The Content Strategy Podcast · Kristina Halvorson

Episode 61: Morgan Marie Quinn - Getting to grips with large language models

October 10, 2023·46 min·4 clips
Morgan Quinn, a content design leader at Google, explains how large language models learn like babies.
AI is the juicy topic. Christina starts with the ordinary mess of recording logistics before bringing in Morgan Marie Quinn, a content design leader at Google working across BARD and Google Assistant. The tone stays close and chatty. Morgan comes in through a winding career path: mommy blogging, ServiceNow, Compass, Intuit, and now large language models. Then the conversation gets into the hard part. Human behavior is hard to predict. So is an LLM response, especially when the content being generated feels organic. In that work, finished is a slippery word. Morgan explains that even something like stopping a conversation can be improved, but it does not become a neat checked box. Christina keeps nudging the topic back to real work. She says LLMs are not built by one person at a desk with ChatGPT, then reaches for the cake joke because, honestly, that lands. Who has to be involved when a large language model gets made? Morgan's answer is wider than the familiar UX, PM, and engineering setup. Data teams, operations, marketing, research, linguists, and more people come into the room. Christina sounds genuinely surprised by parts of that. Linguists make instant sense to her. Marketers and researchers make the whole operation feel bigger, and a little harder to picture. Morgan describes the work as all hands on deck, collaborative, and experimental, the kind of place where different kinds of expertise have to rub up against each other. Christina keeps the technical talk from getting sealed off. She corrects ChatGPT when it is wrong, calls that responsible internet citizenship, and lets the joke do some useful work. The close is about humility. Christina marvels at Morgan's restraint around work that might make someone else yell from mountaintops, then catches herself on confidentiality and her own self-employed distance. The episode leaves LLM work feeling less like magic or hype, and more like messy organizational craft that has to be revisited again and again.

As heard by us

A grounded, useful look at LLM work as collaborative product labor, led by a Google content design perspective.

Christina Halverson turns a buzzy AI topic into a practical conversation about how large language models get built inside product organizations.

Read the full review in PlayNext →

Why you'd press play

You want the cross-functional team behind LLMs, not a glossy AI myth.

Read the full recommendation in PlayNext →
Listen to the show on