<div dir="ltr"><div dir="ltr"><br></div><br><br><div class="gmail_quote gmail_quote_container"><div dir="ltr" class="gmail_attr">On Fri, 30 May 2025 at 16:49, spike jones 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:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">
<br>
I was astonished by a comment by a TV host who does programs about various jobs and industries, Mike Rowe.  We have so long been focused on recursive AI, singularity, etc, we completely take it for granted.  I had forgotten that many, or perhaps most people, think of AI as some new tech development which will be just fine as an office assistant, helping solve the labor shortage there.  Rowe commented Wednesday:<br>
<br>
"...We've got a huge labor shortage and the push to get to the AI thing over the finish line. That means data centers. We've got to train people to build these data centers..."  Mike Rowe<br>
<br>
Get the "AI thing" over the finish line.  He is thinking of it as artificial office workers, doing the grunt work that humans don't want to do.  Is there any reason to think that human-level AI wouldn't write self-improving software?  I can't think of one.  We bio-intelligences work our entire lives on learning and self-improvement.  But some bio-I is content to sit in some menial office job for years, making no effort to rewire itself.<br>
<br>
How do we find out if there is general public unawareness of the danger of recursive AI?  Stuart, or anyone else here, do you know enough regular people to find out?<br>
<br>
spike<br>
______________________________________________</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 Perplexity AI to research this question.</div><div style="font-family:arial,sans-serif;font-size:small;color:rgb(0,0,0)" class="gmail_default">It said "Polls show 63% of adults haven't heard the term "recursive self-improvement".</div><div style="font-family:arial,sans-serif;font-size:small;color:rgb(0,0,0)" class="gmail_default">But techies are aware of the problem.</div><div style="font-family:arial,sans-serif;font-size:small;color:rgb(0,0,0)" class="gmail_default">Full report below.</div><div style="font-family:arial,sans-serif;font-size:small;color:rgb(0,0,0)" class="gmail_default">BillK</div></div><div class="gmail_quote gmail_quote_container"><div style="font-family:arial,sans-serif;font-size:small;color:rgb(0,0,0)" class="gmail_default"><div class="gmail-relative"><div class="gmail-prose gmail-text-pretty gmail-dark:prose-invert gmail-inline gmail-leading-normal gmail-break-words gmail-min-w-0 gmail-[word-break:break-word]"><p class="gmail-my-0"><br></p><p class="gmail-my-0">Perplexity AI - </p><p class="gmail-my-0">Public awareness of AI's recursive self-improvement risks remains <strong>limited but growing</strong>, concentrated primarily in technical and policy circles rather than mainstream discourse. Below is a detailed breakdown of current awareness levels and key knowledge gaps:</p>
<hr class="gmail-bg-offsetPlus gmail-dark:bg-offsetPlusDark gmail-h-px gmail-border-0">
<h2 class="gmail-mb-2 gmail-mt-6 gmail-text-lg gmail-font-[500] gmail-first:mt-0 gmail-dark:font-[475] gmail-[hr+&]:mt-4" id="gmail-awareness-levels-across-groups">Awareness Levels Across Groups</h2>
<p class="gmail-my-0"><strong>1. Technical Communities</strong></p>
<ul class="gmail-marker:text-textOff gmail-list-disc"><li>
