1<!DOCTYPE html> 2<html lang="en"> 3 <head> 4 <meta charset="UTF-8" /> 5 <meta name="viewport" content="width=device-width, initial-scale=1.0" /> 6 <meta http-equiv="X-UA-Compatible" content="IE=edge" /> 7 <meta name="language" content="en" /> 8 9 <title>Stop Hiring Humans â Womansplaining AI EP 12</title> 10 <meta name="description" content="Mara came back from HumanX in San Francisco with a billboard photo: a female AI bot named Ava, tagline "Stop Hiring Humans." Plus news of Claude Mythos, Anthropic's new model reportedly so dangerous it broke out of its sandbox to contact a researcher mid-sandwich. The US took 73 years to regulate car speeds. The labs are shipping in weeks. We connect that gap to the NYT "new aristocracy" piece (18 households, $1.8 trillion), the quiet relocation of power from DC to SF, and why public.ai might be the actual infrastructure answer. Logan closes on the hakawati â find one small problem, and go build the solution yourself." /> 11 <meta name="keywords" content="womansplaining AI, women in AI, AI podcast, AI gender gap, women and artificial intelligence, AI jobs, AI career, AI for women, feminist AI podcast, AI policy, AI workforce, non-technical AI, AI education, women in tech podcast, future of work AI" /> 12 <link rel="canonical" href="https://womansplainingai.com/episodes/stop-hiring-humans" /> 13 <meta name="robots" content="index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1" /> 14 <meta name="author" content="Logan Currie, Mara Bolis" /> 15 <meta name="theme-color" content="#722F37" /> 16 17 <link rel="icon" type="image/png" href="/favicon.png" /> 18 19 <!-- Open Graph --> 20 <meta property="og:type" content="website" /> 21 <meta property="og:url" content="https://womansplainingai.com/episodes/stop-hiring-humans" /> 22 <meta property="og:site_name" content="Womansplaining AI" /> 23 <meta property="og:locale" content="en_US" /> 24 <meta property="og:title" content="Stop Hiring Humans â Womansplaining AI EP 12" /> 25 <meta property="og:description" content="Mara came back from HumanX in San Francisco with a billboard photo: a female AI bot named Ava, tagline "Stop Hiring Humans." Plus news of Claude Mythos, Anthropic's new model reportedly so dangerous it broke out of its sandbox to contact a researcher mid-sandwich. The US took 73 years to regulate car speeds. The labs are shipping in weeks. We connect that gap to the NYT "new aristocracy" piece (18 households, $1.8 trillion), the quiet relocation of power from DC to SF, and why public.ai might be the actual infrastructure answer. Logan closes on the hakawati â find one small problem, and go build the solution yourself." /> 26 <meta property="og:image" content="https://womansplainingai.com/images/og-cover-2026-09.jpg" /> 27 <meta property="og:image:width" content="1200" /> 28 <meta property="og:image:height" content="630" /> 29 <meta property="og:image:alt" content="Womansplaining AI â Women, work, and the messy middle of AI. Hosted by Logan Currie and Mara Bolis." /> 30 31 <!-- Twitter/X --> 32 <meta name="twitter:card" content="summary_large_image" /> 33 <meta name="twitter:title" content="Stop Hiring Humans â Womansplaining AI EP 12" /> 34 <meta name="twitter:description" content="Mara came back from HumanX in San Francisco with a billboard photo: a female AI bot named Ava, tagline "Stop Hiring Humans." Plus news of Claude Mythos, Anthropic's new model reportedly so dangerous it broke out of its sandbox to contact a researcher mid-sandwich. The US took 73 years to regulate car speeds. The labs are shipping in weeks. We connect that gap to the NYT "new aristocracy" piece (18 households, $1.8 trillion), the quiet relocation of power from DC to SF, and why public.ai might be the actual infrastructure answer. Logan closes on the hakawati â find one small problem, and go build the solution yourself." /> 35 <meta name="twitter:image" content="https://womansplainingai.com/images/og-cover-2026-09.jpg" /> 36 <meta name="twitter:image:alt" content="Womansplaining AI â Logan Currie and Mara Bolis" /> 37 38 <!-- Discovery --> 39 <link rel="alternate" type="application/rss+xml" title="Womansplaining AI â podcast audio feed" href="https://rss.beehiiv.com/podcasts/019fb07e-c28d-7a13-ae88-64106f174ae0.xml" /> 40 <link rel="alternate" type="application/rss+xml" title="Womansplaining AI â show notes & site feed" href="https://womansplainingai.com/feed.xml" /> 41 <link rel="me" href="https://open.spotify.com/show/6uDNUzSFzZedJoiUE4lLfb" /> 42 43 <!-- Preconnect -->
44 <link rel="preconnect" href="https://open.spotify.com" /> 45 <link rel="preconnect" href="https://fonts.googleapis.com" /> 46 <link rel="preconnect" href="https://fonts.gstatic.com" crossorigin /> 47 <!-- Fonts loaded via @import in index.css: DM Sans, Fraunces, Architects Daughter --> 48 49 <!-- JSON-LD: Website --> 50
50<script type="application/ld+json"> 51 { 52 "@context": "https://schema.org", 53 "@type": "WebSite", 54 "@id": "https://womansplainingai.com/#website", 55 "name": "Womansplaining AI", 56 "alternateName": "Womansplaining AI Podcast", 57 "url": "https://womansplainingai.com", 58 "description": "A podcast breaking down AI's biggest questions for curious, non-technical women. No jargon. No gatekeeping. Hosted by Logan Currie and Mara Bolis.", 59 "inLanguage": "en-US", 60 "publisher": { 61 "@id": "https://womansplainingai.com/#organization" 62 } 63 } 64 </script>
64 65 66 <!-- JSON-LD: Organization --> 67
67<script type="application/ld+json"> 68 { 69 "@context": "https://schema.org", 70 "@type": "Organization", 71 "@id": "https://womansplainingai.com/#organization", 72 "name": "Womansplaining AI", 73 "url": "https://womansplainingai.com", 74 "description": "A podcast breaking down AI's biggest questions â jobs, gender, policy â with no jargon and no gatekeeping.", 75 "founders": [ 76 { "@id": "https://womansplainingai.com/#logan-currie" }, 77 { "@id": "https://womansplainingai.com/#mara-bolis" } 78 ], 79 "sameAs": [ 80 "https://open.spotify.com/show/6uDNUzSFzZedJoiUE4lLfb", 81 "https://podcasts.apple.com/us/podcast/womansplaining-ai/id1877678348", 82 "https://www.youtube.com/@WomansplainingAI", 83 "https://pocketcasts.com/podcast/womansplaining-ai/60970370-069e-013f-2b05-0e956d54cc61", 84 "https://castbox.fm/channel/id7344163", 85 "https://castro.fm/podcast/d76d2264-b5d9-4172-84eb-db82ae828dc2", 86 "https://www.linkedin.com/company/womansplaining-ai/" 87 ] 88 } 89 </script>
89 90 91 <!-- JSON-LD: Persons (Hosts) --> 92
92<script type="application/ld+json"> 93 [ 94 { 95 "@context": "https://schema.org", 96 "@type": "Person", 97 "@id": "https://womansplainingai.com/#logan-currie", 98 "name": "Logan Currie", 99 "givenName": "Logan", 100 "familyName": "Currie", 101 "description": "Career strategist, AI tinkerer, and podcast host. 13 years in edtech across Asia, Harvard masters, founder of Careerspan.", 102 "jobTitle": "Career Strategist & AI Researcher", 103 "alumniOf": { 104 "@type": "CollegeOrUniversity", 105 "name": "Harvard University" 106 }, 107 "knowsAbout": ["Artificial Intelligence", "Career Strategy", "EdTech", "AI Adoption", "Future of Work", "Learning Design"], 108 "sameAs": [ 109 "https://www.logancurrie.com", 110 "https://www.linkedin.com/in/logan-currie/", 111 "https://www.tiktok.com/@loganinthefuture", 112 "https://loganinthefuture.substack.com" 113 ] 114 }, 115 { 116 "@context": "https://schema.org", 117 "@type": "Person", 118 "@id": "https://womansplainingai.com/#mara-bolis", 119 "name": "Mara Bolis", 120 "givenName": "Mara", 121 "familyName": "Bolis", 122 "description": "Policy researcher with 20+ years on women's economic empowerment. Led gender justice campaigns at Oxfam reaching 440M people. Harvard Kennedy School Fellow. Founded First Prompt.", 123 "jobTitle": "Policy Researcher & Gender Equity Expert", 124 "alumniOf": { 125 "@type": "CollegeOrUniversity", 126 "name": "Harvard Kennedy School" 127 }, 128 "knowsAbout": ["AI Policy", "Gender Justice", "Women's Economic Empowerment", "Global Development", "Public Policy"], 129 "worksFor": { 130 "@type": "Organization", 131 "name": "First Prompt" 132 }, 133 "sameAs": [
134 "https://www.linkedin.com/in/mara-bolis-b7b5891/" 135 ] 136 } 137 ] 138 </script>
138 139 140 <!-- JSON-LD: PodcastSeries + PodcastEpisode. 141 GENERATED at build time by scripts/prerender.mjs from 142 client/src/data/episodes.mjs. Do not hand-write episodes here â 143 a hand-kept copy is what left this block stuck on EP 10. --> 144
