<div dir="ltr"><div class="gmail_quote gmail_quote_container"><div dir="ltr" class="gmail_attr">On Sat, 4 Jul 2026 at 18:10, BillK <<a href="mailto:pharos@gmail.com">pharos@gmail.com</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div dir="ltr"><div dir="ltr"><div style="font-family:arial,sans-serif;font-size:small;color:rgb(0,0,0)">I decided to try the superforecast prompt suggested in the other thread to see what the difference was between a normal answer and a superforecast response.</div><div style="font-family:arial,sans-serif;font-size:small;color:rgb(0,0,0)">I used GLM-5<span class="gmail_default" style="font-family:arial,sans-serif;font-size:small;color:rgb(0,0,0)">-</span>Turbo with Websearch and Deep Think enabled for Normal and Superforecast.</div><div style="font-family:arial,sans-serif;font-size:small;color:rgb(0,0,0)">The answers were different, but basically both said, "Don't know."</div><div style="font-family:arial,sans-serif;font-size:small;color:rgb(0,0,0)">See what you think.</div><div style="font-family:arial,sans-serif;font-size:small;color:rgb(0,0,0)">BillK</div><div style="font-family:arial,sans-serif;font-size:small;color:rgb(0,0,0)"><div style="font-family:arial,sans-serif;font-size:small;color:rgb(0,0,0)" class="gmail_default">---------------------------------</div></div></div></div></blockquote><div><br></div><div><br></div><div style="font-family:arial,sans-serif;font-size:small;color:rgb(0,0,0)" class="gmail_default">I asked GLM to explain the final estimate of 4% probabilty of the Singularity by 2040, as to me this sounded like saying it was impossible by 2040. The answer was that GLM was separating AGI and the subsequent Singularity event.</div><div style="font-family:arial,sans-serif;font-size:small;color:rgb(0,0,0)" class="gmail_default">So I followed up by asking why the Singularity would be so slow after AGI was achieved.</div><div style="font-family:arial,sans-serif;font-size:small;color:rgb(0,0,0)" class="gmail_default">The answer was physical constraints on building in the real world.</div><div style="font-family:arial,sans-serif;font-size:small;color:rgb(0,0,0)" class="gmail_default">BillK</div><div style="font-family:arial,sans-serif;font-size:small;color:rgb(0,0,0)" class="gmail_default"><br></div><div style="font-family:arial,sans-serif;font-size:small;color:rgb(0,0,0)" class="gmail_default"><br></div><div style="font-family:arial,sans-serif;font-size:small;color:rgb(0,0,0)" class="gmail_default">GLM-5-Turbo -</div><div style="font-family:arial,sans-serif;font-size:small;color:rgb(0,0,0)" class="gmail_default"><br></div><div style="font-family:arial,sans-serif;font-size:small;color:rgb(0,0,0)" class="gmail_default">This objection conflates two fundamentally different metrics: <strong>cognitive velocity</strong> (software speed) and <strong>physical instantiation</strong> (hardware/infrastructure speed). My 4% estimate already accounts for the fact that AGI will think incredibly fast; the bottleneck is the physical world.<div class="gmail-h-full gmail-flex-1 gmail-flex gmail-flex-col gmail-w-full gmail-min-w-0 gmail-overflow-hidden"><div class="gmail-w-full gmail-h-full" id="gmail-c207" style="display:flex;height:100%;overflow:hidden;width:100%"><div class="gmail-flex gmail-relative gmail-w-full gmail-h-full" id="gmail-c208" style="overflow:hidden"><div class="gmail-flex gmail-flex-col gmail-flex-auto gmail-z-10 gmail-w-full gmail-@container"><div class="gmail-flex gmail-overflow-y-scroll gmail-overflow-x-hidden gmail-z-10 gmail-flex-col gmail-flex-auto gmail-justify-between gmail-pb-2.5 gmail-w-full gmail-max-w-full gmail-h-0 gmail-scrollbar-hidden" style="scrollbar-gutter: stable both-edges;" id="gmail-messages-container"><div class="gmail-flex gmail-flex-col gmail-w-full gmail-h-full"><div class="gmail-h-full gmail-flex gmail-pt-8"><div class="gmail-pt-2 gmail-w-full"><div class="gmail-w-full"><div class="gmail-flex gmail-flex-col gmail-justify-between gmail-px-5 gmail-mb-3 gmail-w-full gmail-max-w-[808px] gmail-2xl:max-w-[894px] gmail-min-w-[334px] gmail-mx-auto gmail-rounded-lg