The Edtech Podcast

There are a handful of genuinely useful ideas - flipping AI into a Socratic questioner, moving AI ownership from CTO to CPO, the compounding-context inequality point - but they are buried in extended biographical narrative, platitudes about change management, and meandering safeguarding tangents that add little operational value.

“asking the AI to ask you questions rather than getting it to answer questions... ask me one question at a time, waiting for my answer in between, to help me think through this problem”

“the responsibility of rolling out AI needs to move from the chief technology officer to the chief people officer... it is a people problem. It is not a technology problem”

The 'ask AI to question you' reframe and 'brain fry' from agentic AI are moderately fresh angles, but the bulk of the conversation - AI as conversation partner, don't treat it as search, executives need to lead, ban policies backfire - recycles widely circulated takes without adding a genuinely contrarian or first-principles twist.

“a single almond the production of a single element would get you about 30,000 queries on chat GPT”

“there's this concept called brain fry... if you're constantly at the highest part of your intellectual capability... by 10 a.m 11 a.m like i'm already overloaded”

Joshua is a legitimate practitioner - built and sold Super Awesome to Epic Games, now runs an AI transformation company serving Home Depot, Delta, Pearson - which grounds his claims in real operator experience; he is not a career podcast guest, though he is not a tier-one name in AI research or enterprise transformation.

“Some of the biggest companies in the world, Home Depot, Hyatt, Delta Airlines, Pearson, Fortnum Mason”

“that became one of the biggest kids technology companies in the world... ultimately acquired by Epic Games, the creators of Fortnite”

There are concrete reference points - the almond/30,000 queries comparison, the sub-4.5-year skill half-life, the 2% data-centre energy figure, named client companies - but most claims are asserted without sourcing, and the single-almond statistic in particular is dropped without attribution or verification.

“the average half-life of a skill was just going down and down and down and was getting close to about four and a half years at the time”

“the entire world's data centers combined are less than 2% of the world's energy. And that includes AI, which is less than 10% of that”

The host leans heavily on affirmation and biographical cheerleading rather than probing follow-ups; unsupported claims like the 1% figure, the almond statistic, and the 'brain fry' concept pass unchallenged, and the energy/water correction - which deserved real scrutiny - is quickly dropped in favour of agreement.

“I mean, already I'm absolutely in awe of the things that you've been able to do”

“Amazing. And I love that”