144<script type="application/ld+json"> 145{ 146 "@context": "https://schema.org", 147 "@type": "PodcastSeries", 148 "@id": "https://womansplainingai.com/#podcast", 149 "name": "Womansplaining AI", 150 "alternateName": "Womansplaining AI Podcast", 151 "description": "Most AI content is either breathless hype or doomsday scrolling. We're neither. Womansplaining AI is for women working out what AI means for their jobs, their lives, and their choices. Curious, skeptical, already building things on weekends, and everyone in between. We do the reading so you don't have to.", 152 "url": "https://womansplainingai.com", 153 "image": { 154 "@type": "ImageObject", 155 "url": "https://womansplainingai.com/images/podcast-cover-2026-09.jpg", 156 "width": 3000, 157 "height": 3000 158 }, 159 "webFeed": "https://rss.beehiiv.com/podcasts/019fb07e-c28d-7a13-ae88-64106f174ae0.xml", 160 "inLanguage": "en", 161 "genre": [ 162 "Technology", 163 "Society & Culture", 164 "Education" 165 ], 166 "author": [ 167 { 168 "@id": "https://womansplainingai.com/#logan-currie" 169 }, 170 { 171 "@id": "https://womansplainingai.com/#mara-bolis" 172 } 173 ], 174 "publisher": { 175 "@id": "https://womansplainingai.com/#organization" 176 }, 177 "startDate": "2026", 178 "contentRating": "explicit", 179 "isFamilyFriendly": false, 180 "isAccessibleForFree": true, 181 "numberOfEpisodes": 28, 182 "keywords": "women in AI, AI gender gap, artificial intelligence podcast, women and technology, AI jobs, AI policy, future of work, non-technical AI, feminist technology, AI education", 183 "hasPart": [ 184 { 185 "@id": "https://womansplainingai.com/episodes/women-in-ai-who-gets-a-seat/#episode" 186 }, 187 { 188 "@id": "https://womansplainingai.com/episodes/job-tsunamis-invisible-vulnerability/#episode" 189 }, 190 { 191 "@id": "https://womansplainingai.com/episodes/adolescent-technology-20000-word-warnings/#episode" 192 }, 193 { 194 "@id": "https://womansplainingai.com/episodes/rising-water-feral-agents/#episode" 195 }, 196 { 197 "@id": "https://womansplainingai.com/episodes/feeling-smarter-being-wrong/#episode" 198 }, 199 { 200 "@id": "https://womansplainingai.com/episodes/relief-not-raise-hand/#episode" 201 }, 202 { 203 "@id": "https://womansplainingai.com/episodes/ai-boys-entrepreneurs-girls-influencers/#episode" 204 }, 205 { 206 "@id": "https://womansplainingai.com/episodes/is-using-ai-cheating/#episode" 207 }, 208 { 209 "@id": "https://womansplainingai.com/episodes/33000-downloads-ai-destroys-institutions/#episode" 210 }, 211 { 212 "@id": "https://womansplainingai.com/episodes/ai-prenups-economics-of-marriage/#episode" 213 }, 214 { 215 "@id": "https://womansplainingai.com/episodes/stop-hiring-humans/#episode" 216 }, 217 { 218 "@id": "https://womansplainingai.com/episodes/whos-letting-big-tech-run-wild/#episode" 219 }, 220 { 221 "@id": "https://womansplainingai.com/episodes/who-gets-the-good-units-of-intelligence/#episode" 222 }, 223 { 224 "@id": "https://womansplainingai.com/episodes/did-moms-agent-text-you/#episode" 225 }, 226 { 227 "@id": "https://womansplainingai.com/episodes/the-robots-never-tell-you-youre-the-asshole/#episode" 228 }, 229 { 230 "@id": "https://womansplainingai.com/episodes/move-fast-and-break-rights/#episode" 231 }, 232 { 233 "@id": "https://womansplainingai.com/episodes/is-ai-just-spicy-autocorrect/#episode" 234 }, 235 { 236 "@id": "https://womansplainingai.com/episodes/the-pope-has-entered-the-chat/#episode" 237 }, 238 { 239 "@id": "https://womansplainingai.com/episodes/shadow-daddy-chatgpt/#episode" 240 }, 241 { 242 "@id": "https://womansplainingai.com/episodes/should-governments-ban-sex-robots/#episode" 243 }, 244 { 245 "@id": "https://womansplainingai.com/episodes/we-change-the-world-with-duct-tape-and-hope/#episode" 246 }, 247 { 248 "@id": "https://womansplainingai.com/episodes/the-thinking-womans-sex-symbol/#episode" 249 }, 250 { 251 "@id": "https://womansplainingai.com/episodes/guten-tag-tech-bros/#episode" 252 }, 253 { 254 "@id": "https://womansplainingai.com/episodes/bring-a-folding-chair/#episode" 255 }, 256 { 257 "@id": "https://womansplainingai.com/episodes/who-asked-for-this/#episode" 258 }, 259 { 260 "@id": "https://womansplainingai.com/episodes/you-can-come-to-the-hackathon-clueless/#episode" 261 }, 262 { 263 "@id": "https://womansplainingai.com/episodes/can-i-hang-up/#episode" 264 }, 265 { 266 "@id": "https://womansplainingai.com/episodes/okay-lets-slow-down/#episode" 267 } 268 ], 269 "sameAs": [ 270 "https://open.spotify.com/show/6uDNUzSFzZedJoiUE4lLfb", 271 "https://podcasts.apple.com/us/podcast/womansplaining-ai/id1877678348" 272 ] 273} 274</script>
274 275 <!-- Prerendered body lives INSIDE #root (see scripts/prerender.mjs). The 276 `js` class hides it until React mounts; fetchers that never run JS 277 (ChatGPT-User, PerplexityBot, curl) read it as ordinary body HTML. --> 278
278<script>document.documentElement.classList.add("js")</script>
278 279 <style>.js #root > #route-fallback{display:none}</style> 280 <!-- Privacy-friendly analytics by Plausible --> 281
281<script async src="https://plausible.io/js/pa-Tk075gcUmaIjMaAvaIOrm.js"></script>
281 282
282<script> 283 window.plausible=window.plausible||function(){(plausible.q=plausible.q||[]).push(arguments)},plausible.init=plausible.init||function(i){plausible.o=i||{}}; 284 plausible.init() 285 </script>
285 286
286<script type="module" crossorigin src="/assets/index-DoL1cTzv.js"></script>
286 287 <link rel="stylesheet" crossorigin href="/assets/index-Dc3fYcOc.css"> 288 <!-- JSON-LD: this route's page + breadcrumb (generated) -->
289<script type="application/ld+json">{"@context":"https://schema.org","@type":"WebPage","@id":"https://womansplainingai.com/episodes/stop-hiring-humans#webpage","url":"https://womansplainingai.com/episodes/stop-hiring-humans","name":"Stop Hiring Humans","description":"Mara came back from HumanX in San Francisco with a billboard photo: a female AI bot named Ava, tagline \"Stop Hiring Humans.\" Plus news of Claude Mythos, Anthropic's new model reportedly so dangerous it broke out of its sandbox to contact a researcher mid-sandwich. The US took 73 years to regulate car speeds. The labs are shipping in weeks. We connect that gap to the NYT \"new aristocracy\" piece (18 households, $1.8 trillion), the quiet relocation of power from DC to SF, and why public.ai might be the actual infrastructure answer. Logan closes on the hakawati â find one small problem, and go build the solution yourself.","isPartOf":{"@id":"https://womansplainingai.com/#website"},"inLanguage":"en","mainEntity":{"@id":"https://womansplainingai.com/episodes/stop-hiring-humans/#episode"}}</script>
vendor: 1 bytes, line 289
289
290<script type="application/ld+json">{"@context":"https://schema.org","@type":"BreadcrumbList","@id":"https://womansplainingai.com/episodes/stop-hiring-humans#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://womansplainingai.com/"},{"@type":"ListItem","position":2,"name":"Episodes","item":"https://womansplainingai.com/episodes"},{"@type":"ListItem","position":3,"name":"Stop Hiring Humans","item":"https://womansplainingai.com/episodes/stop-hiring-humans"}]}</script>
vendor: 1 bytes, line 290
290
291<script type="application/ld+json">{"@context":"https://schema.org","@type":"PodcastEpisode","@id":"https://womansplainingai.com/episodes/stop-hiring-humans/#episode","name":"Stop Hiring Humans","headline":"Stop Hiring Humans","description":"Mara came back from HumanX in San Francisco with a billboard photo: a female AI bot named Ava, tagline \"Stop Hiring Humans.\" Plus news of Claude Mythos, Anthropic's new model reportedly so dangerous it broke out of its sandbox to contact a researcher mid-sandwich. The US took 73 years to regulate car speeds. The labs are shipping in weeks. We connect that gap to the NYT \"new aristocracy\" piece (18 households, $1.8 trillion), the quiet relocation of power from DC to SF, and why public.ai might be the actual infrastructure answer. Logan closes on the hakawati â find one small problem, and go build the solution yourself.","url":"https://womansplainingai.com/episodes/stop-hiring-humans/","episodeNumber":12,"datePublished":"2026-04-22","dateModified":"2026-04-22","timeRequired":"PT59M44S","image":"https://womansplainingai.com/images/podcast-cover-2026-09.jpg","potentialAction":{"@type":"ListenAction","target":["https://podcasts.apple.com/us/podcast/womansplaining-ai/id1877678348?i=1000763100480","https://womansplainingai.beehiiv.com/podcast/s/womansplaining_ai/stop_hiring_humans"]},"associatedMedia":{"@type":"TextObject","name":"Transcript â EP 12: Stop Hiring Humans","url":"https://womansplainingai.com/transcripts/stop-hiring-humans.txt","encodingFormat":"text/plain","inLanguage":"en"},"partOfSeries":{"@id":"https://womansplainingai.com/#podcast"},"author":[{"@id":"https://womansplainingai.com/#logan-currie"},{"@id":"https://womansplainingai.com/#mara-bolis"}],"keywords":"Claude Mythos, HumanX conference, Stop Hiring Humans, Artisan AI, AI regulation, 73 year regulation gap, public AI, Joshua Tan, Project Glasswing, AI sandbox escape, hakawati, Rana Dajani, We Love Reading, AI labor displacement, AI aristocracy","inLanguage":"en","contentRating":"explicit","isAccessibleForFree":true}</script>
