Feb 12, 2023 · 34m · another-podcast
ChatGPT versus Google
0:00 / 0:00
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
Tech analysts Benedict Evans and Toni Cowan-Brown examine how generative AI challenges Google's long-standing search monopoly, evaluating the profound disruptions across product accuracy, advertising economics, compute infrastructure, and digital publishing.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. The hosts hold 99.8% of the talking time here. How this is scored →
The hosts as informed peer 7.0 Guest teaching 0.1 Guest disagreement 0.1 The hosts pushing back 0.1
speaking balance: gold is the hosts, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 8:50 Rejecting generalized search universality
Evans challenges the prevailing assumption that generalized LLMs seamlessly apply across all categories without complex manual taxonomy.
Hardest push from the hosts ▶ 31:10 Dismantling publisher link monetization claimsEvans dismisses past newspaper demands for link royalties as fundamentally flawed because search engines provided traffic.
Biggest teaching moment ▶ 16:23 Educating on search ad keyword valuationEvans surprises Cowan-Brown by explaining why asbestos personal injury litigation has historically commanded the highest search ad rates.
The host holds their own ▶ 27:50 Technical breakdown of device power budgetsEvans demonstrates deep domain insight by explaining why the battery cost of cellular radio transmission outweighs local silicon compute.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | The hosts as informed peer | Guest teaching | Guest disagreement | The hosts pushing back | Why |
|---|---|---|---|---|---|---|
| The Three Search Dimensions and Library Query Analogy | 7 | 0 | 0 | 0 | Evans breaks down search query taxonomy using a structured three-tier library analogy (navigational, topical, synthesis) and explains the manual engineering required for structured answers. | |
| The Problem of Confident Hallucinations and Domain Risk | 7 | 0 | 1 | 0 | Evans outlines the confident hallucination problem using vivid comparative examples between image generation and high-stakes medical queries. | |
| Vertical Search Habits and Conversational User Interfaces | 6 | 1 | 0 | 0 | Cowan-Brown brings up vertical search behavior across YouTube and TikTok, while Evans frames the interactive multi-turn dialogue as a UI paradigm shift. | |
| Evaluating Verifiability, Intent, and Probabilistic AI | 7 | 0 | 0 | 0 | Evans establishes a 2x2 conceptual matrix evaluating domain verifiability against the consequence of factual errors, referencing GitHub Copilot. | |
| Rethinking Search Monetization and Ad Placement Models | 7 | 0 | 0 | 0 | Evans explains search monetization mechanics, citing the historical ad value of asbestos litigation queries and the friction of integrating ads into synthesized text. | |
| Disruption Dynamics: The Dreadnought and iPhone Analogies | 8 | 0 | 0 | 0 | Evans draws on naval history with HMS Dreadnought and the 2007 iPhone launch to explain how technological resets render incumbent advantages obsolete. | |
| Infrastructure Costs, Capex, and Edge Computing Strategy | 8 | 0 | 0 | 0 | Evans details data center capex, inference costs, Apple local silicon strategy, and radio transmission power budgets versus on-device processing. | |
| Hardware Upgrade Cycles and AI Compute Constraints | 6 | 0 | 0 | 0 | Evans analyzes how local generative AI workloads could revive consumer hardware upgrade cycles by making silicon performance gains perceptible. | |
| The Zero-Click Web, Publisher Traffic, and Copyright Tensions | 7 | 0 | 0 | 1 | Evans contrasts historical publisher complaints about search links with the legitimate copyright challenge posed by zero-click AI synthesis. |
Statements from this episode (14)
Evans: Google has always stayed years ahead of Bing
“Google has been sort of two or three years ahead of Bing, and has always been two or three years ahead of Bing, and Bing never had a hope of catching up.”
Cowan-Brown: Bing has nothing to lose against Google in search
“Microsoft and Bing had absolutely nothing to lose, but Google is on that back foot of we do have something to lose here”
Evans: Google's AI search will likely be better than Microsoft Bing's
“And one should presume that it's not that Google's version of doing this will be worse than Bing's, in fact it wouldn't surprise me if Google's version of doing this will be better”
Evans: ChatGPT's natural language presentation creates an illusion of certainty
“But if you go to ChatGPT and you ask it a question, it gives you one answer, and it presents that in natural language, which looks, which creates this impression of certainty. This is this phrase, it's a confident bullshitter. It might have exactly the same de…”
Evans: Generative AI viability depends on domain tolerance for 10% inaccuracy
“And the point is, there's sort of some domains where, like, a 90% accuracy produces a fantastic result, and you don't see or care the last 10%, and there's other domains where the last 10% is a problem.”
Evans: Single-text AI answers project certainty, unlike multi-link search results
“What the four images and the 10 links are doing is communicating uncertainty, whereas the one answer that's in text presents itself as the certainty far more than you get with than you get with the others, and”
Evans: AI utility relies on a 2x2 of error severity and verifiability
“And so this is why you have GitHub saying, well, we're going to call this Copilot, and like, it's writing code for you, but you're a coder, you can look at it and see. You can look at it and check. And so what are the domains where You can tell it's wrong. Ver…”
Evans: Asbestos lawyer queries are historically search's most valuable terms
“This has been the most valuable search term since forever because it's personal injury. So you've got it, you get a lawyer, you sue, the lawyer makes a lot, you make, you get a big payout, the lawyer, it's a lot. So this has been one of the most valuable searc…”
Evans: Radically new AI UX will reset default consumer search habits
“Presume that the chat GPT will be 10% worse than Google. I mean, maybe not, maybe it'll be 10% better, but the point is consumers will not care anymore. They will, the experience will be so different that you'll kind of reset people's presumptions and people's…”
Evans: Google pays Apple nearly $20B for default iPhone search
“Google pays, Apple pays, Apple probably close to twenty billion dollars to be the default search engine on the iPhone.”
Evans: Google's core search operating margin is around 60% excluding TAC
“If you exclude that, Google operating margins are core operating margins for the, like, the actual core search and advertising business are like 60%.”
Evans: The tech industry will aggressively push AI compute to local devices
“So there will be this massive push to kind of shift as much compute as you can down onto the device.”
Evans: Generative AI processing demands will reignite the PC hardware upgrade cycle
“Will I be able to tell the difference running Keynote and a web browser and email and Excel? No, I will not be able to tell. Would I be able to tell running Stable Dive Fusion? Oh, suddenly I get that image back in 10 seconds instead of 20 seconds. Oh no, that…”
Evans: AI search answers will drain web traffic from lifestyle publishers
“But if I'm going to say, you know, write me an itinerary for hotels, restaurants, and architecture in Barcelona, I'm going to get, that's going to produce less click-through, and so that's going to be less traffic for Condé Nast Traveler, it's going to be less…”