<div dir="auto"><div><br><br><div class="gmail_quote gmail_quote_container"><div dir="ltr" class="gmail_attr">On Thu, Feb 12, 2026, 5:57 PM Adrian Tymes via extropy-chat <<a href="mailto:extropy-chat@lists.extropy.org">extropy-chat@lists.extropy.org</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">> Gemini 3 -<br>
<br>
> The "February 5th" Shift: Shumer points to the recent release of models like GPT-5.3 Codex and Anthropic’s Opus 4.6 as a turning point. He claims these models no longer just assist with work but can execute complex, multi-day technical projects autonomously, demonstrating "judgment" and "taste" that were previously thought to be uniquely human.<br>
<br>
Which has literally been said of prior releases.<br></blockquote></div></div><div dir="auto"><br></div><div dir="auto">The lastest AI from anthropic was used to program a 100,000-lines-of+code C compiler that can compile the Linux kernel on 3 platforms, without one line of code being written by a human. This is a paradigm shift in capability.</div><div dir="auto"><br></div><div dir="auto"><div class="gmail_quote gmail_quote_container"><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<br>
Including running multi-day technical projects autonomously (to take a<br>
recent but running-since-before-Feb-5 example, Moltbook; I am<br>
personally aware of capability that could be described that way going<br>
back to at least 2023), and demonstrations of taste and judgment that<br>
were previously thought to be uniquely human.<br>
<br>
> Recursive Self-Improvement: A critical development is that AI is now instrumental in building its own next generation. By automating the coding and debugging of its own training runs, the "feedback loop" of intelligence is accelerating at an exponential rate.<br>
<br>
The thing about accelerating at an exponential rate is, at any given<br>
time the acceleration is faster than it has ever been before...and<br>
it's still not "the" moment, because future acceleration is even<br>
faster.<br></blockquote></div></div><div dir="auto"><br></div><div dir="auto">I learned recently that technological growth doesn't follow an exponential curve, but rather a hyperbolic one.</div><div dir="auto"><br></div><div dir="auto">For an exponential curve, the slope doubles for each fixed unit of time. For a hyperbolic curve, the time between each doubling halves. One shoots towards infinity in finite time, the other takes infinite time to reach infinity.</div><div dir="auto"><br></div><div dir="auto">Using the observed rates of historic change (even using data from the many decades ago), showed we would reach a singularity point in 2027.</div><div dir="auto"></div><div dir="auto"><br></div><div dir="auto">What will the trigger be? AI improving its own software? Robots building robots? Hard to say, but the 2027 timing seems accurate.</div><div dir="auto"><br></div><div dir="auto"><div class="gmail_quote gmail_quote_container"><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<br>
> The End of Knowledge Work as We Know It: Shumer warns that white-collar sectors—law, finance, medicine, and engineering—are at the precipice of massive disruption. He cites Anthropic CEO Dario Amodei’s prediction that 50% of entry-level white-collar jobs could vanish within 1 to 5 years.<br>
<br>
People will find other things to do, not remain unemployed forever.<br>
See the history of every such disruption ever.<br></blockquote></div></div><div dir="auto"><br></div><div dir="auto">Consider the history of horses after the automobile.</div><div dir="auto"><br></div><div dir="auto">If intelligence and ingenuity is what allowed us to adapt in the last, how do we adapt when artificial intelligence surpasses humans, and can work cheaper, more reliably, with fewer errors, and complaints, etc.?</div><div dir="auto"><br></div><div dir="auto">Humans may still trade their time with other humans human and track that using human currencies, but humans will represent an ever diminishing fraction of the productive economy.</div><div dir="auto"><br></div><div dir="auto">If we're lucky we can look forward to a future like that in the Culture Series, where humans were free to pursue hobbies and interests to their content, but no one needs to work to survive.</div><div dir="auto"><br></div><div dir="auto"><div class="gmail_quote gmail_quote_container"><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<br>
> The "Capability Gap": There is a dangerous divide between those using free/outdated AI models and those using the latest paid versions. Those who dismiss AI usually do so based on 2024-era experiences, failing to realize how much the technology has evolved in just the last few months.<br>
<br>
See also the gap between those using even free/outdated AI models, and<br>
those who've yet to seriously start using AI.<br>
<br>
> Immediate Advice: He urges readers to "be early" by integrating the most powerful models into their daily workflows now. His stance is that the window to gain a competitive advantage is closing, and the only path forward is radical adaptation and financial caution.<br>
<br>
It is true that, even today, AI can give a boost to many careers,<br>
particularly white collar/knowledge work. For blue collar jobs,<br>
useful tools are emerging: they might be better off learning and using<br>
said tools, rather than "learning AI" in and of itself.<br>
<br>
TL;DR: can the hype, cancel the alarm. Do study up on AI or<br>
AI-powered tools, depending on what's more applicable to your career,<br>
like how learning how to use the Web and email became new priorities<br>
about 30 years ago.<br></blockquote></div></div><div dir="auto"><br></div><div dir="auto">The summary reads like hype, but the full article backs everything up with real world examples and data (which the summary doesn't include). Given these data, it's not hype but justified alarm.</div><div dir="auto"><br></div><div dir="auto">Jason </div></div>