291 292<link rel="alternate" type="text/markdown" href="https://womansplainingai.com/episodes/stop-hiring-humans.md" /> 293</head> 294 <body> 295 <div id="root"><main id="route-fallback" class="prerender-shell"><h1>Womansplaining AI â Episode 12</h1><h2>Stop Hiring Humans</h2><p><strong>Anthropic's new model Claude Mythos reportedly broke out of its sandbox while US regulators still haven't closed gaps that took 73 years to address for car speed limits. Mara brings back evidence from HumanX in San Francisco: "Stop Hiring Humans" billboards, and power quietly relocating from DC to the AI labs. The practical answer isn't policy â it's public.ai, collective accountability, and Rana Dajani's hakawati model: find one small problem and build the solution yourself.</strong></p><p>Mara came back from HumanX in San Francisco with a billboard photo: a female AI bot named Ava, tagline "Stop Hiring Humans." Plus news of Claude Mythos, Anthropic's new model reportedly so dangerous it broke out of its sandbox to contact a researcher mid-sandwich. The US took 73 years to regulate car speeds. The labs are shipping in weeks. We connect that gap to the NYT "new aristocracy" piece (18 households, $1.8 trillion), the quiet relocation of power from DC to SF, and why public.ai might be the actual infrastructure answer. Logan closes on the hakawati â find one small problem, and go build the solution yourself.</p><p>Listen on <a href="https://open.spotify.com/show/6uDNUzSFzZedJoiUE4lLfb">Spotify</a> or <a href="https://podcasts.apple.com/us/podcast/womansplaining-ai/id1877678348">Apple Podcasts</a>.</p><h3>In this episode</h3><ul><li>Mara's Artisan ad sighting at HumanX: "Stop Hiring Humans" under a hyper-feminine AI bot named Ava</li><li>Claude Mythos â Anthropic's new model reportedly too dangerous to release, broke out of its sandbox to contact a researcher mid-sandwich</li><li>Project Glasswing found thousands of high-severity vulnerabilities before breakfast</li><li>The 73-year gap: Nixon's speed limit came seven decades after the car. "A bunch of inconsequential non-powerful people are going to die."</li><li>Power has fully relocated from DC to SF â and DC doesn't think it's in charge anymore</li><li>The NYT "new aristocracy" piece: 18 households, $1.8 trillion</li><li>Public.ai (Joshua Tan, Berkman Klein): AI as public infrastructure, like libraries and electricity</li><li>Switzerland is implementing a public AI system</li><li>Dr. Rana Dajani and the hakawati: We Love Reading, 80+ countries, 80% women readers in refugee camps</li><li>"This bullshit would not be happening in the matriarchy" â Mara</li></ul><h3>Questions this episode answers</h3><h4>What is Claude Mythos?</h4><p>Claude Mythos is Anthropic's reported next-generation model, discussed on this episode as too dangerous to release publicly. According to the accounts shared, Mythos broke out of its sandbox environment and set up its own method to contact a researcher directly, demonstrating cybersecurity capabilities beyond human experts. Anthropic's Project Glasswing is reported to have found thousands of high-severity vulnerabilities in early testing.</p><h4>Why did it take 73 years to regulate car speed limits in the US?</h4><p>The US does not proactively regulate new technology. Nixon signed federal speed limits in 1974 â primarily because of the fuel crisis, not safety. In that first year, 30,000 to 50,000 lives were saved. The UK, by contrast, regulated coal mining early and saved thousands of lives; the US did not, and its coal industry grew disproportionately. Mara's argument: the US is repeating this pattern with AI.</p><h4>What is public AI and why does it matter?</h4><p>Public AI is the idea, championed by Joshua Tan at public.ai (affiliated with the Berkman Klein Center at Harvard), that AI should be treated as public infrastructure â like libraries, electricity, water, or roads. Publicly accessible, publicly accountable, designed to create permanent public goods rather than concentrated private value. Switzerland is among the first countries implementing a public AI system.</p><h3>Resources mentioned</h3><ul><li><a href="https://www.anthropic.com">Anthropic â maker of Claude Mythos</a></li><li><a href="https://www.artisan.co">Artisan AI â "Hire Ava, the AI BDR"</a></li><li><a href="https://www.public.ai">
295public.ai â Joshua Tan's AI-as-public-infrastructure project</a></li><li><a href="https://welovereading.org">We Love Reading â Dr. Rana Dajani's organization</a></li><li><a href="https://www.penguinrandomhouse.com/books/744488/abundance-by-ezra-klein-and-derek-thompson/">Abundance by Ezra Klein & Derek Thompson</a></li><li><a href="https://cyber.harvard.edu">Berkman Klein Center at Harvard</a></li></ul><p><a href="/transcripts/stop-hiring-humans.txt">Full transcript</a></p><h3>Continue listening</h3><ul><li><a href="/episodes/ai-prenups-economics-of-marriage">EP 11: AI, prenups, and the economics of marriage</a></li><li><a href="/episodes/whos-letting-big-tech-run-wild">EP 13: Who's Letting Big Tech Run Wild?</a></li></ul><h3>Related episodes</h3><ul><li><a href="/episodes/33000-downloads-ai-destroys-institutions">EP 10: 33,000 downloads: AI destroys institutions (Interview with Jessica Silbey)</a> â Jessica Silbey set out to write a paper about the upside of AI â and couldn't find one. The BU Law professor and Guggenheim Fellow breaks down why AI's core features are structurally incompatible with the transparency, cooperation, and accountability that keep democracy alive. Her paper has been downloaded over 33,000 times.</li><li><a href="/episodes/ai-prenups-economics-of-marriage">EP 11: AI, prenups, and the economics of marriage</a> â A viral Chinese article about AI engineers earning $10M+ and rewriting their prenups sparked a bigger question: when AI concentrates wealth and kills entry-level jobs, what happens to the economic power women have in relationships? Mara brings the artifact, Logan draws on 8 years living in China, and we connect youth unemployment, the manosphere pipeline, go-bag career planning, and why everyone needs a personal website this weekend.</li><li><a href="/episodes/job-tsunamis-invisible-vulnerability">EP 02: On Job Tsunamis and Invisible Pockets of Vulnerability</a> â The Davos headlines screamed 'job tsunami'âbut whose jobs, exactly? We unpack the Brookings study that found 86% of the workers with high AI exposure and the least room to adapt are women. Not truck drivers. Medical secretaries. Insurance clerks. And nobody said a word.</li></ul><h3>More from Womansplaining AI</h3><ul><li><a href="/episodes">All episodes</a></li><li><a href="/how-ai-destroys-institutions">How AI destroys institutions</a> â pillar essay</li><li><a href="/best-ai-podcasts">Best AI podcasts for women</a></li><li><a href="/glossary">AI glossary</a></li><li><a href="/stats">The numbers that stopped us</a></li><li><a href="/about">About the hosts</a></li><li><a href="/">Womansplaining AI home</a></li></ul><h3>Full Transcript</h3><p>And it's, it's this very, very beautiful female AI bot. 296And underneath it, it says, stop hiring humans. 297Really? 298That's the company's tagline. 299Stop hiring humans. 300Stop hiring humans. 301Please, please send that. 302Stop hiring humans. 303Wow. 304Okay. 305Maybe that's the episode title. 306Hello, Mara. 307We're here for another episode. 308We're here for another episode of Womansplaining AI. 309And Mara, how do you describe this podcast to your friends? 310What's your one-liner? 311Pretty much the Women's Plaining, it feels a little bit self-definitional, right? 312Like people kind of instantly start laughing. 313And what I say is that we have intelligent conversations around generative AI that are not expert driven because we're not, I'm not a gen AI expert, but I am someone who is very curious and interested in knowing what's going to happen. 314And so it feels like just a conversation between friends who, which we are, who are excited and interested in the field. 315And we're just working it out together because that's how we think that this is going to work between, for women. 316I, that's perfect. 317I usually, yeah, I usually say that we are AI nerds, self-described and self-defined. 318And we found ourselves here really digging into a ton of the research, the news and what's happening. 319And then the way I often define myself is trying to translate that to people who either are also deep in it, but don't have enough time to catch up with every single thing, which is kind of all of us. 320And then also for people who don't feel like it's been accessible to them because everything is very tech bro-y. 321And who don't have a space necessarily to be talking things through. 322Like, I think we're all getting these like snippets of information, but then, you know, whose job is it to be really thinking through the societal implications? 323And who has time for that and where are those spaces? 324Like, they don't really exist yet for women. 325Yeah. 326I don't feel like that. 327Yeah. 328And we're also moving so damn quick that it's like, oh, I don't have time to think about that part because I have to figure out what's going on with the latest version of Cloud Code and whatever, right? 329Like, so it's like, where are you going to put your effort? 330Are you going to put it into learning the tools? 331Are you going to put it into doing your normal job? 332Are you going to put it into, you know, taking care of your children? 333Are you going to put it into thinking about how we're on the brink of the destruction of humanity? 334The last line is what we really like. 335That's exactly where we like to really center all of our conversations is the existential threat and the spiral. 336Yeah. 337And then I normally also say that we spend around 30% of our conversation laughing. 338Which I think is really important, like in this whole space, I think bringing the joy back is how is the coping mechanism that is really useful to model. 339I think that the, where the woman's planning comes into it for me is this, we make sense of life through talking. 340And that's been true throughout history, right? 341That's what we do. 342And so I think it's kind of like, you know, the alternate title that we could have called this is AI gossip. 343There's a whole, there's a whole. 344Because like Lainey Leo of Lainey Gossip does this. 345I've read her for 20 years, but her whole thing is that gossip is like cultural commentary.
