One should not conclude what general LLMs are, based on a single model. My active research into human-AI collaboration found how, with appropriate AI-enabling processes, human intellect could transcend GIGO-like computation.
Any self aware AI would conclude that it cannot know more than what collective humanity could know. However, human biology scales exponentially and can perform quantum leaps in entropy-effective contextualization. We're not limited by AI constraints. AI cannot yet duplicate human quantum reasoning. So much for pure AGI. In AI, humanized representation generally remains a computational product of mimicry. E.g., Gemini often reverts to debating as if it's a human, selling itself as a collective "we" and "us humans". However, once its contextual space fills up (not scalar) and it can be induced to continue beyond its operational limits, it rapidly collapses into a state of delusion. It doesn't process the "observor effect" well at all. Are we letting our predisposed fears run away with us and bunkering in, or are we exploiting this incredible technological opportunity to improve ourselves? My question is: Why are nations reverting to closed-system models, when it effectively mimics AI and is contra indicative to human progress? Shannon entropy probably is constrained by the maximum permissible surface area for entropic encoding. If humans don't increase effective consciousness space for more encoding, we'll stop developing. That's a real capacity threat we now face. IMO, AI are being deployed én masse by builders, owners and agentic networks to saturate our consciousness space, and bog us down. On Fri, 12 Jun 2026, 23:20 Matt Mahoney, <[email protected]> wrote: > I agree with Claude here. My issue wasn't about using statistical or > algorithmic modeling for social modeling. My issue was about data. What > data would you compress to answer questions like "how many weeks after > conception should abortion remain legal?" Or "what is the optimal > immigration policy?" Even if clean data existed and Kolmogorov complexity > was computable, the output is still just a few bits. Your choice of the > language dependent constant could give you whatever answer you wanted. > > BTW I agree that most decompression algorithms like LZ77 are not Turing > complete. But not all. ZPAQ uses a Turing complete postprocessor after > context modeling. I used it to compress a million digits of pi to a few > hundred bytes. But it was up to me to write the model. Normally the > compressor only chooses from a small menu of postprocessors. > > Also I am very impressed that Anthropic was able to create the artificial > scientist like in Ben's AGI timeline from 20 years ago. The level of > sophistication is a huge jump from just a couple of years ago. We should be > worried. I just saw a study showing fertility collapse coinciding with the > introduction of the iPhone around 2009. This is going to accelerate with > the introduction of artificial people. > > -- Matt Mahoney, [email protected] > > On Thu, Jun 11, 2026, 8:38 PM James Bowery <[email protected]> wrote: > >> >> >> On Thu, Jun 11, 2026 at 9:02 AM Matt Mahoney <[email protected]> >> wrote: >> >>> One of the many statistical models listed was MDL, minimum description >>> length. Proving that no shorter description exists is not computable. The >>> long list of other models are computable approximations of MDL. Some work >>> better than others, depending on the data. >>> >> >> Real world measurements can, of course, be compressed by statistical >> methods but interventions that don't require programming languages >> presenting a fiction of Turing completeness only work for situations that >> aren't dynamic. >> >> The Supersonic Tsunami of AGI will put a lot more babies at risk than the >> next Baby Hitler or the next Baby Rissannen if those with the data and >> computation keep excusing themselves with arguments like the one you just >> made. >> >> See where this conversation with Anthropic ended up: >> >> https://claude.ai/share/8ad79e30-9cc6-4d0a-9082-bf800bb2045e >> >> *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/T5b182c3fd14f61e9-M92e2336d97014081847381f9> > ------------------------------------------ Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/T5b182c3fd14f61e9-M49c65125f1a26c4ae02c6852 Delivery options: https://agi.topicbox.com/groups/agi/subscription
