AdExchanger

A handful of genuinely interesting observations - the LLM-answer-pasted-back-into-Reddit feedback loop and the GEO skepticism - but the episode is padded with Cannes small-talk, a mid-roll ad break, and generic Reddit advocacy. Novel ideas per minute is low.

“There's this new kind of behavior which is, okay, get, maybe get an LLM answer, and then I might cut and paste that. It's almost like a new post. Type into Reddit and say, well, what do you think?”

“I am skeptical that there's like a program where you do X and get Y and have a formal”

The framing of LLM outputs as syntheses with 'an objective function of your retention and happiness' is sharp and uncommon, and GEO skepticism is a mildly contrarian take for a platform that benefits from it. But the bulk of the episode is standard Reddit self-promotion and broadly recycled AI-disrupts-search narratives.

“It's a synthesis sometimes articulated as a perspective. With an objective function of your retention and happiness.”

“I am skeptical that there's like a program where you do X and get Y”

Jen Wang is a genuine senior operator - COO of Reddit for eight years through its IPO - with real institutional knowledge of a platform central to the AI training data conversation. However, she is visibly in Cannes PR mode, deflecting on deal specifics and keeping answers safely promotional.

“We didn't know that, um, Reddit would be such a high portion of citations”

“I started eight years ago”

The host surfaces concrete deal values ($60M Google, $70M OpenAI) and the Mission Mindset report yields two usable stats (79%, 62%), but the guest volunteers almost no original specifics, deflects on negotiations, and the named figures come from a commissioned WPP co-branded study rather than raw internal data.

“it's 60 million a year with Google, right. And 70 million with OpenAI”

“79% of Redditors say they do more research before making a purchase now than they did a year ago”

The host demonstrates real topic fluency - correctly identifying the LLM feedback loop dynamic and pushing back with 'But it's kind of a new creature' - but also self-answers frequently, lets vague deflections pass unchallenged, and spends the final minutes on German memes and gardening subreddits.

“But it's kind of a new creature.”

“What a world, though. So Reddit is feeding LLM answers, and then people are taking those answers and putting them back into Reddit”