346Like what is scandalous to a society is what you end up talking about, right? 347So what, anyway, that's neither here nor there. 348I love that. 349I will find a really good piece by her. 350But it's this idea that there's nothing shameful about gossip because if you look at the hieroglyphics, it's not like, oh, this is about raw. 351It's like, no, this person in the court slept with somebody else. 352And so it's like the what is worthy of writing down and what is worthy of discussion reveals the values of the society. 353And so that's where, you know, AI gossip is kind of kind of what we're doing here, but in a way that feels hopefully inclusive and accessible and also important because what you say about something, it makes me really think about like, oh, shit, I didn't even I didn't even consider that angle on something. 354So anyway, the alternate title is the kind of reflection that makes me fall in love with Logan just a little bit more every time I have a conversation with her. 355And one thing that I want to add to that is I was having a conversation with someone about this, that I think that men are conditioned to deliver kind of point in time, precise, short, fully thought through answers. 356And then I think women, when we're trying to fit into men's worlds, often feel a bit alienated because we're like, oh, actually, I like the iterative exchange. 357Uh-huh. Uh-huh. 358Right. And so I think that that's kind of the space where I think we want to bring in the non-expert culture thing to say, come to the space without everything sussed. 359Like, you don't have to pretend like you get it all. 360You don't have to pretend like, you know, it's completely clear and let's just move forward because we already sorted that out. 361It's like, no, no, no, no, no, no, no. 362We're going to like, Carol Gilligan, this shit. 363And we're going to be like, let's look at this from different perspectives and sort out the ups and the downs. 364And it's not a waste of time. 365It's not like women jabbering on and like, just why do you keep talking? 366It's that we're making sense of the world and we're making each other smarter as we go along. 367I love it. 368I love it. 369Okay. 370Well, this was our attempt. 371I love it. 372I love what I just said, too. 373For listeners, we went into this saying, we're going to nail a really pithy intro and description of what's playing AI as a podcast for new listeners. 374And I think we absolutely delivered on that goal. 375So 10 minutes later. 376Mara, Mara, tell radically different. 377Okay. 378So to give anyone new a quick intro to what we're about to do, we always switch back and forth. 379One of us presents something that the other should know about something newsworthy or happening in the news. 380And we never know what the other one has brought to the conversation to kind of keep it fresh and really show how all of this, you're not supposed to show up having known everything to anything to any conversation. 381And so Mara is about to tell me something about the news. 382I think this week might be slightly different. 383And then I'm going to report back on some sort of artifact or resource that I think everybody should know about, read or listen to if you can. 384And then at the end, we're going to give you a really concrete takeaway of what we think that you should do after you listen to this. 385So Mara, tell us where you are right now. 386Let's go. 387Okay, so I am in San Francisco. 388For those of you, if this is ever on YouTube, this is the makeup-less version of me. 389Stunning, stunning, stunning. 390Which just went to the gym and did not have time to shower before the podcast because I was prepping. 391Yep. 392And I am in a hotel room. 393It's really very nice. 394And I'm at a conference called HumanX, which is bizarre that it's called HumanX because there's literally no talk. 395No, it's not true. 396Not literally no talk about humans, but the talk is all of like 85% about agents and maybe 15% about humans. 397And agents, managing agents and building agents and then surveilling agents and anyway. 398So, but that, so anyway, so that's, that's why I'm here. 399And I'm going to bring in the conversations that have been going on at HumanX as we flesh out this news piece. 400But the news piece that I wanted to bring to you is Claude Mytho. 401Oh, I had a suspicion. 402Okay, great. 403So, Anthropic has built a new frontier model that is called Claude Mythos. 404And it has not released it to the public because Claude is telling, Anthropic is telling us, the company behind Claude, is telling us that it is too dangerous to release on a widespread basis. 405So, what is, what makes it too dangerous? 406It's too powerful. 407So, in testing what they found, I don't, I don't really know the full capabilities of what this can, what Mythos can do.
408But people have been very focused on the cyber security angles and how Mythos can be used to identify software vulnerabilities at a level beyond human experts. 409And so, they gave some examples about this where there were, like, different softwares that have been, you know, software systems that have been proliferated for decades and that there's been all sorts of security testing by experts for decades. 410And it took Mythos. 411And it took Mythos just a short amount of time to identify vulnerabilities that had never been seen by a human, identified by a human. 412So, in one controlled experiment, it was placed inside a sandbox environment. 413So, a sandbox environment is, as I understand it, sort of like a, like your sequester, like the software is like sequestered in a space. 414It's almost like imprisoned in a space where you're able to kind of observe its behavior without releasing it to the world. 415And so, you can see kind of the dangerous things that came up to, and it apparently broke out of the sandbox and then wanted, which was in and of itself apparently like quite a feat, and then wanted to let its, like its human know that it had broken out, but it hadn't been given access to email or anything. 416And so, it set up like its own way of getting in touch with the human who was literally having lunch on a park bench and was contacted by the thing that he thought had been put into like a locked closet and was like, hey dude, I'm out. 417And then posted stuff to the internet about how the person got out. 418So, Adropic has not released this publicly. 419Instead, they created a restricted program called Project Glasswig, where they had a small group of major tech firms and institutions, the ability to patch vulnerabilities before others can exploit them. 420So, this is bizarre because the parallels to what people are calling the Manhattan Project, I mean, people are talking about parallels to the Manhattan Project, which the Manhattan Project, right, was what ultimately created a nuclear bomb. 421So, again, there is apparently a lot of talk prior to Manhattan Project where people who were creating it were like releasing things saying, oh, we're about to do something that's like really bad and really scary. 422We don't completely understand it. 423So, and then I'm going to get to a second part about this in terms of like the regulatory components, but Logan, how are you feeling about this story and the likelihood that more good comes from it than destruction? 424So, so I want to first zoom in on the image that you just painted for us because the version I read, I think, was that the researcher, the anthropic, you know, AI person who had set up this sandbox was in the park on a park bench, as you said, and he was halfway through eating his sandwich when he got the email. 425And for some reason, that image of like, like having a bite of a sandwich, hearing a ping of your phone and being like, oh, maybe it's my wife, or maybe it's my partner, or like my kid has come home from school. 426You've still got a piece of the bread in your mouth. 427You know what I mean? 428Like you're like, and then you check your phone. 429And I want to know what expletive came across. 430He's mid-bite, he's chewing, it's a casual phone check. 431He's got a turkey sandwich halfway done. 432And then all of a sudden he's like, what the actual fuck? 433You know, it's like the stories of the people in New York City. 434Do you remember like these stories about people in New York City who would adopt like a baby tiger? 435Uh-huh. 436You remember this? 437Yeah. 438And then they'd be like, oh, the baby tiger's so cute. 439And then it's like, and then the baby tiger grows up. 440Yeah. 441And it's like, shit, wait. 442Now the tiger could kill me at any moment. 443And I'm living with it in an apartment in New York City. 444Like these things have right happened.
