ATD Accidental Trainer

The three impact zones and the 14-question implementation canvas are modestly useful reframes, but the episode is heavily padded with biographical backstory, book promotion, and ATD Intensive plugs. The ratio of genuinely novel ideas to filler is low for a 36-minute runtime.

“sometimes I worry that we spend so much time talking about zone one and zone two, that we're not taking the time we save in zone one and zone two to go engage in zone three”

“Zone three is when the business uses AI to change how it does business, how a process happens or a new offering or something. And in that zone, L&D is not leading the conversation.”

The impact-zone segmentation and the WIsE@AI acronym are minor original contributions, but the underlying arguments - trust matters for adoption, L&D has transferable skills, AI ethics needs governance - are well-worn takes recycled in AI-for-L&D discourse. No genuinely contrarian or first-principles claims are made.

“Tech adoption lives at the intersection of psychology, strategy, and design”

“We can give everybody a tool. Woo-hoo, everybody logged into Copilot. Yay. But are they using it? Do they trust it? Do they trust it appropriately?”

Megan Torrance is a legitimate practitioner - founder of TorranceLearning, former Andersen Consulting/Accenture, three published technical books - and draws on real client engagements. She is not a celebrity thought-leader, but her credibility comes from actual project delivery rather than conference-circuit positioning.

“pre-generative AI, one of my clients was an insurance company, and they brought in a machine learning algorithm to do near immediate underwriting of life insurance plans below a certain limit”

“We just used it at a very large project. It was a zone two, so AI in front of the learner where our team mediated conversations with huge subject matter expert”

The insurance underwriting anecdote and the internal burnup-chart automation failure are the only concrete, grounded examples; both are described without metrics, timelines, or dollar figures. The rest of the episode operates at framework and acronym level with minimal named companies, data points, or measurable outcomes.

“they brought in a machine learning algorithm to do near immediate underwriting of life insurance plans below a certain limit. It was very carefully boxed, right? And we had a very large training program, went all around the country”

“every spreadsheet is stored in a slightly different place on the network. It's not consistent because some of our projects are six years old and some just started last week”

The host is clearly a colleague and collaborator of the guest, resulting in a warm promotional chat rather than a rigorous interview. Questions are mostly open invitations to explain pre-planned frameworks, with no pushback on vague claims and frequent affirmations that stall genuine depth.

“Oh, I love that idea.”

“I love it, Megan, because when, I'm doing exactly what you said that people do”