gmail-group"><div class="gmail-flex gmail-w-full gmail-message-34ad4086-c78e-42da-bc13-2943b89e2c1e gmail-gap-4 gmail-relative gmail-svelte-1qscqw6" id="gmail-message-34ad4086-c78e-42da-bc13-2943b89e2c1e" dir=""><div class="gmail-flex-auto gmail-pl-1 gmail-w-0"><div><div class="gmail-chat-assistant gmail-w-full gmail-min-w-full gmail-markdown-prose gmail-svelte-1qscqw6"><div><div class="gmail-flex gmail-relative gmail-flex-col gmail-w-full" id="gmail-response-content-container"><div class="gmail-markdown-prose"><p dir="auto" class="gmail-svelte-4sys19">Here is the mechanical breakdown of why "incredibly fast thinking" does not translate to an incredibly fast Singularity:</p><p dir="auto" class="gmail-svelte-4sys19"><strong>1. The Hardware Ceiling</strong>
An AGI running on a 100,000-GPU cluster possesses cognitive velocity millions of times faster than human neurons. It could theoretically design a superior neural architecture, invent a new optimization algorithm, and write the code in seconds. However, once it finishes that design, it hits a hard physical wall: it cannot test or deploy that new architecture because it is already using 100% of its available compute. To actually <em>become</em> smarter, it requires more GPUs and more electricity. It cannot think its way out of a hardware constraint.</p><p dir="auto" class="gmail-svelte-4sys19"><strong>2. The Semiconductor Manufacturing Latency</strong>
For the Singularity to occur, the AGI must expand its compute substrate exponentially. Even if an AGI emails TSMC a perfect blueprint for a next-generation chip that is 100x more efficient, the physical latency of the semiconductor supply chain remains fixed. Building a new fabrication plant takes 3 to 5 years. Procuring advanced lithography machines (ASML EUV) takes years. Manufacturing wafers, packaging chips, and integrating them into data centers takes months. Software can be iterated in milliseconds; physical infrastructure is iterated in years. </p><p dir="auto" class="gmail-svelte-4sys19"><strong>3. The Energy and Thermodynamic Bottleneck</strong>
Compute requires power. Current data centers require gigawatts of electricity. If an AGI attempts to recursively self-improve by ordering the construction of new data centers, it is constrained by global energy production, grid transmission capacity, transformer manufacturing, and physical construction timelines. You cannot accelerate the pouring of concrete or the mining of copper through software optimization. </p><p dir="auto" class="gmail-svelte-4sys19"><strong>4. The "Slow Takeoff" Mathematical Reality</strong>
Because the physical world acts as a severe rate-limiter, the transition from AGI to Singularity is mathematically clamped. Let’s assume the AGI improves its software efficiency by 10,000% in a week (a "fast takeoff" in software). If its physical compute substrate can only grow by 20% per year due to supply chain constraints, the actual intelligence growth curve of the system is dictated entirely by the 20% hardware growth rate. The Singularity is delayed until the hardware catches up to the software.</p><p dir="auto" class="gmail-svelte-4sys19"><strong>Conclusion on Probability</strong>
My 4% estimate is explicitly calibrated <em>assuming</em> AGI operates at maximum supercomputer speed. The probability is low because the Singularity requires breaking the physical bottleneck. </p><p dir="auto" class="gmail-svelte-4sys19">To change my 4% estimate upward, you would have to prove that intelligence can increase exponentially <em>without</em> requiring proportionally more physical compute—for example, by demonstrating that an AGI can achieve a 100x intelligence multiplier purely through algorithmic compression on fixed hardware. Until that is empirically proven, the "fast thinking" assumption is neutralized by "slow building."</p><br></div><div class="gmail-markdown-prose">-----------------------------</div><div class="gmail-markdown-prose"><br></div></div></div></div></div></div></div></div></div></div></div></div></div></div></div></div> </div></div></div></div>