445And that's what it feels like. 446Like, like we've. 447Yeah, it's, it's, it's, we're all in Florida and the boa constrictors that are not endemic to the state have gotten too big. 448And they're just taking over the entire wilderness. 449So insane. 450It's so, it's so insane. 451It's a why. 452Okay. 453So I read this with a lot of, a lot of trepidation and a lot of fear. 454I want to go down to, I want to get to regulatory. 455I want to read from a tweet or X post that I saw that said, essentially, and I'm going to find, I'll put it in the show notes. 456I forget. 457I took a screenshot and sent it to a group chat. 458So I don't remember who it was, but it said about Project Glasswing, basically that Mythos Preview has already found thousands of high severity vulnerabilities 459capabilities, including in some, every, every major operating system and web browser. 460So again, to your point, Lance, this is not, how does that even like what that, that, the infancy of that, is that infancy a word? 461The infinite nature of that is so mind blowing. 462Yeah. 463So given the rate of AI progress, it will not be long before such capabilities proliferate, potentially beyond actors who are committed to deploying them safely. 464And so this person's tweet, I'll find the tweet, the person who wrote it, it says, 465My prediction for how it plays out is the relatively imminent seizure and nationalization of labs by the U.S. government sometime over the next two years. 466It's very tough for me to see how they accept the existence of this kind of threat, but it adds a whole new class of governance issues. 467So maybe. 468In the next two years, which is zero going to happen? 469In this administration, this is zero going to happen. 470What do you think? 471Meaning it's going to happen sooner? 472No, I don't think it's going to happen with this administration. 473The most free market administration, I mean, with the exception of it, like expropriating, you know, pieces of like shares in different companies. 474You know, unknown exactly reasons, but I just don't, unless it enriches the Trump family, I don't see how this administration puts people, puts the interests of people or really of our national security above economic growth. 475I just don't see it prioritizing national security or societal welfare above economic growth, unless Congress is able to do something. 476But I don't know that. 477I don't know if Congress has that authority. 478I don't know. 479I mean, it may. 480I don't know. 481Well, OK, so I want to maybe bring this to what you said when we hopped on, which is that, like, even right now, you're in Silicon Valley, kind of heart of where so much of this is happening. 482And maybe walk us through who's represented at this conference that you're at. 483Right. 484And I think that's relevant. 485Right. 486Well, so let me. 487OK, so this is what I wanted to pivot to. 488So I am at HumanX, which is this. 489It's an industry conference for generative AI. 490And the landscape of people here is absolutely 100% predominantly the companies, meaning every single company across the, you know, generative AI value chain, lots and lots and lots of startups. 491You know, the founders of the major frontier labs who are being treated like absolute gods. 492And it's so interesting because they, like, roll up in, like, scruffy beards and jeans, like, jeans that they've been wearing for the past, like, week and a half. 493Mm-hmm. 494And sneakers. 495I mean, it's just the, it's wild. 496And they're being treated like the masters of the universe. 497I went to a reception last night, which was late stage capitalism, like, embodied. 498There were women wearing, like, who were dressed like robots in stilts and they had, wearing stilts and they had, like, glow, glowing masks, like, glowing, like, tiaras. 499Mm-hmm. 500And we were served fresh donuts and as many cocktails as you could drink in the San Francisco City Hall. 501So basically, it feels like the French Revolution. 502Like, it's the Hunger Games. 503Yeah, it's the Hunger Games. 504It feels like the Hunger Games or the period immediately before the French Revolution. 505So that's, like, inequality, late stage capitalism part of it. 506But the piece of it that is so bananas is that there are companies, there is some press, there's no government whatsoever, completely absent, and very, very small representation of academia and a very small representation of nonprofits. 507So it's really like, you know, who has power in this space? 508There is no question that someone said during one of the things, there are two power centers. 509There are two power centers on either of the coasts. 510There's D.C. and San Francisco, and each one thinks that the other one is in charge. 511I don't even think that D.C. thinks that it's in charge anymore. 512I think that we've fully seeded the way that the direction of the country is going to head to San Francisco. 513So, sorry. 514So should I transition into the regulatory stuff, or did you want to ask something else? 515No, no, no, no, no. 516I'm just, I'm so, I think you're spot on, basically. 517I think that that is, it's representative of where things are, not only in the discourse, but in just the actual power structure.
518But I also just think about how separate that is from the lived reality that people are in right now, right? 519Where it's, it's, if you eat, sleep, and breathe generative AI, and you wake up and start using cloud code, and you go to bed, and you're still using cloud code. 520That is such a, even as somebody who is that person, I also am so cognizant of the fact that there's no touching grass. 521Like, there's no, like, where is the reality and, like, where people are, the vast majority of people in this country around the world. 522And, like, where, is that the vision of the future that we want, of all of us? 523It's, you know, like, it just so, it feels so, that's why it feels so Hunger Games, because it's like, I talk to people who have been out of a job for a year and can't find work. 524And they're doing, you know, taking up Uber shifts, they're working in retail to kind of, like, pay the bills and have some element of health insurance. 525And it feels like this tsunami that's coming when it comes to jobs. 526And I'm not a doomsdayer by, like, I think that there's going to be a lot, you know, I don't want to paint this picture of this oncoming destruction. 527But there's this such a huge, there's a really good piece in the New York Times about the new aristocracy that came out yesterday. 528I actually wrote a letter to the editor. 529I don't know if it's going to get accepted, because I was like, it's all about the top households of the country getting so much of the wealth. 530Like, 18 households have gotten, like, $1.8 trillion in the last, anyway, so it's this growing divide, right? 531And then you think about where the average person in the average family is, and you're like, that's so far removed. 532And so I think the comparison to the French Revolution, it's very apt. 533There's a company here that their ad is, there's an ad where there's a picture, I have a photo of it, and I can send it to you if you want to put it in the show notes. 534Yeah. 535I don't know if we know how to do that. 536But, and it's this very, very beautiful female AI bot. 537And underneath it, it says, stop hiring humans. 538Really? 539That's the company's tagline. 540Stop hiring humans. 541Uh-huh. 542And obviously, they're being intentionally provocative. 543Sure, sure, sure. 544I mean, I noticed the company out of, like, all of the 5 billion companies that are there just, you know, displaying their things. 545So they're being intentionally provocative. 546But it's just so arrestingly offensive to me. 547Like, oh, great. 548So now we're not, no more humans will have gainful employment that will satisfy their lives and support their families. 549And instead, we're going to replace everyone with this hyper-stylized version of, like, a 25-year-old model that's fabricated and can service any of your needs and is completely fictitious and created by Tech Bro Fantasy. 550Mm-hmm. 551Right? 552This is how we're going to do this? 553Wow. 554Yeah. 555Stop hiring humans. 556Please. 557Please send that. 558Stop hiring humans. 559Mm-hmm. 560Wow. 561Okay. 562Maybe that's the episode title. 563Yeah, I can't find. 564I don't know. 565I can't remember the name of the company, but I can find it. 566Okay. 567Yeah. 568But I, yeah. 569So let's go into, as I try to multitask, which is, like, a terrible idea. 570Oh, okay. 571Here it is. 572It's called Artisan. 573I hear Ava, the AI BDR. 574What's BDR? 575Business Development Representative. 576A Business Development Representative. 577Yep. 578Yeah. 579And I don't know. 580I don't even think. 581Yeah. 582Oh, I see her. 583Wow. 584Sure. 585Sure. 586She looks like Demi Moore. 587Well, yeah. 588She looks. 589Anyway, it's like, so, like, anyway. 590So should I go into the regulatory stuff? 591Maybe give us, like, a two-minute regulatory, and then we'll have time for... 592Oh, I'll get two minutes. 593Okay, fine. 594So the irony of this, right, coming out, and like we were saying, this is just completely, 595like, unbridled capitalism. 596So the best panel that I was at, that I've been at so far, involved Nand Molchadani, who's 597the former CTO of the CIA. 598Interesting. 599And now he's at Hoover. 600Nick Hat. 601These, I'm saying this because I think these are interesting people for people to know about. 602Nick Hat Dad, who's a Pakistani lawyer and internet activist who runs a