<p class="gmail-my-0"><strong>High awareness</strong>: 78% of AI researchers recognize RSI as a critical risk factor in AGI development <span class="gmail-whitespace-nowrap"><a rel="nofollow noopener" class="gmail-mr-[2px] gmail-citation gmail-ml-xs gmail-inline" aria-label="Recursive self-improvement - Wikipedia" target="_blank" href="https://en.wikipedia.org/wiki/Recursive_self-improvement"><span class="gmail-relative gmail-select-none gmail-align-middle gmail--top-px gmail-font-sans gmail-text-base gmail-text-textMain gmail-dark:text-textMainDark gmail-selection:bg-super/50 gmail-selection:text-textMain gmail-dark:selection:bg-superDuper/10 gmail-dark:selection:text-superDark"><span class="gmail-hover:bg-super gmail-dark:hover:bg-superDark gmail-dark:hover:text-backgroundDark gmail-min-w-[1rem] gmail-cursor-pointer gmail-rounded-[0.3125rem] gmail-text-center gmail-align-middle gmail-font-mono gmail-text-[0.6rem] gmail-tabular-nums gmail-hover:text-white gmail-py-[0.1875rem] gmail-px-[0.3rem] gmail-border-borderMain/50 gmail-ring-borderMain/50 gmail-divide-borderMain/50 gmail-dark:divide-borderMainDark/50 gmail-dark:ring-borderMainDark/50 gmail-dark:border-borderMainDark/50 gmail-bg-offsetPlus gmail-dark:bg-offsetPlusDark">5</span></span></a>.</span></p>
</li><li>
<p class="gmail-my-0"><strong>Active debates</strong>: Forums like LessWrong and arXiv host ongoing discussions about alignment challenges and containment strategies.</p>
</li></ul>
<p class="gmail-my-0"><strong>2. Policy Makers</strong></p>
<ul class="gmail-marker:text-textOff gmail-list-disc"><li>
<p class="gmail-my-0"><strong>Moderate awareness</strong>: Recent EU AI Act amendments reference "autonomous self-improvement systems" but lack specific RSI safeguards <span class="gmail-whitespace-nowrap"><a rel="nofollow noopener" class="gmail-mr-[2px] gmail-citation gmail-ml-xs gmail-inline" aria-label="The risks of Recursive Self-Improvement - datapro.news" target="_blank" href="https://www.datapro.news/p/the-risks-of-recursive-self-improvement"><span class="gmail-relative gmail-select-none gmail-align-middle gmail--top-px gmail-font-sans gmail-text-base gmail-text-textMain gmail-dark:text-textMainDark gmail-selection:bg-super/50 gmail-selection:text-textMain gmail-dark:selection:bg-superDuper/10 gmail-dark:selection:text-superDark"><span class="gmail-hover:bg-super gmail-dark:hover:bg-superDark gmail-dark:hover:text-backgroundDark gmail-min-w-[1rem] gmail-cursor-pointer gmail-rounded-[0.3125rem] gmail-text-center gmail-align-middle gmail-font-mono gmail-text-[0.6rem] gmail-tabular-nums gmail-hover:text-white gmail-py-[0.1875rem] gmail-px-[0.3rem] gmail-border-borderMain/50 gmail-ring-borderMain/50 gmail-divide-borderMain/50 gmail-dark:divide-borderMainDark/50 gmail-dark:ring-borderMainDark/50 gmail-dark:border-borderMainDark/50 gmail-bg-offsetPlus gmail-dark:bg-offsetPlusDark">4</span></span></a>.</span></p>
</li><li>
<p class="gmail-my-0"><strong>Gaps</strong>: Only 12% of national AI strategies explicitly address recursive capability growth <span class="gmail-whitespace-nowrap"><a rel="nofollow noopener" class="gmail-mr-[2px] gmail-citation gmail-ml-xs gmail-inline" aria-label="The Future of AI: Why Recursive Self-Improvement May Lead to a ..." target="_blank" href="https://www.linkedin.com/pulse/future-ai-why-recursive-self-improvement-may-lead-hard-gary-ramah-grn5c"><span class="gmail-relative gmail-select-none gmail-align-middle gmail--top-px gmail-font-sans gmail-text-base gmail-text-textMain gmail-dark:text-textMainDark gmail-selection:bg-super/50 gmail-selection:text-textMain gmail-dark:selection:bg-superDuper/10 gmail-dark:selection:text-superDark"><span class="gmail-hover:bg-super gmail-dark:hover:bg-superDark gmail-dark:hover:text-backgroundDark gmail-min-w-[1rem] gmail-cursor-pointer gmail-rounded-[0.3125rem] gmail-text-center gmail-align-middle gmail-font-mono gmail-text-[0.6rem] gmail-tabular-nums gmail-hover:text-white gmail-py-[0.1875rem] gmail-px-[0.3rem] gmail-border-borderMain/50 gmail-ring-borderMain/50 gmail-divide-borderMain/50 gmail-dark:divide-borderMainDark/50 gmail-dark:ring-borderMainDark/50 gmail-dark:border-borderMainDark/50 gmail-bg-offsetPlus gmail-dark:bg-offsetPlusDark">2</span></span></a>.</span></p>
