<div dir="ltr">On Thu, 26 Feb 2026 at 19:26, Jason Resch via extropy-chat <<a href="mailto:extropy-chat@lists.extropy.org">extropy-chat@lists.extropy.org</a>> wrote:<br>> Indeed, the change I've noticed in AI in the past few months feels like today's tools are on a completely different level than they were from a few months ago.<br>><br>> <snip><br>> Jason<br>> _______________________________________________<br><div><br></div><div><br></div>I asked Gemini to discuss the exponential rate of AI improvement.<br>BillK<br><br>Gemini 3.1 Pro - <br><div><br></div><div><div class="gmail-markdown gmail-markdown-main-panel gmail-stronger enable-updated-hr-color gmail-preserve-whitespaces-in-response" style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px" id="gmail-model-response-message-contentr_1b481719cce99c3f" dir="ltr"><p style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px">The article you shared captures a pivotal moment in the "AI timeline." To expand on this, we have to look past the word "fast" and understand the specific mechanics of <b style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px">exponential growth</b>—and why the experts cited in that article (from DeepMind and METR) are behaving as if they are watching a tidal wave approach while the rest of the world is still sunbathing.</p><p style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px">Here is a deeper look at the three pillars of this imminent exponential progress.</p><h3 style="font-family:Google Sans,sans-serif;line-height:1.15;margin-top:0px">1. The Shift from "Chat" to "Time Horizons" (The Agentic Leap)</h3><p style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px">The most significant data point in the article is the <b style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px">METR "Time Horizon" benchmark</b>.</p><ul style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px"><li style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px"><p style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px"><b style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px">The Old Way:</b> We used to measure AI by "accuracy" on short questions (e.g., "Can it pass the Bar Exam?").</p></li><li style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px"><p style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px"><b style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px">The New Way:</b> We now measure AI by <b style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px">autonomous duration</b>. In mid-2024, the best AIs could only "focus" on a task for a few minutes before losing the thread or hallucinating.</p></li><li style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px"><p id="gmail-p-rc_5e5bca6ccd3f48c8-19" style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px"><span class="gmail-citation-9" style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px"></span><b style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px"><span class="gmail-citation-9" style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px">The Exponential Jump:</span></b><span class="gmail-citation-9" style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px"> As of early 2026, models like </span><b style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px"><span class="gmail-citation-9" style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px">Claude Opus 4.6</span></b><span class="gmail-citation-9" style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px"> have hit a "50% time horizon" of </span><b style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px"><span class="gmail-citation-9" style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px">14.5 hours</span></b><span class="gmail-citation-9 gmail-citation-end-9" style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px">.</span> <span class="gmail-citation-8 gmail-citation-end-8" style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px">This means the AI can work autonomously on a messy, complex software project for nearly two full workdays and succeed half the time.</span></p></li></ul><p id="gmail-p-rc_5e5bca6ccd3f48c8-20" style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px"><span class="gmail-citation-7" style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px"></span><b style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px"><span class="gmail-citation-7" style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px"><span class="gmail_default" style="font-family:arial,sans-serif;font-size:small;color:rgb(0,0,0)"></span></span></b></p><p style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px"><b style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px"><span class="gmail-citation-7" style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px"><span class="gmail_default" style="font-family:arial,sans-serif;font-size:small;color:rgb(0,0,0)">1. </span>Why this is exponential:</span></b><span class="gmail-citation-7 gmail-citation-end-7" style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px"> If the doubling rate of ~7 months holds, by 2027, an AI might have a "time horizon" of a month.</span> It won't just be an assistant; it will be an autonomous "digital employee" capable of managing entire projects from start to finish.</p><div class="gmail-source-inline-chip-container gmail-ng-star-inserted" style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px"><button class="gmail-button gmail-ng-star-inserted" aria-label="View source details for citation from Reddit. Opens side panel." style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px"></button></div><p></p><h3 style="font-family:Google Sans,sans-serif;line-height:1.15;margin-top:0px">2. The Measurement Crisis: "Flying Blind"</h3><p id="gmail-p-rc_5e5bca6ccd3f48c8-21" style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px"><span class="gmail-citation-6 gmail-citation-end-6" style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px">The article mentions that researchers are "nervous" about their own charts.