Hard Fork
Hard Fork

Breaking Bard + Who Owns Your Face? + Gamer News!

Hey Bard, can you take a look at my Gmail and psychoanalyze me? Turns out, that’s not a question Google’s chatbot, Bard, can answer with any veracity despite new features that allow it to plug into your Gmail, Google Drive and more . Kevin and Casey on why Bard isn’t answering all their questions co

Featured Speakers

The New York Times HostKashmir Hill Guest

Topics Discussed

Episode Summary

Executive Summary: The episode opens with a comic rat-trap anecdote before focusing on Google Bard’s new integrations with Gmail, Drive, and travel services, and the chatbot’s still-flawed ability to search, summarize, and “double-check” personal data. The second major segment features Kashmir Hill discussing her book on Clearview AI, facial recognition, privacy harms, legal gaps, and the risks of making identity-search tools broadly available. A final “Gamer News” segment covers Unity’s controversial pricing changes and Microsoft gaming leaks.

Main Topics: Bard Extensions and AI in personal data (Priority: 5/5): Kevin and Casey test Google Bard’s new ability to connect to Gmail, Docs, Drive, Maps, Flights, Hotels, and YouTube, finding it useful for some narrow tasks but unreliable when asked to analyze or summarize personal emails or complex travel plans. AI hallucinations and the double-check feature (Priority: 5/5): The hosts discuss Bard’s new citation/highlighting tool meant to help users verify answers, arguing that it improves confidence in some cases but still forces users into time-consuming fact-checking workflows. Clearview AI and facial recognition as privacy threat (Priority: 5/5): Kashmir Hill explains how Clearview scraped billions of images, built a face-search app for law enforcement, and helped normalize a technology that can identify strangers from a single photo, with troubling uses in policing, retail, protests, and dating. Law, regulation, and ethical boundaries for facial recognition (Priority: 4/5): The conversation covers Illinois’ biometric privacy law, the absence of strong federal protections, and the need for guardrails before facial recognition becomes more deeply embedded in everyday life. Open-source vs closed control over dangerous AI (Priority: 4/5): The discussion connects Clearview’s rise to broader debates about whether big tech companies should keep some tools closed and restrained, or whether open release accelerates innovation at the cost of safety. Gaming industry shakeups: Unity and Microsoft (Priority: 3/5): In Gamer News, Unity’s new runtime fees provoke backlash from developers, while leaked Microsoft documents reveal future Xbox plans and a possible discussion of buying Nintendo, highlighting consolidation and platform power in gaming.

Key Arguments: Bard’s new integrations are a milestone because they bring generative AI into the places users already spend time, especially Gmail and YouTube. The double-check feature is helpful, but the need to verify citations still makes Bard feel like a research project rather than a true assistant. Bard is currently better at retrieving and summarizing obvious data than analyzing large personal email archives or performing complex tasks like psychoanalyzing a user. Clearview AI demonstrates that facial recognition can cross a major ethical line by enabling identification of strangers without consent. Facial recognition has legitimate uses in some law-enforcement contexts, but it should never be treated as sufficient evidence on its own. There is a dangerous mismatch between the speed of deployment and the weakness of U.S. legal protections around biometric data. Open release and startup competition can push technologies across ethical boundaries that incumbent big tech companies may avoid. Unity’s pricing change is risky because it breaks the expectations of developers who built games under a royalty-free model. Microsoft’s leaks underscore how much major tech companies want a larger stake in gaming and how consolidation raises competitive concerns.

Data Points: Price of rat bucket kit: $25 - Kevin jokes that he bought a rat bucket kit from Amazon to deal with backyard rats. Bard Extensions services: Gmail, Drive, Docs, Maps, YouTube, Flights, Hotels - Google’s new Bard integrations can access personal Google data and travel information. Clearview database size: 30 billion faces - Kashmir Hill says Clearview claims to have built a database of roughly this size from scraped online images. Clearview claimed accuracy: 99% - Hill describes Clearview’s claim that its face-search app works with about 99% accuracy. Illinois biometric penalty: Up to $5,000 - Illinois’ Biometric Information Privacy Act can impose this penalty for unauthorized biometric use. Year Illinois law passed: 2008 - The strongest U.S. state-level facial recognition law discussed was passed in Illinois in 2008. Human feedback scale: Millions more people - Google’s reason for releasing imperfect Bard extensions is to get feedback from far more users. Unity fee timing: January - Unity’s new runtime fee was scheduled to take effect in January. Unity pricing unit: A couple cents per install - The runtime fee would charge developers based on installs rather than upfront licensing. Book title: Your Face Belongs to Us - Kashmir Hill’s book about Clearview AI and facial recognition.

Pivotal Quotes: "I think in 50 years, there's only going to be two things left on Earth. It's going to be chat GPT and rats, and we'll let them duke it out for the future of the planet." — Kevin Roose: Opening banter about the host’s new rat trap and broader fears about AI and animal intelligence. "It works incredibly well and he had written this memo for the company to explain to police why using Clearview AI was not illegal and that they wouldn't break any state or federal laws." — Kashmir Hill: Hill recounts the legal memo that first alerted her to Clearview AI’s capabilities and law-enforcement ambitions. "I truly cannot imagine a more dystopian outcome for the path that we are on." — Kashmir Hill: Hill argues that ubiquitous face-search technology would severely erode privacy and freedom in public life.

Implications: The episode suggests AI is moving from abstract hype into daily infrastructure, but reliability, privacy, and abuse remain unresolved. Expect more pressure for regulation, better verification tools, and pushback against technologies that de-anonymize people or surprise developers.

🔓 Sign Up for Unlimited Episode Search

About Hard Fork

“Hard Fork” is a show about the future that’s already here. Each week, journalists Kevin Roose and Casey Newton explore and make sense of the latest in the rapidly changing world of tech. Unlock full access to New York Times podcasts and explore everything from politics to pop culture. Subscribe today at nytimes.com/podcasts or on Apple Podcasts and Spotify. Also, for more podcasts and narrated articles, download The New York Times app at nytimes.com/app.

View all episodes from Hard Fork