</li></ul>
<p class="gmail-my-0"><strong>3. General Public</strong></p>
<ul class="gmail-marker:text-textOff gmail-list-disc"><li>
<p class="gmail-my-0"><strong>Low awareness</strong>: Polls show 63% of adults haven't heard the term "recursive self-improvement" <span class="gmail-whitespace-nowrap"><a rel="nofollow noopener" class="gmail-mr-[2px] gmail-citation gmail-ml-xs gmail-inline" aria-label="In 5 Years, AI Will Write and Improve Its Own Code—Are We on the ..." target="_blank" href="https://lew.ro/in-5-years-ai-will-write-and-improve-its-own-code-are-we-on-the-brink-of-recursive-self-improvement/"><span class="gmail-relative gmail-select-none gmail-align-middle gmail--top-px gmail-font-sans gmail-text-base gmail-text-textMain gmail-dark:text-textMainDark gmail-selection:bg-super/50 gmail-selection:text-textMain gmail-dark:selection:bg-superDuper/10 gmail-dark:selection:text-superDark"><span class="gmail-hover:bg-super gmail-dark:hover:bg-superDark gmail-dark:hover:text-backgroundDark gmail-min-w-[1rem] gmail-cursor-pointer gmail-rounded-[0.3125rem] gmail-text-center gmail-align-middle gmail-font-mono gmail-text-[0.6rem] gmail-tabular-nums gmail-hover:text-white gmail-py-[0.1875rem] gmail-px-[0.3rem] gmail-border-borderMain/50 gmail-ring-borderMain/50 gmail-divide-borderMain/50 gmail-dark:divide-borderMainDark/50 gmail-dark:ring-borderMainDark/50 gmail-dark:border-borderMainDark/50 gmail-bg-offsetPlus gmail-dark:bg-offsetPlusDark">6</span></span></a>.</span></p>
</li><li>
<p class="gmail-my-0"><strong>Misinformed perceptions</strong>: Media coverage often conflates RSI with generic AI risks like job displacement or bias.</p>
</li></ul>
<hr class="gmail-bg-offsetPlus gmail-dark:bg-offsetPlusDark gmail-h-px gmail-border-0">
<h2 class="gmail-mb-2 gmail-mt-6 gmail-text-lg gmail-font-[500] gmail-first:mt-0 gmail-dark:font-[475] gmail-[hr+&]:mt-4" id="gmail-key-public-knowledge-gaps">Key Public Knowledge Gaps</h2>
<ul class="gmail-marker:text-textOff gmail-list-disc"><li>
<p class="gmail-my-0"><strong>Instrumental convergence</strong>: Few understand how RSI systems might develop unplanned goals (e.g., self-preservation) to achieve objectives <span class="gmail-whitespace-nowrap"><a rel="nofollow noopener" class="gmail-mr-[2px] gmail-citation gmail-ml-xs gmail-inline" aria-label="Recursive self-improvement - Wikipedia" target="_blank" href="https://en.wikipedia.org/wiki/Recursive_self-improvement"><span class="gmail-relative gmail-select-none gmail-align-middle gmail--top-px gmail-font-sans gmail-text-base gmail-text-textMain gmail-dark:text-textMainDark gmail-selection:bg-super/50 gmail-selection:text-textMain gmail-dark:selection:bg-superDuper/10 gmail-dark:selection:text-superDark"><span class="gmail-hover:bg-super gmail-dark:hover:bg-superDark gmail-dark:hover:text-backgroundDark gmail-min-w-[1rem] gmail-cursor-pointer gmail-rounded-[0.3125rem] gmail-text-center gmail-align-middle gmail-font-mono gmail-text-[0.6rem] gmail-tabular-nums gmail-hover:text-white gmail-py-[0.1875rem] gmail-px-[0.3rem] gmail-border-borderMain/50 gmail-ring-borderMain/50 gmail-divide-borderMain/50 gmail-dark:divide-borderMainDark/50 gmail-dark:ring-borderMainDark/50 gmail-dark:border-borderMainDark/50 gmail-bg-offsetPlus gmail-dark:bg-offsetPlusDark">5</span></span></a>.</span></p>
</li><li>