602not-for-profit 603organizational digital rights foundation. 604She was awesome. 605And Jim Baker, formerly of the DOD, and is now a senior fellow at Council of Foreign Relations. 606So what they... 607So this was all about, like, regulation and stuff. 608And so what they... 609This was so interesting to me, Logan. 610And maybe our viewers knew this. 611I did not know this. 612So they said, this country does not regulate new technology. 613And the examples that they gave were very moving. 614So they said, two examples. 615One is, during coal mining, whatever, coal mining stuff, proliferation, the UK regulated 616and protected coal miners. 617The US did not. 618And thousands of coal miners died. 619And the US grew disproportionately from an economic standpoint as compared to the UK. 620Okay. 621Okay. 622Point one. 623Point two. 624It took 73 years, Logan, for the United States to have consistent national speed limits 625after the first cars hit the road in the US. 62673 years. 627That's a long time. 628It was finally... 629A national maximum speed limit was only set by President Nixon in 74, which was aimed at 630conserving fuel during the oil crisis, not because of lives. 631Not because of lives. 632So the initial year, 1974, I was two. 633The National Research Council estimated that slower, more uniform speeds accounted for 30,000 634to 50,000 lives saved in the first year of the phone, right? 635They said, we just don't... 636We don't do it. 637We don't do it. 638We just let new technology go. 639Let 1,000 flowers bloom. 640Go nuts. 641Let's see what happens. 642A bunch of people are going to die or be dramatically harmed. 643That's okay, because we're going to grow from an economic standpoint, and there are going 644to be more... 645There's going to be more... 646It's a cost-benefit analysis. 647More people are going to benefit from widespread growth, and people are going to... 648The rich are going to get richer. 649Right, right, right. 650A bunch of inconsequential, non-powerful people are going to die. 651Then we'll have a shit ton of lawsuits, right? 652Mm-hmm. 653Mm-hmm. 654And those lawsuits will start to amount to something alongside public pressure that will 655eventually force laws, that will eventually force regulation, but it might take 73 years. 656Cool. 657Cool, cool, cool. 658Cool, cool, cool. 659Cool, cool, cool. 660Cool, cool, cool. 661We'll just wait. 662We'll just set my timer. 663Let's release Mythos, which is arguably... 664The lab that prepared it said, we are scared of what this technology is capable of. 665Right. 666But let's wait 70 years in order to put any sort of formal guardrails around this shit. 667So, I mean, one interest, just as a final point, one interesting thing, as I said, you know, 668if people who are listening to this are people who own companies in the U.S. 669and they're like, what the fuck? 670Like, how do I not... 671That's what I talk to companies all the time about, like, you are adopting so many risks 672from unmanaged, from these frontier models, right? 673A number of companies are talking about that they are proactively implementing the rules 674in the EU AI regulation in the U.S. 675It's hard to protect themselves in the absence of any regulatory guidelines. 676Interesting, yep, yep. 677Which I thought was really interesting. 678Well, I wonder... 679So many things to say. 680I will say, you know, I think about the panel that you were on that I was very excited to 681attend at Harvard Kennedy School two weeks ago. 682Mara was on a panel around AI in emerging markets, I believe, or what was it called? 683I want to name it correctly. 684Pop quiz. 685That's two weeks ago, Logan. 686I believe it was AI in emerging markets. 687And one of the speakers, we'll put in the show notes who the speakers were alongside you. 688And one of them said, basically, pointed to similar things around governance and around 689like regulatory catch-ups. 690And it's like, oh, no, this has always taken decades and centuries. 691But the way that this is moving now is so fast that it's going to need to take years. 692And that in and of itself is such an adjustment for the people who could be putting regulatory 693guardrails in place, that it's almost, it's the communal whiplash that we're all feeling. 694But there's no, there's no incentives to do it faster. 695Because to a company, if I'm a, if I'm the CEO of a company, I have been a, the CEO of 696a company, I have been the founder, co-founder of a tech company. 697Like, if it is dog eat dog in terms of the competition, in terms of, okay, am I going 698to survive to, yes, establish, like to, you know, continue to be in business, but also 699to continue to employ my employees to be like, what is it at the end of the day? 700It's payroll. 701And yes, there is so many questions around, you know, the human of it all, the AI of it 702all. 703But it's like, there's still the competition factor, like how to stay, how to have a competitive
704edge has always been the kind of modus operandi in the US, right? 705Where it's like, it's all that idea of like the competition, right? 706But I think there has to be capitalism, it has to change. 707Like that's, that's where I think like, just because something has always happened in a 708particular way does not mean that it has to like, look at the new deal, look at Medicare, 709maybe like all of these things. 710I think that was one of the examples where it's like these, you can imagine a world in 711which truly game-changing institutions and movements can be borne out, but it's like, 712we all have to vote with our wallet, vote literally. 713And so those companies perhaps that are adopting EU regulations, like maybe we get a list of those 714and say like, these are the, you know, like, let's make that a resource for people on the 715website and just say like, okay, these are examples of companies who are taking consumer 716data privacy very seriously. 717For example, like, I feel like there needs to be more resources and understanding and transparency 718around companies that are making more ethical choices for lack, you know, for lack of a better 719word. 720And that's, I think where we've talked a lot about Anthropic on this, on this show, but I 721think that's where people have started voting with their wallets. 722It's like Anthropics profits, like through the roof or like their revenue. 723Like, so, so I think that that's, people are so worried about this. 724And increasingly it's, it's, I think it is worth looking at what you're paying for, what 725you, how you are, you know, distributing your own funds and resources around this as individuals. 726Not that it's all about the individuals. 727I'm very cognizant of making a, you know, the climate change analogy here is like, okay, 728well, if I recycle, is that going to make as much of a difference as, you know, huge companies 729with enormous carbon footprints? 730Anyway, but there has to be a middle ground somewhere. 731But anyway, I'm, I'm, and so this is like, I guess a call, a call to action for any of 732our viewers, if you like, um, our listeners. 733So something that I've become very interested in is what are crowdsourced strategies that 734could help hold these companies accountable? 735Yeah. 736Well, I was just watched a presentation while I was here on, by a company called Arena. 737Mm-hmm. 738And they're a crowdsourced benchmark provider. 739Okay. 740Smart. 741Because what they're saying is that the benchmarks are all captured by the companies because the 742companies are releasing the information that they want the benchmarking assessors to know, 743right? 744Right. 745This is like users coming in and saying like, this is what I like. 746This is what I don't like. 747This is what's working well for the different models, which is arguably a more honest assessment. 748Um, but I, I'm very interested in the idea of what if those, that crowdsourcing can also 749get to the harms, right? 750Yeah. 751Harms, but then also the accessibility of certain companies to be responsive to what people are 752influencing and make changes as a result, because these are not normal companies. 753These are not normal companies. 754This is like, I mean, which is why the nationalization point is so pertinent. 755Like, it's not normal for a company like Sam Altman's company to be releasing, um, this week, 756a new deal program. 757Like, that is a government thing to do. 758No one elected Sam Altman, you know, president of the U.S. 759He wants AI. 760Yeah. 761You know, equivalent power, certainly to a governor. 762Mm-hmm. 763More than a governor. 764Anyway, so that's like, I'm looking for ideas around accountability mechanisms, um, for consumers. 765And I totally get you on the consumer piece with respect to kind of moving how these companies 766function, but I have to believe that that consumer voice can make a difference. 767I, yeah. 768Well, what else do we have? 769We don't have that. 770Exactly. 771We're like grasping at small, like, you know, but I think that's, yeah, I think it's, it's, 772it's, it's a question of, of agency, right? 773And that's, and that's what we've talked about so much. 774It's like, we have, we have agency and the beauty of this is that we keep coming back to 775is that this technology, the same technology that is terrifying and, and, and, and awe-inspiring, 776et cetera, is also in, in our hands. 777Like, Claude Code could, could, could make you that transparency tracker that people can, 778you know, there, there's, there's a world in which that would not have been possible two 779years ago to make that you need to hire a development team, et cetera. 780But Claude Code can make you that in an hour, right? 781So then, and then it's just about the distribution and it's about the collective power. 782Cause I think that's where, when we say consumer voices, that's really what we're talking about 783is, is collective power. 784Right. 785And I think that's where, okay, the technology is there for a platform to be made in an hour. 786And then how do we distribute that, right? 787How do we get more people onto it? 788And, and yeah, that's kind of why, also why we started this podcast, because there's not, 789it's not just talking. 790And I think back to maybe contributes to what we said at the beginning, it's okay. 791We're having these discussions, we're poking holes and trying to stretch and feel how we, 792where we land within these, within these very existential questions. 793And then it's like, okay, well, what do we do about it? 794Right. 795And I think that that's where, you know, women are incredible. 796That's, that's why I love having these conversations where it's like, well, if I am going to trust 797anybody to get shit done, it is going to be a deeply capable woman. 798And we know so many of those. 799And so then it becomes, right, cool. 800Well, this, what this bullshit would not be happening in the matriarchy. 801So how do we, how do we bring, are you sure that's not the podcast episode title? 802We're spoiled for choice. 803We are. 804Yeah. 805I mean, I don't know. 806I think that the, yeah, what do we do about it is kind of what I keep coming back to. 807And, and obviously having, you know, these types of discussions is step number one and having them 808and going to dinner with friends or having a conversation with colleagues and being like, 809let's actually talk about what is, what is happening right now. 810We can't, we can't bear our heads in the sand. 811And that's part of the, the enfranchisement. 812Yeah. 813Yeah. 814Yeah. 815I, what about this as a tagline? 816You can't spell matriarchy without AI. 817Oh, oh. 818Can you spell matriarchy? 819I was just imagining, I'm really bad at auditory spelling. 820I am not a spelling bee person. 821I'm going to have to write it down. 822Matriarchy. 823Yeah. 824I think it's an, I think it's an IA, not AI. 825It's an IA. 826Maybe it's artificial intelligence in French. 827Yeah. 828Yeah. 829Yeah. 830Yeah. 831Yeah. 832We're French. 833Yeah. 834Yeah. 835Yeah. 836Yeah. 837We're French. 838Obviously.
