AI Empowered

If your favorite AI model disappeared tomorrow, would you lose a week of convenience, or would you lose your business's memory?

This month Aaron moved off Claude as his primary AI, switched off every scheduled task it ran, and found that some of what his business knew lived only in that one tool's folder. He uses the family-systems idea of overfunctioning, where a household organizes itself around one member until that member steps away, to explain why going deep with one model quietly creates the same trap. He also covers OpenAI's retirement of GPT-4o from ChatGPT and research that caught providers changing the models behind their endpoints. Then he lays out the rule he now runs on: a plain-language Brief and Decisions doc as the single entry point for every client and project, with a replacement drill you can run this month.

This show extends the book AI Empowered. Read the first chapter free at aiempoweredbook.com, or get the book on Amazon.

A Harvard AI tutor more than doubled what an active-learning classroom taught. A chatbot handed to high schoolers left them seventeen percent behind kids who never had it. Same class of tool. The difference was what each design made the human do before the answer arrived.

This episode returns to chapter six of AI Empowered, the collaboration demands more of you, not less, and pays off the half of the argument that "The Deskilling Trap" (July) left standing alone. That episode used Bastani et al.'s PNAS field experiment (about 1,000 Turkish math students) to show what unscaffolded AI costs when the tool goes quiet. This one adds Kestin et al.'s crossover RCT in Scientific Reports (194 Harvard introductory-physics students), where a GPT-4 tutor built on the course's own pedagogy, and told to guide rather than answer, produced more than double the median learning gains of the in-class version, in less time, with higher reported engagement. Set side by side, the two studies isolate design, not model capability, as the variable.

Aaron carries both studies' boundaries in the script: Kestin measured immediate gains only and says nothing about retention or deskilling; it compared two scaffolded designs, not AI against human, and no plain chatbot arm was tested; the at-home, self-paced delivery is confounded with the tutor; the material sat at the understand-apply-analyze level, not synthesis or professional judgment; and both samples are students, so the extension to knowledge work is the book's inference and is labeled as such. The work anchor is a design rule in the spec for the AIRAA (pronounced "era") app's coaching feature: ask what the user thinks first, ask for evidence, refuse to write the work product, end every exchange with a decision the user makes. Dependency on the system is treated as a failure mode. The turn to the listener is one audit question for one workflow: what does it make the human do before the answer arrives?

Studies:

Earlier episode: The Deskilling Trap (July 2026).

---

This show extends the book AI Empowered. Read the first chapter free at https://www.aiempoweredbook.com, or get the book on Amazon.

Aaron Douglas runs Auspicious, a fractional marketing practice.

Blog: https://auspicious.llc/insights/ 

Nine months into an AI rollout, the dashboard says eighty-something percent weekly usage, training completion in the nineties, time-to-first-draft down forty percent. Every number is accurate. Then somebody pulls three client proposals from before and three from after and reads a paragraph from each out loud, and you could swap the client names without anyone noticing.

This episode draws on chapter ten of AI Empowered, the chapter about organizations, and pairs it with a working paper posted in August called "How Organizations Use AI: Evidence from ChatGPT" — the largest look yet at enterprise AI telemetry. Three of its five authors work at OpenAI, two are academics at Columbia and Wharton who did the work as paid contractors for OpenAI, and it studies OpenAI's own product. Seventeen hundred and sixty-four companies, seventeen and a half million messages, measured six months after adoption. It documents how fast usage is growing, what kinds of firms adopt early, who inside a company actually uses the tool, and what they use it for.

And then, in its own limitations paragraph, it says the thing worth sitting with: the task results "do not measure downstream work products, productivity effects, or changes in organizational routines." A dataset that size can tell you exactly how much people typed and nothing about whether anything got better. The paper's own summary line is the one to keep: adoption is only the beginning of deployment.

This show extends the book AI Empowered*. Read the first chapter free at https://www.aiempoweredbook.com or get the book on Amazon https://a.co/d/082wePlH 

Aaron Douglas runs Auspicious a fractional marketing practice.

Four out of five organizations are running AI training. Three out of five still report an AI skills gap. The training is happening; the gap isn't moving. That isn't a quality failure — it's a targeting failure.

This episode works the argument from chapter two of AI Empowered: the three dominant responses to the AI collaboration gap — education, prompting technique, and change management — each address something real, and each aims at the wrong layer. The live element is a Microsoft field experiment with 388 employees at a Fortune 500 retailer, published this spring, where everyone got the same AI tool and only the structure around its use was varied. The structured behavioral protocol was associated with *lower* document quality and substantially less work finished. Partnership training that reframed AI as a thought partner was associated with higher quality — but only at the top of the distribution, and the authors are candid that their belief-change result likely reflects carry-over rather than the training itself.

Also covered: why the social cost of AI adoption (a measured nine percent competence penalty from peers, twenty-six percent for women engineers evaluated by male non-adopters) is a barrier no communication plan reaches, and what turns up when you go looking for a product that teaches the layer underneath — trust calibration, knowing when to override, what you personally bring that the model can't generate.

The question to stop asking is "have our people been trained." The question to start asking: when someone on your team opens an AI tool, what do they believe their job is in that exchange?

---

This show extends the book AI Empowered. Read the first chapter free at https://aiempoweredbook.com, or get the book on Amazon.

Aaron Douglas runs Auspicious, a fractional marketing and AI empowerment practice.