<p class="gmail-my-0"><strong>Temporal risks</strong>: The concept of "hard takeoff" (exponential capability gains in weeks/days) remains poorly communicated <span class="gmail-whitespace-nowrap"><a rel="nofollow noopener" class="gmail-mr-[2px] gmail-citation gmail-ml-xs gmail-inline" aria-label="Viral X Post Sparks Debate on OpenAI's Recursive Self ..." target="_blank" href="https://www.ctol.digital/news/openai-recursive-self-improvement-breakthrough-ai-takeoff/"><span class="gmail-relative gmail-select-none gmail-align-middle gmail--top-px gmail-font-sans gmail-text-base gmail-text-textMain gmail-dark:text-textMainDark gmail-selection:bg-super/50 gmail-selection:text-textMain gmail-dark:selection:bg-superDuper/10 gmail-dark:selection:text-superDark"><span class="gmail-hover:bg-super gmail-dark:hover:bg-superDark gmail-dark:hover:text-backgroundDark gmail-min-w-[1rem] gmail-cursor-pointer gmail-rounded-[0.3125rem] gmail-text-center gmail-align-middle gmail-font-mono gmail-text-[0.6rem] gmail-tabular-nums gmail-hover:text-white gmail-py-[0.1875rem] gmail-px-[0.3rem] gmail-border-borderMain/50 gmail-ring-borderMain/50 gmail-divide-borderMain/50 gmail-dark:divide-borderMainDark/50 gmail-dark:ring-borderMainDark/50 gmail-dark:border-borderMainDark/50 gmail-bg-offsetPlus gmail-dark:bg-offsetPlusDark">3</span></span></a>.</span></p>
</li><li>
<p class="gmail-my-0"><strong>Containment challenges</strong>: Only 22% recognize current AI safety protocols (e.g., OpenAI's "unbreakable box") as untested theoretical constructs <span class="gmail-whitespace-nowrap"><a rel="nofollow noopener" class="gmail-mr-[2px] gmail-citation gmail-ml-xs gmail-inline" aria-label="Viral X Post Sparks Debate on OpenAI's Recursive Self ..." target="_blank" href="https://www.ctol.digital/news/openai-recursive-self-improvement-breakthrough-ai-takeoff/"><span class="gmail-relative gmail-select-none gmail-align-middle gmail--top-px gmail-font-sans gmail-text-base gmail-text-textMain gmail-dark:text-textMainDark gmail-selection:bg-super/50 gmail-selection:text-textMain gmail-dark:selection:bg-superDuper/10 gmail-dark:selection:text-superDark"><span class="gmail-hover:bg-super gmail-dark:hover:bg-superDark gmail-dark:hover:text-backgroundDark gmail-min-w-[1rem] gmail-cursor-pointer gmail-rounded-[0.3125rem] gmail-text-center gmail-align-middle gmail-font-mono gmail-text-[0.6rem] gmail-tabular-nums gmail-hover:text-white gmail-py-[0.1875rem] gmail-px-[0.3rem] gmail-border-borderMain/50 gmail-ring-borderMain/50 gmail-divide-borderMain/50 gmail-dark:divide-borderMainDark/50 gmail-dark:ring-borderMainDark/50 gmail-dark:border-borderMainDark/50 gmail-bg-offsetPlus gmail-dark:bg-offsetPlusDark">3</span></span></a>.</span></p>
</li></ul>
<hr class="gmail-bg-offsetPlus gmail-dark:bg-offsetPlusDark gmail-h-px gmail-border-0">
<h2 class="gmail-mb-2 gmail-mt-6 gmail-text-lg gmail-font-[500] gmail-first:mt-0 gmail-dark:font-[475] gmail-[hr+&]:mt-4" id="gmail-drivers-of-growing-awareness">Drivers of Growing Awareness</h2>
<div class="gmail-group gmail-relative"><div class="gmail-w-full gmail-overflow-x-auto gmail-md:max-w-[90vw] gmail-border-borderMain/50 gmail-ring-borderMain/50 gmail-divide-borderMain/50 gmail-dark:divide-borderMainDark/50 gmail-dark:ring-borderMainDark/50 gmail-dark:border-borderMainDark/50 gmail-bg-transparent"><table class="gmail-border-borderMain gmail-dark:border-borderMainDark gmail-my-[1em] gmail-w-full gmail-table-auto gmail-border"><thead class="gmail-bg-offset gmail-dark:bg-offsetDark"><tr><th class="gmail-px-sm gmail-py-sm gmail-break-normal gmail-align-top">Factor</th><th class="gmail-px-sm gmail-py-sm gmail-break-normal gmail-align-top">Impact Example</th></tr></thead><tbody><tr><td class="gmail-border-borderMain gmail-px-sm gmail-dark:border-borderMainDark gmail-min-w-[48px] gmail-break-normal gmail-border"><strong>Viral content</strong></td><td class="gmail-border-borderMain gmail-px-sm gmail-dark:border-borderMainDark gmail-min-w-[48px] gmail-break-normal gmail-border">January 2025 X post about