This week, Kimberly Becker and Jessica Parker dig into the “AI bubble”—why it keeps inflating even as skepticism grows inside the industry.

We unpack the growing disconnect between massive investment and unclear payoffs, including a widely discussed Goldman Sachs research question: what $1 trillion problem will AI actually solve? From there, we connect the dots between two very different narratives:

• Dario Amodei’s essay framing “powerful AI” as an imminent civilization-level risk—and a reason to race ahead (carefully… “to some extent”).
• Cory Doctorow’s argument that this is a familiar tech bubble pattern, with a predictable ending—and that we should focus on what can be salvaged from the wreckage.

Along the way, we define what makes a bubble a bubble (and how this one differs from dot-com), talk about growth-stock dynamics and why no one in power wants to be responsible for “popping” it, and explore what AI hype looks like when it hits real workplaces—especially through Doctorow’s concept of the reverse centaur: a human reduced to a machine’s accountable appendage.

We also go nerdy (in the best way): training corpora, “WEIRD” cultural assumptions baked into data, model-collapse fears from AI eating AI-generated output, and why the internet itself feels increasingly polluted by synthetic text patterns.

In this episode:

• The “$1T problem” question and why the AI ROI story feels thin right now
• Why “AI is inevitable” functions like a strategy (not a neutral prediction)
• Growth stocks vs. mature companies—and the incentive to keep inventing the next hype cycle
• Reverse centaurs, liability, and why “AI replaces jobs” often means “humans take the blame.”
• “TINA” (There Is No Alternative) as a trap—and a demand dressed up as an observation
• Corpus 101: what it is, why it matters, and how bias shows up in “universal” models
• Model collapse / photocopy-of-a-photocopy: when AI trains on AI outputs
• Regulation talk that centers on “economic value” (and whose value that really is)
• Pit & Peach: slowing down, pausing, gratitude, and building without growth pressure

Sources:

• Goldman/AI bubble discussion (Deep View): https://archive.thedeepview.com/p/goldman-sachs-publishes-blistering-report-on-ai-bubble
• Goldman Sachs “$1T spend” framing: https://www.goldmansachs.com/insights/top-of-mind/gen-ai-too-much-spend-too-little-benefit
• Amodei essay: https://www.darioamodei.com/essay/the-adolescence-of-technology
• Doctorow (The Guardian): https://www.theguardian.com/us-news/ng-interactive/2026/jan/18/tech-ai-bubble-burst-reverse-centaur
  
  

Leave us a comment or a suggestion!

Support the show

Contact us: https://www.womentalkinboutai.com/

In this episode, Kimberly and Jessica debrief Jessica’s interview with Arlyn (founder of Tobey’s Tutor) and unpack what it looks like to build AI products as “non-technical” founders. They reflect on their own journey building Moxie: bootstrapping vs raising money, the pressure-cooker effect of investors, the messy realities of UX/UI and platform migration, the world of APIs and subscriptions, and why “friction” can be an ethical design choice, especially in AI for education.

In this episode, we talk about

• Why “non-technical founder” is a misleading label
• The hope in AI (and how “both can be true”: benefits + harms at once)
• Bootstrapped “mom-and-pop” AI companies vs venture-backed growth expectations
• The founder reality: burnout, delegation, and why money changes decision-making
• The startup metrics whirlwind: LTV, CAC, churn, stickiness, payback period
• What building an AI product costs in practice: tools, subscriptions, and constant ops
• UX/UI psychology: heatmaps, “rage clicking,” onboarding friction, and conversion decisions
• Why “friction” can be good (consent, safety, pacing, limits, especially for kids)
• “Building on rented land”: what happens when OpenAI/Google/Anthropic change terms
• The bigger ethical question: solving a problem vs optimizing a broken system

Suggested listener action

If you’re building, using, or researching AI in education: reach out. And if you’re using AI tutoring with kids (or yourself), ask questions about data, limits, mistakes, and oversight.

Leave us a comment or a suggestion!

Support the show

Contact us: https://www.womentalkinboutai.com/

In this episode of Women talkin’ ’bout AI, Jessica sits down with Arlyn Gajilan, founder of Tobey’s Tutor, an AI-powered learning support platform she originally built for her son, who has ADHD and dyslexia.

This conversation is a deep dive into what it actually looks like to build an AI product as a non-technical, bootstrapped founder, from vibe coding and early prototypes to onboarding, safety systems, and pricing decisions.

Jessica fully geeks out with Arlyn as they unpack:

• Building AI to solve a deeply personal problem
• What “vibe coding” can (and can’t) do
• Designing responsibly for children and learning differences
• UX vs. UI decisions that matter
• Bootstrapping, pricing, and intentionally staying small
• Why “AI wrapper” criticism misses the point
• The reality of building while parenting and working full-time

Mentioned in the Episode

• Tobey’s Tutor: https://tobeystutor.com/
• Scientific American (article mentioning Tobey’s Tutor): https://www.scientificamerican.com/article/how-one-mom-used-vibe-coding-to-build-an-ai-tutor-for-her-dyslexic-son/
• Mobbin (UX/UI inspiration library); https://mobbin.com/
• Empire of AI by Karen Hao: https://www.penguinrandomhouse.com/books/743569/empire-of-ai-by-karen-hao/
  
  

Leave us a comment or a suggestion!

Support the show

Contact us: https://www.womentalkinboutai.com/