but may not be for very much longer. Penrose also argues that true machine knowledge creation is impossible because current machines lack consciousness and genuine understanding. He questions the use of the term "Intelligence" within AI. Is the ultimate test for machine understanding, true knowledge creation? On Tue, 11 Aug 2026, 14:41 Matt Mahoney, <[email protected]> wrote: > Penrose argued in 1989 that AI is not computable because the brain uses > quantum effects that non quantum computers can't model. > https://en.wikipedia.org/wiki/The_Emperor's_New_Mind > > And instead of looking for evidence that his theory was wrong, as any good > scientist should do, he looked for evidence that his theory was right, > which is what everyone actually does. > > And since 2023 we had computers that pass the Turing test, learning the > full range of human emotions, using text prediction on neural networks > running on classical computers. So the neural model was right and Penrose > was wrong. > > -- Matt Mahoney, [email protected] > > > > On Tue, Aug 11, 2026, 6:21 AM Quan Tesla <[email protected]> wrote: > >> Matt. Penrose often hides his original thoughts behind suggestive >> questions. IMO, what he means by this is that a computer cannot duplicate >> human consciousness architecture. I support that view. >> >> I also support your view that extended LLMs could be enabled with a >> relative version of a "consciousness" architecture. >> >> I agree with John's implication that a specific Turing complete machine >> objective is worthy of pursuing. >> >> However, the industry obsession with trying to equal intelligent machines >> to humans is proving counter productive to Western competitive >> advancement. As a Western society, not only as the elite ruling the >> society. >> >> The Pacific-East don't seem to suffer from such delusions. They strive >> purely to optimize machine intelligence reality as interactive enabler of >> human reality. >> >> In the empirical literature this becomes abundantly clear. Starting >> nearly 30 years behind the global AI dev curve, the Pacific-East achieved >> AI dominance within a specific cycle of the last 18 years. >> >> Smart Westerners would rapidly adopt such empirical evidence and catch >> up. Even so, careful curation of AI developmental knowledge by the East >> seems to be intentional, to keep the Western adoptive lag by at least 3 >> years. >> >> The only real competitive option for the West seems to point to adapting >> and jumping the curve as the Pacific-East did. >> >> However, Western socio-economic infrastructure was designed for mistrust >> and name-brand dominent profiteering, not collaboration-based collective >> progress. IMO, self defeating. >> >> It may yet prove easier to develop AI with human-like reasoning and >> machine consciousness, than fixing the human-human collaborative >> infrastructrural chasm in the West. >> >> We all need a "partizan hero". We've been brainwashed to need it. Without >> it, society tends to collapse into chaos. In general, the West don't tend >> to encourage and enable human agency as peer members of a neural network. >> >> On the contrary, organized socialized, neural networks are generally >> viewed as "potential threats", which are actively being neutralized and >> controlled via disruptive and destructive AI-enabled social-media like >> structures and selective access control to critical resources. >> >> By direct contrast, I think the Pacific-East employ their coherent >> collective-interest driven collaborative resources to architect competitive >> socio-technical developmental resilience. >> >> Based on my analysis, for the world this remains a People <=> >> Organization <=> Process synthesis problem, not a Technology problem. Yet, >> we focus on Technology as our champion, as our "hero". >> >> We are an intentionally divided civilization. How can we expect different >> resilience outcomes when we institutionalize the same intentional errors? >> >> Einstein referred to such behavior as madness. Is this civilization >> certifiably mad? >> >> On Tue, 11 Aug 2026, 01:08 Matt Mahoney, <[email protected]> wrote: >> >>> Penrose believed that the human mind does things that are not >>> computable, stuff like consciousness, feelings, intuition, creativity, >>> insight, free will, qualia. To prove it, he looked for quantum effects in >>> the brain that could not be simulated on a non quantum computer. He and >>> Hameroff found quantum superposition in the protein structure of >>> microtubules in neurons and argued that no computer could simulate this. >>> >>> Of course he was wrong. We now have LLMs that pass the Turing test, >>> which means they have all the properties that he argued were not >>> computable. It turns out you don't have to model microtubules to build a >>> neural network, just like you don't have to model quantum tunneling of gate >>> electrons in SSD memories to model the digital logic of a computer. >>> >>> -- Matt Mahoney, [email protected] >>> >>> On Mon, Aug 10, 2026, 9:53 PM John Rose via AGI <[email protected]> >>> wrote: >>> >>>> On Tuesday, July 14, 2026, at 10:38 PM, Quan Tesla wrote: >>>> >>>> Thanks John. I like this a lot. >>>> >>>> The Orch-OR premise would be, when you observe locally, you collapse >>>> the [local] wave function. >>>> >>>> >>>> If I’m correct consciousness is going to be protocol agnostic, able to >>>> be instantiated across substrates. Lowest level >>>> pancomputational/panphsychist substrates might be quantum fields, a qubit >>>> framework, quantum state topologies, Planck level automata, even ether. Was >>>> reading Ken Wheeler on magnetism and ether, very interesting not sure the >>>> state of that lately. Not sure of the most correct contemporary reality >>>> substrate there seems to be many... And I like when intelligence is >>>> described as a physically defined entity for example a field with dynamical >>>> laws in spacetime. Then higher order intelligences and consciousnesses like >>>> human an AI can be described as structures built on a base. Text is mostly >>>> UI and data storage in my opinion. >>>> >>>> *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/T906c878e4ba411be-M228922edf3113d872779099a> > ------------------------------------------ Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/T906c878e4ba411be-M637b7311e02b0e5967345707 Delivery options: https://agi.topicbox.com/groups/agi/subscription