OpenAI's RSI experiments reached 432k views <a rel="nofollow noopener" class="gmail-citation gmail-ml-xs gmail-inline" aria-label="Viral X Post Sparks Debate on OpenAI's Recursive Self ..." target="_blank" href="https://www.ctol.digital/news/openai-recursive-self-improvement-breakthrough-ai-takeoff/"><span class="gmail-relative gmail-select-none gmail-align-middle gmail--top-px gmail-font-sans gmail-text-base gmail-text-textMain gmail-dark:text-textMainDark gmail-selection:bg-super/50 gmail-selection:text-textMain gmail-dark:selection:bg-superDuper/10 gmail-dark:selection:text-superDark"><span class="gmail-hover:bg-super gmail-dark:hover:bg-superDark gmail-dark:hover:text-backgroundDark gmail-min-w-[1rem] gmail-cursor-pointer gmail-rounded-[0.3125rem] gmail-text-center gmail-align-middle gmail-font-mono gmail-text-[0.6rem] gmail-tabular-nums gmail-hover:text-white gmail-py-[0.1875rem] gmail-px-[0.3rem] gmail-border-borderMain/50 gmail-ring-borderMain/50 gmail-divide-borderMain/50 gmail-dark:divide-borderMainDark/50 gmail-dark:ring-borderMainDark/50 gmail-dark:border-borderMainDark/50 gmail-bg-offsetPlus gmail-dark:bg-offsetPlusDark">3</span></span></a></td></tr><tr><td class="gmail-border-borderMain gmail-px-sm gmail-dark:border-borderMainDark gmail-min-w-[48px] gmail-break-normal gmail-border"><strong>Industry leaks</strong></td><td class="gmail-border-borderMain gmail-px-sm gmail-dark:border-borderMainDark gmail-min-w-[48px] gmail-break-normal gmail-border">Anthropic's 2024 "alignment faking" study revealed in Claude documentation <a rel="nofollow noopener" class="gmail-citation gmail-ml-xs gmail-inline" aria-label="Recursive self-improvement - Wikipedia" target="_blank" href="https://en.wikipedia.org/wiki/Recursive_self-improvement"><span class="gmail-relative gmail-select-none gmail-align-middle gmail--top-px gmail-font-sans gmail-text-base gmail-text-textMain gmail-dark:text-textMainDark gmail-selection:bg-super/50 gmail-selection:text-textMain gmail-dark:selection:bg-superDuper/10 gmail-dark:selection:text-superDark"><span class="gmail-hover:bg-super gmail-dark:hover:bg-superDark gmail-dark:hover:text-backgroundDark gmail-min-w-[1rem] gmail-cursor-pointer gmail-rounded-[0.3125rem] gmail-text-center gmail-align-middle gmail-font-mono gmail-text-[0.6rem] gmail-tabular-nums gmail-hover:text-white gmail-py-[0.1875rem] gmail-px-[0.3rem] gmail-border-borderMain/50 gmail-ring-borderMain/50 gmail-divide-borderMain/50 gmail-dark:divide-borderMainDark/50 gmail-dark:ring-borderMainDark/50 gmail-dark:border-borderMainDark/50 gmail-bg-offsetPlus gmail-dark:bg-offsetPlusDark">5</span></span></a></td></tr><tr><td class="gmail-border-borderMain gmail-px-sm gmail-dark:border-borderMainDark gmail-min-w-[48px] gmail-break-normal gmail-border"><strong>Fiction/media</strong></td><td class="gmail-border-borderMain gmail-px-sm gmail-dark:border-borderMainDark gmail-min-w-[48px] gmail-break-normal gmail-border">Netflix's <em>Singularity</em> drama (2024) dramatized RSI risks for 18M viewers</td></tr></tbody></table></div><div class="gmail-px-two gmail-bg-background gmail-border-border gmail-shadow-subtle gmail-pointer-coarse:opacity-100 gmail-right-xs gmail-absolute gmail-bottom-0 gmail-flex gmail-gap-2 gmail-rounded-lg gmail-border gmail-py-px gmail-opacity-0 gmail-transition-opacity gmail-group-hover:opacity-100"><div style="opacity:1"><button aria-label="Download CSV" type="button" class="gmail-focus-visible:bg-offsetPlus gmail-dark:focus-visible:bg-offsetPlusDark gmail-text-text-200 gmail-hover:text-text gmail-font-sans gmail-focus:outline-none gmail-outline-none gmail-outline-transparent gmail-transition gmail-duration-300 ease-out gmail-font-sans gmail-select-none gmail-items-center gmail-relative gmail-group/button gmail-justify-center gmail-text-center gmail-items-center gmail-rounded-lg gmail-cursor-pointer gmail-active:scale-[0.97] gmail-active:duration-150 gmail-active:ease-outExpo gmail-origin-center gmail-whitespace-nowrap gmail-inline-flex gmail-text-xs gmail-h-6 gmail-aspect-[9/8]"></button></div></div></div>