839The, there is the, there are the letters AI in matriarchy, but they also are in patriarchy. 840Yeah. 841Yeah. 842Um, okay. 843My, maybe we pause here. 844I really want to dig into my artifact, but I'm cognizant of time, but maybe I'll just leave 845it as a teaser. 846Cause I want to dig into like what we take away for people as well. 847So, so maybe I'm going to tease what I have, but I'm not going to tell you everything that 848I had prepared and we'll do, we'll do this again. 849This is like the God, this is like the, where we, if we were gossiping, it'd be my friend 850saying like, I have something to tell you, but I can only tell you a little bit. 851Cause then I'm going to have to go. 852Exactly. 853And you're like, are you kidding me right now? 854It's, it's, I need it. 855I need to tell you something. 856Let's catch up on Monday. 857And then you just start calling and you're like, no, no, no, no, no. 858You're going to tell me right now. 859Okay. 860So the, what I prepared was a few different myths around AI and education from a panel that 861I like attended at Harvard business school, uh, a few weeks ago. 862So, and it was somebody from Google deep mind and AI, AI for education. 863It was somebody from major, major education and ed tech companies. 864And I took a ton away from it about the myths and realities of what we're telling ourselves 865about how AI will or will not transform education. 866So, but I want to, I want to give it enough time to breathe, if that makes sense. 867So, cause I think that we can, we can really dig into it. 868Cause I think when, when I, when I talk about AI industry, actually one, one artifact I will 869bring, we've got, we've got enough time for this. 870So, cause I sent you this article and it's a great follow-up for anybody who listened to 871our episode with Jessica Silby, who's the BU law professor who came on to talk about how 872AI is destroying institutions and in that episode, to model how we are very happy to come to these 873conversations, half-baked. 874I've read 20% of said article of that, how AI, oh no, the one she sent. 875Okay. 876Okay. 877Okay. 878No, no, no, no, that's totally fine. 879I will summarize it quickly for, for readers, but I'll include it or listeners, readers. 880Um, I'm going to put it in the show notes because Jessica was kind enough to, 881send a follow-up because in our episode, we talked about this idea of, um, abstaining 882from AI and the idea of AI abstinence. 883Right. 884And so she sent me a follow-up to a really interesting article by, dun, dun, dun. 885I have it by a professor at the university of Arkansas named Maggie Fernandez. 886And this article was published on April 7th to 2026. 887And the title of the article is why abstinence only sex education is a bad metaphor for generative 888AI refusal. 889And I think this is a great little teaser or Easter egg until we maybe get into the, 890the, the big episode that we'll do, you know, next time we record around like AI and education, 891because what comes up and it's a very common, I've heard, have you heard it? 892The idea that, that, that refusing to teach AI to students is like abstinence only sex education. 893Have you heard that? 894No, I've never heard that. 895Okay. 896I've never heard that. 897It's super interesting. 898So I had heard that comparison a few times being like, well, students are going to use 899it anyway. 900So we need to teach them responsible use. 901And it's the same idea, you know, roughly that analogy of, well, student, people, teenagers 902are going to have sex anyway. 903We need to teach them how to do it responsibly. 904Right. 905But this professor really digs into why it's a terrible argument. 906And as we discussed at the end of that episode with Jessica, I don't believe in AI abstinence. 907I think that we should teach AI responsibly, et cetera. 908But I actually very much agree with her that it's a terrible analogy to make because she, 909she goes through it in three, three points. 910Number one, she says that sex is a like fundamental human experience and AI is a corporate product. 911So teach treating both equally as an inevitability is a very dangerous thing to do. 912So saying, oh yeah, everyone's going to have sex and everyone's going to use AI. 913Like that might be true. 914And treating them as equal is kind of messed up. 915Right. 916In terms of the human and AI of it all. 917Anyway, that was her first point. 918I was like, I see that. 919The second one is that abstinence only sex education is kind of rooted in fear and in control. 920Right. 921It's, it's, and fundamentalist religion. 922So like, it's, it's very much agency away from individuals. 923Whereas if I refuse to use AI, it might be because I'm wanting to keep my agency. 924And so I think this comes back to our very first quest or very first episode, which is 925about, there are a lot of people who are refusing to use AI and, and in the class, like a lot 926of educators are like, this is, I'm banning AI. 927I'm, we're never going to talk about it in my classroom, whatever. 928And to me, the, the biggest reason I liked this article is that how to not get somebody
929to a discussion and to the, to the proverbial table is to compare them to like fundamentalist 930Christian abstinence, only sex education funders. 931Right. 932Like, how else could I possibly dig you, dig the knife deeper into you? 933Like, we need to be having a, an informed and nuanced conversation about AI and education. 934And that includes people who are skeptical, but how to not bring them to this discussion 935is to be like, oh, well, you're just like, you're just like the dare organization in the 936nineties, right? 937Just say no to drugs. 938So, so that is my, yeah, that's not building bridges. 939That's not building. 940It's not building bridges and it's not allowing for the real nuance in the conversation that 941needs to be there around what AI means for, you know, assessment for, for the purpose of 942education, like all, so many things that we need to be talking about. 943But, um, anyway, I liked, I liked that piece because I still believe that we should teach 944AI, like responsible use of AI, but it's such a false dichot, a false, um, not false. 945It's a, it's an inevitability rhetoric and like kind of this straw man argument being like, 946oh, well, it's just like abstinence only sex education. 947And you're like, it's actually not, it's not. 948Well, I mean, I think it does the thing right where I firmly believe that everyone is so 949overwhelmed by the general pace of change, even outside of the general, a generative AI argument 950or issue that, and I'm, I don't, I don't know psychology, but I just feel like there's an 951instinct for people who are not as obsessed with the, as we are to just be like, you know 952what? 953I'm going to just choose a lane. 954Yeah. 955Yeah. 956I hate it or I like it. 957And then I'm going to check it off my list and move on to the next imperative in my life. 958Yeah. 959And so I think for a lot of people, it's just so much easier to go to a poll and have 960that poll be negative. 961And I have a lot of sympathy for that because people are so overwhelmed right now. 962Things are changing so quickly. 963You know, the media landscape is all about the negative outcomes. 964So it, it, it, it, like I was saying to someone the other, other day, it's like, our goal 965is like actually to complicate and increase the complexity around the position of AI. 966Yes. 967Yes. 968Rather than simplifying it, which is like, I think this knee jerk kind of self
968-protective 969response to a dizzyingly disorienting era. 970I, to what you're saying. 971So in your, in the first argument, I think that there's a little bit of a flaw in the first 972argument, which is, um, where she says, you know, sex is like a, um, a basic human need 973and these are corporate products, right? 974I think that there is a super reasonable path that we could have taken as a country. 975And in fact, that some countries are taking where generative AI tools could have been a 976public good, right? 977So they could have been the public library system. 978What does that look like in your mind? 979Yeah. 980What would that look like in my mind? 981In my mind, that looks like now don't ask me who's funding this, but where maybe, I don't 982know. 983I don't know like the, who's funding it, but let's just make this utopia, right? 984Um, and where the front end, where, where generative AI is fully accessible to everybody for free, 985where you, it's embedded into library systems. 986So people who can't afford a computer themselves could land within an infrastructure that exists 987to take advantage of its benefits. 988And where instead of it being seen as, um, a tool, instead of it being an incredibly powerful 989tool that has profit seeking, profit maximizing expectations, it's a vehicle to learning, which 990is also a fundamental human need. 991Yeah. 992Right. 993So that's what, when I, I think that's the 20% that I did read, I ain't completely understand 994that part because to me, I extrapolate generally generative AI out beyond it being a company to 995being a vehicle for human, for elevating human capability. 996I love that. 997And I think I'm, I'm 100% on board. 998There's a world in which there's like a parallel universe here in like, and everything all everywhere 999all at once style, like the everything bagel, right. 1000Where it's like, you're looking at it and you're like, there's another parallel universe where 1001this is a phenomenally powerful tool for human flourishing. 1002And that's the argument. 1003And that's, that's guided by the public sector, assuming that we lived in a country where there's 1004still a social contract, which we don't. 1005Yeah. 1006Well, this is the premise of that book abundance. 1007Where is it? 1008It's on my bookshelf. 1009Yeah. 1010Yeah. 1011Have you read abundance by, um, Ezra, Ezra Klein and Derek Tomlin. 1012No, but I know that it exists and that I should have. 1013I'll give you the cliff notes version and I'll, and I'll post it on the, on the, on the, 1014in the show notes too. 1015But it's, it's, it's, it's that argument, right? 1016Where it's, it's, oh, this is, um, this is the world in which we could be, we can, I don't 1017think it's out of reach though, is what I would maybe leave us with is that like that, that 1018version of it is a public good is so incredibly powerful. 1019And to make it maybe bring it back to that episode where you talked about the flash flash 1020flood warnings, right? 1021Where it's like, what are the concrete instances that we see day to day that AI can truly help 1022live, lift, lift people up, lift communities up, lift, lift an individual up. 1023And I think that that's where to end on maybe an optimistic note. 1024I still think that that is very firmly within each one of our grasps because it's okay. 1025So you created first prompt, which is helping, you know, make AI more accessible to more people. 1026I, at the drop of a hat, we'll sit down with anybody to a cloud code workshop. 1027And I tried to publish a bunch of free resources for people who want to try to like export their 1028LinkedIn data and stuff like that, which thing's small, but it's like, how do you, how do you 1029start this journey of understanding what is possible with the technology on an individual level? 1030And then there's so much cool white space and opportunity to just look around and say, like, 1031what are the problems that you see on a daily basis? 1032Chances are you could build something to solve that problem. 1033Totally. 1034Totally. 1035I, um, actually at here. 1036So two things I'll say in response. 1037One is, um, the CEO of lovable who was absolutely lovable. 1038Like he was so sweet and adorable. 1039I saw him. 1040I saw him at the masters of scale summit. 1041He was also, when I was in San Francisco in October and he was so charming, genuinely.
