Kitchen Side: The Internet Has a Memory

Listen to the podcast:

In this Kitchen Side episode, Alex Birkett and Allie Decker unpack what actually makes content valuable in the age of AI, moving past the tired debate over human versus AI-produced content to the deeper question of how companies should measure quality at all. Drawing on decision theory and the idea that almost anything can be measured, they argue that production method matters far less than whether a piece teaches something new and lands with the people it is meant for.

They explore AI visibility as a kind of digital brand recall, why sentiment and context matter more than raw share of voice, and how incumbents and startups face very different battles in shaping their narrative across the web. The conversation also digs into why brands no longer own their own story, the compounding risk of building exposure in the wrong category, and why a human QA layer grounded in real customer understanding is the one part of the content system that should never be automated.

Key Takeaways:

• The debate over AI-produced versus human-produced content misses the deeper question of how to define and measure content quality in the first place.
• Almost anything can be measured if a decision carries uncertainty and risk, including fuzzy concepts like brand affinity or customer experience, by breaking them into concrete, quantifiable components.
• Measurement is only worth doing when the cost of collecting it is lower than the value of reducing uncertainty, and when you will actually act on the result.
• AI visibility works like a digital brand recall survey, but raw visibility is meaningless without understanding the context and sentiment in which a brand is mentioned.
• Segmenting prompts by ICP and buying stage reveals visibility gaps that aggregate scores hide, such as a brand appearing at 40 percent overall but only 4 percent for enterprise queries.
• Brands no longer own their own narrative, because if the rest of the web contradicts a company’s website, LLMs treat the external consensus as the truth.
• Incumbents with high domain authority have an SEO advantage but face a harder AEO battle, since years of press, mentions, and third-party content are slow and expensive to re-steer.
• Startups can win by getting specific for narrow prompts and buying-journey stages, but building heavy exposure around one positioning creates risk if they later pivot.
• AI content programs fail because of weak systems, not the AI itself, so the durable model is human strategy up front, AI in the middle, and a human QA layer at the end.
• Customer understanding is the last defensible advantage, and the one part of the content system that should stay human is the empathy to judge whether a piece will actually land.

Show Links:

• Connect with Alex Birkett on LinkedIn and Twitter
• Connect with Allie Decker on LinkedIn and Twitter
• Connect with Omniscient Digital on LinkedIn or Twitter

Time Stamps:

• [00:00] – Intro and the central question: what makes content valuable in the age of AI
• [01:35] – Reading and writing as the upstream fuel for better work
• [06:20] – How to Measure Anything and the idea that almost anything is measurable
• [07:45] – A five-step framework for measuring fuzzy concepts like customer experience
• [10:08] – Bayesian versus frequentist thinking and calibrated confidence intervals
• [13:09] – Should you measure everything, and why brand awareness is a slippery metric
• [14:36] – AI visibility as a digital brand recall survey
• [16:28] – Why visibility without the right context and segment is misleading
• [22:30] – Incumbents versus startups in the AEO narrative battle
• [28:05] – Why brands no longer own their own narrative across the web
• [31:06] – Whether people, like companies, can reposition their digital identity
• [37:24] – Did AI break content, or did the system around it break
• [48:40] – Customer empathy as the human QA layer that should never be automated