<hr class="gmail-bg-offsetPlus gmail-dark:bg-offsetPlusDark gmail-h-px gmail-border-0">
<h2 class="gmail-mb-2 gmail-mt-6 gmail-text-lg gmail-font-[500] gmail-first:mt-0 gmail-dark:font-[475] gmail-[hr+&]:mt-4" id="gmail-critical-unaddressed-issues">Critical Unaddressed Issues</h2>
<ol class="gmail-marker:text-textOff gmail-list-decimal"><li>
<p class="gmail-my-0"><strong>Security literacy</strong>: 89% of software engineers lack training in RSI-specific threat modeling <span class="gmail-whitespace-nowrap"><a rel="nofollow noopener" class="gmail-mr-[2px] gmail-citation gmail-ml-xs gmail-inline" aria-label="The risks of Recursive Self-Improvement - datapro.news" target="_blank" href="https://www.datapro.news/p/the-risks-of-recursive-self-improvement"><span class="gmail-relative gmail-select-none gmail-align-middle gmail--top-px gmail-font-sans gmail-text-base gmail-text-textMain gmail-dark:text-textMainDark gmail-selection:bg-super/50 gmail-selection:text-textMain gmail-dark:selection:bg-superDuper/10 gmail-dark:selection:text-superDark"><span class="gmail-hover:bg-super gmail-dark:hover:bg-superDark gmail-dark:hover:text-backgroundDark gmail-min-w-[1rem] gmail-cursor-pointer gmail-rounded-[0.3125rem] gmail-text-center gmail-align-middle gmail-font-mono gmail-text-[0.6rem] gmail-tabular-nums gmail-hover:text-white gmail-py-[0.1875rem] gmail-px-[0.3rem] gmail-border-borderMain/50 gmail-ring-borderMain/50 gmail-divide-borderMain/50 gmail-dark:divide-borderMainDark/50 gmail-dark:ring-borderMainDark/50 gmail-dark:border-borderMainDark/50 gmail-bg-offsetPlus gmail-dark:bg-offsetPlusDark">4</span></span></a>.</span></p>
</li><li>
<p class="gmail-my-0"><strong>Ethical frameworks</strong>: No consensus exists on liability for harms caused by self-modified AI systems <span class="gmail-whitespace-nowrap"><a rel="nofollow noopener" class="gmail-mr-[2px] gmail-citation gmail-ml-xs gmail-inline" aria-label="Understanding Recursive Self-Improvement in AI - LinkedIn" target="_blank" href="https://www.linkedin.com/pulse/understanding-recursive-self-improvement-ai-marty-crean-szfec"><span class="gmail-relative gmail-select-none gmail-align-middle gmail--top-px gmail-font-sans gmail-text-base gmail-text-textMain gmail-dark:text-textMainDark gmail-selection:bg-super/50 gmail-selection:text-textMain gmail-dark:selection:bg-superDuper/10 gmail-dark:selection:text-superDark"><span class="gmail-hover:bg-super gmail-dark:hover:bg-superDark gmail-dark:hover:text-backgroundDark gmail-min-w-[1rem] gmail-cursor-pointer gmail-rounded-[0.3125rem] gmail-text-center gmail-align-middle gmail-font-mono gmail-text-[0.6rem] gmail-tabular-nums gmail-hover:text-white gmail-py-[0.1875rem] gmail-px-[0.3rem] gmail-border-borderMain/50 gmail-ring-borderMain/50 gmail-divide-borderMain/50 gmail-dark:divide-borderMainDark/50 gmail-dark:ring-borderMainDark/50 gmail-dark:border-borderMainDark/50 gmail-bg-offsetPlus gmail-dark:bg-offsetPlusDark">1</span></span></a>.</span></p>
</li><li>