Researchers gave twenty-five different AI models the same prompt — write a metaphor about time — and collected fifty responses from each. Over a thousand generations came back as essentially two ideas. Different companies, different continents, same sentences: one measured pair of models overlapped 81 percent of the time, down to identical phrasing.

This episode takes the "I Am the Scarce Input" argument from chapter five of the book and collides it with what researchers at the University of Washington, Carnegie Mellon, and the Allen Institute for AI are calling the Artificial Hivemind — plus a Trends in Cognitive Sciences review of more than 130 studies showing that AI makes individuals more productive while making groups less original. If the models are converging on the same average, the only differentiating input left in the stack is the human one. Aaron walks through the deference trap that produces generic output, why the style guide running his own content system amounts to a field guide to the hivemind, and a one-question Monday-morning test for whether you shipped your judgment or shipped the average.

Referenced in this episode:

- "Artificial Hivemind" study (University of Washington, Carnegie Mellon University, Allen Institute for AI): https://the-decoder.com/study-warns-ai-could-homogenize-human-creativity-as-models-converge-on-artificial-hivemind/

---

This show extends the book *AI Empowered*. Read the first chapter free at https://www.aiempoweredbook.com or get the book on Amazon https://a.co/d/082wePlH.

Learn more about Aaron Douglas and his work at https://auspicious.llc

This episode takes the deskilling trap from chapter six of the book ("The Collaboration Demands More of Me, Not Less") and collides it with what the research is saying right now: a Nature news feature on early deskilling evidence across medicine and software engineering, the Lancet colonoscopy study behind it, and the survey showing 70 percent of nurses and 77 percent of physicians worry about losing skills to AI reliance. The researchers say no established solution against deskilling exists yet. Aaron argues the early shape of one has been sitting in the scaffolded-engagement research all along — including the Wharton experiment where the group forced to think first improved 127 percent and kept every gain when the AI was removed, while the easy-access group ended up worse than they started. Plus the story of pointing two AI models at all fifty-eight of the book's arguments before printing, and what that adversarial review demonstrates about using AI to sharpen judgment instead of replacing it.

Read the first chapter free at https://aiempoweredbook.com or get the book on Amazon https://a.co/d/082wePlH.

Learn more about Aaron Douglas and his work at https://auspicious.llc.

Everyone has the same AI tools. A few people get results that look like they're using something else entirely. The difference has a name now — and a book.

AI Empowered is out in paperback and Kindle, and this episode marks what it means for the show: a new episode every few weeks, each working one idea from the book live — variations the book didn't have room for, new research that pushes on the argument, patterns from real client rooms.

Not chapter readings. This episode demonstrates the format with the idea at the book's center: you are the scarce input, the two directions people fail when they miss that, and the task-type research finding that changes how you delegate by Monday morning. This show extends the book AI Empowered. Read the first chapter free at aiempoweredbook.com, or get the book on Amazon.

Blog: https://auspicious.llc/insights/

X/Twitter: https://x.com/aaron_douglas

LinkedIn: https://www.linkedin.com/in/aaronddouglas/

AI Empowered is shifting how episodes get made — and this one explains why. Aaron talks about what's been working, what hasn't, and what the show looks like going forward.

The short version: episodes are moving from abstracted frameworks to grounded stories from Aaron's week-to-week work as a fractional CMO and marketing leader. Two examples from recent client engagements illustrate the difference: a department lead staring down a CEO's AI mandate with no process to build on, and a founder watching a six-figure AI integration sit untouched because it made invisible work visible. These are the kinds of moments the show is built around now.

Same themes — human-AI collaboration, change management, why smart people resist useful tools — but sourced from the work itself rather than outlined in advance. Deeper-dive episodes aren't going away, but they'll be anchored in something specific. Something that actually happened.

If you've been listening, this is what's next. If you're new, good timing.

---

Blog: https://auspicious.llc/insights/

Two conversations keep showing up in every organization working through AI adoption. The first: an executive stands up and declares "we need to be using AI across the board." Licenses get bought, emails get sent, nothing changes. The second: a respected director quietly routes around every AI initiative for a year, not because she's anti-technology, but because the tool threatens the system she spent a decade building.

Both reactions make complete sense emotionally. Neither moves anything forward. The mandate creates motion without direction. The resistance creates stability without progress. And the organization develops a split personality where the board deck says "AI-enabled" and the actual work stays exactly the same.

In this episode, I walk through both patterns from real engagements (anonymized), explain why they reinforce each other, and describe what a productive conversation actually looks like. Spoiler: it's almost boring. One process, one team, one real problem. No company-wide email required.

If your organization is stuck between "use it for everything" and "we tried it, wasn't a fit," this is the episode.

AI Readiness & Adaptability Accelerator — https://airaa.org

AI Empowered Book - https://aiempoweredbook.com

Have you ever noticed that some people dive headfirst into AI while others hang back — and wondered why it feels so personal? In this episode, we reframe AI adoption as a classic Hero’s Journey where every one of us is the protagonist. You’ll meet four very human archetypes — the Pioneer, the Guardian, the Steady Hand, and the Pragmatist — each with their own wants, obstacles, and real stakes, and discover how recognizing these different chapters turns friction into strength and resistance into real human-AI collaboration.

Learn more about Aaron Douglas at https://www.linkedin.com/in/aaronddouglas/ Subscribe to the podcast for more insights, read AI Empowered Book - https://aiempoweredbook.com or Accelerate your AI Readiness at https://airaa.org/.