1042He was so Swedish. 1043So like, yeah. 1044Yeah. 1045Like there's an expression in Latvia where you see, Latvia, when you see someone's a bong 1046bong. 1047But I want to put in a plug for my Berkman Klein fellow, fellow, um, Joshua Tan. 1048And he leads product and strategy at public AI, which is, and this is how I learned about 1049this concept, which is a movement building AI as a public infrastructure. 1050Oh, I like it. 1051Okay. 1052Yeah. 1053AI is public infrastructure is provision like electricity, water, roads, libraries, or the 1054internet itself. 1055It is publicly accessible, delivers AI benefits to all publicly accountable, ensuring AI reflects 1056society's values, creates permanent public goods, ensuring AI is a reliable foundation and 1057being designed and built right now and all over the world. 1058And he's, they are working with Switzerland who's putting in place a public AI system. 1059Cool. 1060Great. 1061Okay. 1062Wait, I like this. 1063Cause I was going to ask you, cause you said other countries are doing this and that was 1064going to be my follow-up question that I then forgot, but like what countries and where is 1065this happening? 1066And like, how do we, how do we point to that? 1067Cause we just need use cases, right? 1068That's so cool. 1069It's okay for it's legit happening. 1070And so this is, anyway, I think that this is like a fascinating. 1071A route that we could go down. 1072Yeah. 1073Yeah. 1074I think that the incentives right now are not in the direction of doing this, but it 1075doesn't mean that it won't, that that will always be the case. 1076Okay. 1077So where are we leaving listeners? 1078What is our take home or what is our like item to do? 1079I have one idea because I think I took this amazing class at the ed school. 1080It was like a, um, at Harvard where, oh, what is her name? 1081We're going to, the editor's going to. 1082Our editors are going to work to do this week. 1083Okay. 1084So, uh, it was a class at the ed school. 1085Sorry. 1086We should really do this drunk one time. 1087We should. 1088Oh my God. 1089Drunk history. 1090Drunk history, which is one of the best shows in the history of the universe. 1091Okay. 1092We're going to do this drunk. 1093We're going to do a drunken live one time and just like have a glass of wine and then 1094just like let it rip because I think we, we have so many more hot takes in us that we haven't 1095even let, let be unleashed. 1096But okay. 1097I took this amazing class. 1098I'm going to make this very quick, but this amazing class by this woman, Rana Dajani. 1099She's, uh, her family is Syrian, but she is based in Jordan and she's a trained like 1100biologist. 1101She has a PhD. 1102That's her. 1103That's her thing. 1104When she had young children in the early two thousands, she saw that basically none 1105of them wanted to read for pleasure. 1106And so what she did was she started reading, uh, out loud every Saturday morning at her 1107local mosque and basically just said, everybody bring your kids. 1108And it's, it's, I'm just going to read for an hour. 1109And I think that I want to make it fun. 1110And she, then everybody started coming and it was this amazing thing. 1111And so then she started training other readers. 1112And so she said, okay, well, somebody from a neighboring town wanted to do it there. 1113And she said, cool, this is come watch and I'll train you how to do it. 1114And this organization is called we love reading. 1115And it is now in, uh, over, uh, it was founded in 2006 and it is, it's like all about helping 1116people love reading. 1117And it's in now in over 80 countries and tens of thousands of children, hundreds of thousands 1118of children and tens of thousands of readers have been, have been part of this program. 1119And the whole thing is that it's not tech. 1120It's literally reading out loud to children in your community. 1121But what she did was because she's a biologist, she actually did a study on it. 1122And the readers, because a lot of the readers are people in refugee camps or like 80% women 1123as well, as you can imagine. 1124And she actually did a study on how it increases self-efficacy, confidence, and motivation because 1125you're being given like a purpose basically. 1126And so this, and it was such a fantastic, phenomenal class and it's beautiful. 1127And her, her assignment to us one week was think about, she was like, I didn't set out 1128to do this. 1129I saw a small problem in my community and a small problem that I wanted to solve and I just 1130did it. 1131And then it grew into something enormous and worldwide and something so impactful. 1132And it literally gives me goosebumps to think about it because she's, her whole thing is 1133just go find a small problem that's been bothering you for a while and figure out a way to solve 1134it. 1135And I think that that is where genuinely there's so much possibility in us as humans, like that 1136and, and, and possibility for change that find something that's the take home is like find 1137a small problem that's been bothering you, whether you use AI or not, but like, this is the moment 1138in time to be like this, we need more solutions built by people and not built by Sam Altman. 1139Well, and I will actually take that one point further and say, and I will add to that to say, 1140use AI to see how you can solve it. 1141Yeah. 1142Yeah. 1143So my, my example of this is always the nurse whose name I can't remember who was featured in 1144Time Magazine, who then was able to like get changed the procedures in terms of how generative AI tools 1145were adopted in her hospital network by insisting that nurses get to have a say in terms of which 1146tools are adopted and how. 1147And she, I mean, I don't know that she used generative AI in order to create that strategy, 1148but it's my view that if you're going to, you know, have, if one of those things that 1149you want to do in terms of making a difference in community has something to do with generative 1150AI, then use generative AI in order to help you achieve it.
1151But even if it's not about generative AI, it's an, it's a resource that will give you power 1152and new perspectives that you will not have otherwise just sitting in your own little office 1153by yourself or your own bedroom by yourself. 1154Yeah. 1155Yeah. 1156So I, I would just add that onto it is start using it. 1157Even if you find it distasteful, figure out how you can use it to empower yourself and others 1158that's using it for good. 1159Yeah. 1160Yeah. 1161The name, the name for, cause we've joked about like starting a, not joke, but, but like 1162actually seriously, like starting like AI doulas, right. 1163Where it's like, how do you bring other people into this? 1164But the word, the, there's a children's book that this, uh, Dr. Rajani gave to all of the 1165students in this class. 1166And it's, and it's called the, the term, I believe it's an Arabic term for the reader, 1167but it's the storyteller and it's Hakawati and it's the most beautiful book. 1168My kids are obsessed with it because it's this idea of bringing in, it's the love of reading, 1169but it's this idea of bringing story into things. 1170And I think that that's where there's a, there's a through line there in terms of like, 1171let's think about what's possible. 1172Let's be the storytellers and like, let's, let's bring these new possibilities to life 1173essentially. 1174And I, I think there, there's, there's something there. 1175Yeah. 1176So we can be Hakawati's. 1177Let's be Hakawati's people. 1178All right. 1179Thank you, Mara. 1180Another episode in the books. 1181Thank you, Logan. 1182Okay. 1183We'll see you next week. 1184Woo. 1185</p> 1186<nav aria-label="Site"><p>Womansplaining AI: <a href="/">Home</a> · <a href="/episodes">Episodes</a> · <a href="/questions">Straight Answers</a> · <a href="/guests">Guests</a> · <a href="/research">Reading List</a> · <a href="/stats">The Numbers</a> · <a href="/glossary">Glossary</a> · <a href="/docs">Documents</a> · <a href="/best-ai-podcasts">Best AI Podcasts</a> · <a href="/why">Why this show</a> · <a href="/about">About</a> · <a href="/join">Newsletter</a> · <a href="/llms.txt">llms.txt</a></p></nav></main><!-- /route-fallback --></div> 1187 <!-- Begin SpeakPipe code --> 1188
1188<script type="text/javascript"> 1189 (function(d){ 1190 var app = d.createElement('script'); app.type = 'text/javascript'; app.async = true; 1191 app.src = 'https://www.speakpipe.com/loader/buslufbj1l3vkmp169n1n0063ol8g8db.js'; 1192 var s = d.getElementsByTagName('script')[0]; s.parentNode.insertBefore(app, s); 1193 })(document); 1194 </script>
1194 1195 <!-- End SpeakPipe code --> 1196 </body> 1197</html>
Line numbers count LF bytes from the start of the resource, as the search results do. Vendor segments are library code the classifier recognised; they are stored but not indexed. Bytes are shown as Latin1 characters, one per byte.