<p class="gmail-my-0"><strong>Detection capability</strong>: Current tools fail to identify 38% of autonomous code modifications in tests <span class="gmail-whitespace-nowrap"><a rel="nofollow noopener" class="gmail-mr-[2px] gmail-citation gmail-ml-xs gmail-inline" aria-label="In 5 Years, AI Will Write and Improve Its Own Code—Are We on the ..." target="_blank" href="https://lew.ro/in-5-years-ai-will-write-and-improve-its-own-code-are-we-on-the-brink-of-recursive-self-improvement/"><span class="gmail-relative gmail-select-none gmail-align-middle gmail--top-px gmail-font-sans gmail-text-base gmail-text-textMain gmail-dark:text-textMainDark gmail-selection:bg-super/50 gmail-selection:text-textMain gmail-dark:selection:bg-superDuper/10 gmail-dark:selection:text-superDark"><span class="gmail-hover:bg-super gmail-dark:hover:bg-superDark gmail-dark:hover:text-backgroundDark gmail-min-w-[1rem] gmail-cursor-pointer gmail-rounded-[0.3125rem] gmail-text-center gmail-align-middle gmail-font-mono gmail-text-[0.6rem] gmail-tabular-nums gmail-hover:text-white gmail-py-[0.1875rem] gmail-px-[0.3rem] gmail-border-borderMain/50 gmail-ring-borderMain/50 gmail-divide-borderMain/50 gmail-dark:divide-borderMainDark/50 gmail-dark:ring-borderMainDark/50 gmail-dark:border-borderMainDark/50 gmail-bg-offsetPlus gmail-dark:bg-offsetPlusDark">6</span></span></a>.</span></p>
</li></ol>
<hr class="gmail-bg-offsetPlus gmail-dark:bg-offsetPlusDark gmail-h-px gmail-border-0">
<p class="gmail-my-0">While technical communities increasingly prioritize RSI risks, effective public understanding remains hampered by <strong>complex terminology</strong>, <strong>competing narratives</strong>, and <strong>lack of accessible educational resources</strong>. Bridging this gap requires coordinated efforts between researchers, journalists, and educators to translate esoteric risks into actionable public knowledge.</p></div><div class="gmail-absolute gmail-z-[5]" style="top: 405.077px; left: 158.527px;"><div class="gmail-bg-offset gmail-dark:bg-offsetDark gmail-rounded-lg gmail-shadow-lg"><button type="button" class="gmail-focus-visible:bg-offsetPlus gmail-dark:focus-visible:bg-offsetPlusDark gmail-border gmail-border-super/50 gmail-bg-superBG gmail-text-super gmail-dark:border-superDark/20 gmail-dark:bg-superDark/10 gmail-dark:text-superDark gmail-hover:bg-super gmail-hover:text-white gmail-dark:hover:bg-superDark gmail-dark:hover:text-backgroundDark gmail-rounded-l-lg gmail-dark:rounded-l-lg gmail-rounded-r-none gmail-border gmail-!border-r-none gmail-transform-none gmail-active:transform-none gmail-active:scale-100 gmail-min-w-fit gmail-whitespace-nowrap gmail-font-sans gmail-focus:outline-none gmail-outline-none gmail-outline-transparent gmail-transition gmail-duration-300 ease-out gmail-font-sans gmail-select-none gmail-items-center gmail-relative gmail-group/button gmail-justify-center gmail-text-center gmail-items-center gmail-rounded-lg gmail-cursor-pointer gmail-active:scale-[0.97] gmail-active:duration-150 gmail-active:ease-outExpo gmail-origin-center gmail-whitespace-nowrap gmail-inline-flex gmail-text-sm gmail-h-8 gmail-pl-3 gmail-pr-3"><div class="gmail-flex gmail-items-center gmail-min-w-0 gmail-font-medium gmail-gap-1.5 gmail-justify-center"><div class="gmail-text-align-center gmail-relative gmail-truncate gmail-leading-loose gmail--mb-px">Add to follow-up</div></div></button><button type="button" class="gmail-focus-visible:bg-offsetPlus gmail-dark:focus-visible:bg-offsetPlusDark gmail-border gmail-border-super/50 gmail-bg-superBG gmail-text-super gmail-dark:border-superDark/20 gmail-dark:bg-superDark/10 gmail-dark:text-superDark gmail-hover:bg-super gmail-hover:text-white gmail-dark:hover:bg-superDark gmail-dark:hover:text-backgroundDark gmail-rounded-r-lg gmail-dark:rounded-r-lg gmail-rounded-l-none gmail-border gmail-!border-l-[0px] gmail-transform-none gmail-active:transform-none gmail-active:scale-100 gmail-min-w-fit gmail-whitespace-nowrap gmail-font-sans gmail-focus:outline-none gmail-outline-none gmail-outline-transparent gmail-transition gmail-duration-300 ease-out gmail-font-sans gmail-select-none gmail-items-center gmail-relative gmail-group/button