Bing’s Revenge + Google’s AI Faceplant

Episode Summary

Executive Summary: The episode covers Microsoft’s launch of an AI-enhanced Bing and Edge browser powered by OpenAI models, contrasting it with Google’s botched Bard reveal. The hosts argue AI search could reshape research, publishing, and everyday productivity, but warn about hallucinations, attribution gaps, safety risks, and a looming arms race to deploy powerful models responsibly.

Main Topics: Microsoft’s AI-powered Bing relaunch (Priority: 5/5): A deep dive into Bing’s new chat-based search and Edge browser features, including longer prompts, contextual follow-ups, cited answers, and drafting tools that turn search into an interactive assistant. Google’s Bard misstep and competitive pressure (Priority: 5/5): The hosts criticize Google’s rushed AI announcements, especially Bard’s incorrect demo response and the stock market reaction, framing it as evidence that Google is panicking under pressure from Microsoft and OpenAI. Impact on publishers and search economics (Priority: 4/5): The discussion explores how AI-generated answers may reduce clicks to blue links, threaten SEO-driven sites, and force a rethink of attribution, referral traffic, and monetization for publishers. Accuracy, citations, and trust (Priority: 5/5): The conversation emphasizes that Bing’s footnotes and source links are an improvement over ChatGPT, but that AI still makes mistakes and current citations are not always easy to verify or sufficiently prominent. Safety, moderation, and deployment pace (Priority: 5/5): The interview with Sam Altman and Kevin Scott focuses on responsible AI, the need for societal guardrails, moderation, and the risks of rushing powerful systems into products used by billions. AI-generated media and bias (Priority: 4/5): The hosts examine Nothing Forever, an AI-generated endless Seinfeld-style stream that was suspended after producing homophobic and transphobic content, using it to illustrate model choice, training data, and moderation failures. Disclosure and the future of AI-authored content (Priority: 4/5): The episode closes with a call for clearer disclosure when AI is used to generate emails, streams, or other content, arguing that users should know what model produced what they’re consuming.

Key Arguments: Bing’s new AI search is more useful than ChatGPT because it is integrated into search and browser workflows and can cite sources directly. Google’s Bard launch looked rushed and weak, especially because its only public demo contained a factual error. AI search may reduce raw traffic to publishers, but the traffic that remains could be higher-intent and more valuable. Current AI systems still hallucinate and should not be fully trusted for important decisions without human verification. The industry needs broad social and regulatory guardrails, not company-only rules, for powerful models. OpenAI and Microsoft argue that releasing systems to the public is necessary to learn, improve safety, and understand misuse in real-world conditions. AI-generated content will increasingly blur the line between human- and machine-authored media, making disclosure important for trust.

Data Points: Characters allowed in Bing search box: up to 1,000 characters - The new Bing search interface accepts longer prompts than standard search queries. Characters allowed in Bing chat tab: up to 2,000 characters - Inside Microsoft Edge, the Bing chat feature supports longer conversational prompts. Google Bard launch timing: Monday announcement / Wednesday event - Google announced Bard shortly before Microsoft’s event and followed up with a Paris presentation. Alphabet stock move after Google event: about 7% fall - Shares dropped after Google’s AI presentation underwhelmed investors. ChatGPT mainstream attention: a little over two months - Sam Altman notes the technology has only recently entered mainstream awareness. GPT-3 launch reference: 3 years ago - Altman contrasts the current pace with the state of AI progress three years earlier. Nothing Forever runtime before suspension: a couple of months - The AI-generated Seinfeld-like stream ran for months before producing offensive material and being pulled. Family subscription extras: room for two others - An ad read mentions New York Times Family subscription includes multiple logins.

Pivotal Quotes: "“the new Bing gives you an answer to your question, it adds a footnote.”" — Casey Newton: Explaining why Bing’s AI search feels more trustworthy than ChatGPT because it cites sources. "“Google still has barely let anyone actually try this thing.”" — Kevin Roose: Critiquing Bard’s launch as a blog-post-driven reveal rather than a real product users can test. "“The way this should work is that there are extremely wide bounds of what these systems can do that are decided by not Microsoft or OpenAI, but society, governments, something like that.”" — Kevin Scott: Discussing the need for external guardrails and shared societal control over AI behavior.

Implications: AI search is moving from novelty to infrastructure, reshaping how people research, write, and consume content. Expect stronger demand for citations, more pressure on publishers, and escalating scrutiny over safety, bias, and disclosure.