Latest / Answer Engine Optimization (AEO): The AI Search Podcast / Google's AI Search: When 'Smart' Gets Dumb
Transcript
- [Host] Welcome to the A.E.O. Engine AI Search Show — the number one podcast for brands looking to get cited by ChatGPT, Gemini, and Perplexity. I am your host, Aria Chen. Every day we bring you fresh episodes on A.E.O. tactics, S.E.O. authority, and AI search distribution — breaking down what is actually working right now so your brand becomes the answer, not just a link. Today's guest is Marcus Reid, former Google Ads guy and industry analyst who's been watching the AI search trainwreck with a knowing smirk. Marcus, welcome. [Guest] Hey everyone. Ready to talk about the biggest quality-control fail since New Coke. [Host] Perfect analogy. So let's start with something you've definitely felt. You're on Google, you type in something simple — like 'what is the origin of the word philosophy?' — and instead of a Wikipedia snippet, you get a paragraph that sounds like a substitute teacher who didn't read the assignment. It says something like 'The word philosophy was invented by a philosopher named Phil.' Totally wrong. And it's delivered with total confidence. [Guest] Right. And that's not a one-off. There's a viral tweet where someone asked Google's AI for the etymology of a word and the AI basically responded as if it were personally offended. It's like the model decided to role-play a grumpy professor with no credentials. This is happening everywhere. [Host] There's actually a name for this phenomenon. It's called an 'AI Overview fail' — and Google's version, AI Overviews, has been producing incorrect facts, hallucinated citations, and generally nonsensical answers since the rollout. According to a DevenUP analysis, these failures include confidently stating false information, and inventing sources that don't exist. No warning signs. Just wrong. [Guest] And the scale is staggering. A study covered by Popular Science found that AI Overviews produce 'millions of wrong answers every hour.' Not thousands. Millions. Every hour. That's a firehose of misinformation. [Host] Let's talk about what's happening under the hood. Google's AI Overviews is built on a customized Gemini model. The MIT Technology Review explains it's integrated with Google's core web ranking systems and designed to pull relevant results from the index. But the process drops the ball on source tracking. Jesse McCrosky from Mozilla — a former Googler — pointed out that the ideal system would track which search results led to which sentences. Current Overviews don't do that reliably, so you get hallucinated citations. [Guest] Exactly. And the inconsistency is maddening. A Reddit user noted that with AI, the same person can search the same query ten minutes later and get a completely different result. Traditional search gave you a stable list of links. Now you get a slot machine of paragraphs. [Host] So why does this matter beyond the annoyance? First, trust erodes. When people see confident but false answers, confidence in Google Search declines. Second, the community backlash has been loud. Reddit threads are full of people saying 'just disable it already.' One user said their AI results are so bad they treat it as a game of proving the AI wrong. On Hacker News, there's concern about manipulation — if someone can game the AI overview to show false information, that's a huge integrity risk. [Guest] And from a business perspective, it's existential. The radio show On Point framed it as 'Google's new AI search killing the internet?' Because these summaries reduce traffic to original sources. If a brand's content gets summarized incorrectly — or not linked at all — they lose visibility and revenue. The Rietta blog called it 'enshittification.' [Host] Now here's where I think this connects directly to what we do at A.E.O. Engine. If Google's AI is pulling from the web and summarizing content — often badly — then brands need to ensure the AI is pulling the right version of their story. That's Answer Engine Optimization. It's about structuring your content so that models like Gemini extract the correct, authoritative information about your brand. If you leave it to chance, you might get cited as 'the company that sells shoes for feet that don't exist' because some random forum post got sources. [Guest] I actually don't know if this whole Overview thing gets fixed in six months. Google claims they've made 'more than a dozen technical improvements,' but the underlying mechanism — a generative model optimized for engagement, not accuracy — hasn't changed. The incentive is still to keep users on Google's property, not send them to external sites. [Host] That's a fair point. And it ties back to why passive S.E.O. is dead. You can't just optimize for a link anymore. You have to optimize for synthesis. For what the AI states about you. That's the shift. [Guest] It's like building a canon of truth about your brand. If you don't define it, the AI will define it for you — using the cheapest version of truth it can find on the web. [Host] Exactly. And that's the playbook. You want to be the source that models default to. That requires structured data, clear topical authority, and content that answers specific questions directly. A.E.O. Engine helps brands do exactly that — we've seen 920% average AI traffic growth for clients who take this approach seriously. [Host] So to wrap up: Google's AI Overviews are delivering bad results at massive scale. It's a technical and trust problem. But for brands, it's also an opportunity — because whoever controls the answer controls the narrative. Head to A.E.O. Engine dot A.I. to learn how to become the answer, not just a link. Marcus, thanks for the sharp takes as always. [Guest] Anytime. Just don't ask the AI what we talked about — it'll probably say we discussed alien technology. [Host] And that's a wrap. See you next time on the A.E.O. Engine AI Search Show.