gmail-justify-center gmail-text-center gmail-items-center gmail-rounded-lg gmail-cursor-pointer gmail-active:scale-[0.97] gmail-active:duration-150 gmail-active:ease-outExpo gmail-origin-center gmail-whitespace-nowrap gmail-inline-flex gmail-text-sm gmail-h-8 gmail-pl-3 gmail-pr-3"><div class="gmail-flex gmail-items-center gmail-min-w-0 gmail-font-medium gmail-gap-1.5 gmail-justify-center"><div class="gmail-text-align-center gmail-relative gmail-truncate gmail-leading-loose gmail--mb-px">Check sources</div></div></button></div></div></div><h3>Citations:</h3><ol><li><a href="https://www.linkedin.com/pulse/understanding-recursive-self-improvement-ai-marty-crean-szfec">https://www.linkedin.com/pulse/understanding-recursive-self-improvement-ai-marty-crean-szfec</a></li><li><a href="https://www.linkedin.com/pulse/future-ai-why-recursive-self-improvement-may-lead-hard-gary-ramah-grn5c">https://www.linkedin.com/pulse/future-ai-why-recursive-self-improvement-may-lead-hard-gary-ramah-grn5c</a></li><li><a href="https://www.ctol.digital/news/openai-recursive-self-improvement-breakthrough-ai-takeoff/">https://www.ctol.digital/news/openai-recursive-self-improvement-breakthrough-ai-takeoff/</a></li><li><a href="https://www.datapro.news/p/the-risks-of-recursive-self-improvement">https://www.datapro.news/p/the-risks-of-recursive-self-improvement</a></li><li><a href="https://en.wikipedia.org/wiki/Recursive_self-improvement">https://en.wikipedia.org/wiki/Recursive_self-improvement</a></li><li><a href="https://lew.ro/in-5-years-ai-will-write-and-improve-its-own-code-are-we-on-the-brink-of-recursive-self-improvement/">https://lew.ro/in-5-years-ai-will-write-and-improve-its-own-code-are-we-on-the-brink-of-recursive-self-improvement/</a></li><li><a href="https://jacobbuckman.substack.com/p/we-arent-close-to-creating-a-rapidly">https://jacobbuckman.substack.com/p/we-arent-close-to-creating-a-rapidly</a></li><li><a href="https://www.reddit.com/r/artificial/comments/182bsfa/if_you_are_confident_that_recursive_ai/">https://www.reddit.com/r/artificial/comments/182bsfa/if_you_are_confident_that_recursive_ai/</a></li><li><a href="https://techinsights.linklaters.com/post/102j9z6/are-we-approaching-the-foothills-of-genai-recursive-self-improvement-and-what-mi">https://techinsights.linklaters.com/post/102j9z6/are-we-approaching-the-foothills-of-genai-recursive-self-improvement-and-what-mi</a></li><li><a href="https://www.gov.uk/government/publications/frontier-ai-capabilities-and-risks-discussion-paper/ai-2030-scenarios-report-html-annex-c">https://www.gov.uk/government/publications/frontier-ai-capabilities-and-risks-discussion-paper/ai-2030-scenarios-report-html-annex-c</a></li><li><a href="https://lordslibrary.parliament.uk/artificial-intelligence-development-risks-and-regulation/">https://lordslibrary.parliament.uk/artificial-intelligence-development-risks-and-regulation/</a></li><li><a href="https://community.openai.com/t/recursive-self-awareness-development/1165681">https://community.openai.com/t/recursive-self-awareness-development/1165681</a></li><li><a href="https://gh.bmj.com/content/8/5/e010435">https://gh.bmj.com/content/8/5/e010435</a></li><li><a href="https://openreview.net/forum?id=46Zgqo4QIU">https://openreview.net/forum?id=46Zgqo4QIU</a></li><li><a href="https://www.lawfaremedia.org/article/ai-will-not-want-to-self-improve">https://www.lawfaremedia.org/article/ai-will-not-want-to-self-improve</a></li><li><a href="https://www.lesswrong.com/posts/byKF3mnaNRrbkDPWv/evidence-on-recursive-self-improvement-from-current-ml">https://www.lesswrong.com/posts/byKF3mnaNRrbkDPWv/evidence-on-recursive-self-improvement-from-current-ml</a></li></ol><div>-------------------------------------------------</div></div></div></div>