<sup class="gmail-superscript" style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px;font-size:16px;background-color:transparent"></sup></span> This is because AI is now <b style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px">outpacing its own yardsticks</b>.</p><div class="gmail-source-inline-chip-container gmail-ng-star-inserted" style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px"><button class="gmail-button gmail-ng-star-inserted" aria-label="View source details for citation from MKFM. Opens side panel." style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px"></button></div><p></p><ul style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px"><li style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px"><p style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px"><b style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px">Benchmark Saturation:</b> For years, tests like MMLU (general knowledge) were the gold standard. AIs have now effectively "maxed out" these tests.</p></li><li style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px"><p style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px"><b style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px">The Confidence Interval Problem:</b> When a model is as capable as Claude 4.6, it’s hard to find tasks that are difficult enough to test its limits. This creates a "measurement fog." We know the curve is steep, but we are losing the ability to see exactly where the top is.</p></li><li style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px"><p style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px"><b style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px">Implication:</b> We might reach Superintelligence (ASI) or AGI (Artificial General Intelligence) before we even have a test capable of proving it exists.</p></li></ul><h3 style="font-family:Google Sans,sans-serif;line-height:1.15;margin-top:0px">3. The "Everything" Moment vs. The "Energy Wall"</h3><p style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px">The discussion of "Nothing, nothing, nothing, everything" refers to the deceptive nature of exponential curves.</p><ul style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px"><li style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px"><p style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px"><b style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px">The Economic Lag:</b> Currently, the economy feels "normal" because a tool that fails 50% of the time is hard to integrate into a bank or a hospital. However, on an exponential curve, the jump from "50% success" to "99% success" happens much faster than the jump from 0% to 50%.</p></li><li style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px"><p style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px"><b style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px">The Energy Bottleneck:</b> This is the only "brake" on the exponential. While the software is getting 2x better every few months, the <b style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px">power grid</b> is not. Data centers are hitting a "Power Wall." Reports from early 2026 show that the electricity required for the next generation of training runs could rival the energy consumption of entire nations.</p></li></ul><h3 style="font-family:Google Sans,sans-serif;line-height:1.15;margin-top:0px">4. The "10x Industrial Revolution" Impact</h3><p style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px">When Demis Hassabis (DeepMind) says AI will have <b style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px">10 times the impact of the Industrial Revolution in 1/10th of the time</b>, he is referring to the <b style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px">Cognitive Surplus</b>.</p><ul style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px"><li style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px"><p style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px">The Industrial Revolution replaced <b style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px">muscle</b>.</p></li><li style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px"><p style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px">The AI Revolution is replacing <b style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px">problem-solving</b>.</p></li></ul><p style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px">If progress is truly exponential, we are entering a "Phase Shift." In the Industrial Revolution, society had 100 years to adapt. If this happens in 10 years, the "chaos" the METR researcher mentioned refers to the fact that our laws, schools, and tax systems are designed for linear change, not a world where the "state of the art" resets every seven months.</p><h3 style="font-family:Google Sans,sans-serif;line-height:1.15;margin-top:0px">Summary: Is the Hype Real?</h3><p id="gmail-p-rc_5e5bca6ccd3f48c8-22" style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px"><span class="gmail-citation-5" style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px">The data from </span><b style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px"><span class="gmail-citation-5" style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px">METR</span></b><span class="gmail-citation-5 gmail-citation-end-5" style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px"> suggests the hype is backed by a very specific, measurable trend line (an R² of 0.93, meaning the trend is statistically "clean").</span> The question isn't whether the progress is happening—it’s whether our physical infrastructure (electricity) and our social structures (jobs/laws) can survive the speed of the "everything" moment.</p><p style="font-family:Google Sans Text,sans-serif;line-height:1.15;margin-top:0px"><span class="gmail_default" style="font-family:arial,sans-serif;font-size:small;color:rgb(0,0,0)">--------------------------</span></p></div></div></div>