Is Avocado… Toast?

What It Covers

Meta's Avocado AI model faces delays until at least May after failing to match rivals Google, OpenAI, and Anthropic in internal benchmarks. The episode also covers a landmark social media addiction trial nearing verdict, streaming apps converging on TikTok-style vertical video, and a counterintuitive analysis of how AI will actually displace workers.

Key Questions Answered

• •Meta AI Gap: Meta's Avocado model outperforms its predecessor Llama 4 and Gemini 2.5 from March, but falls short of Gemini 3.0 released in November. Meta is actively discussing temporarily licensing Google's Gemini to power its own products while Avocado catches up — a significant admission of competitive weakness from a company that invested $4.3 billion in Scale AI alone.
• •Social Media Liability Benchmark: The bellwether trial against Meta and YouTube in Los Angeles sets precedent for thousands of similar lawsuits. Jurors need only find negligence was a "substantial factor" in harm — not the sole cause — and just nine of twelve jurors must agree per count. TikTok and Snap already settled before trial, signaling financial exposure is real.
• •Streaming Convergence Strategy: Peacock and Disney Plus are both launching TikTok-style vertical video feeds. Peacock's Bravoverse feature generates over 600 billion possible clip combinations from 5,000 hours of footage. Disney Plus's Verz feed lets users swipe clips and jump directly into full shows. Netflix launched a similar vertical feed in 2024, confirming this is now an industry-wide retention tactic.
• •AI Paradigm vs. Task Automation: The critical distinction for understanding AI's economic impact is paradigm replacement versus task automation. ATMs automated teller tasks within existing branch banking; the iPhone eliminated the paradigm entirely, cutting US bank teller employment from 332,000 in 2010 to 164,000 by 2022. AI slotted into existing workflows will underdeliver — the displacement comes when entirely new organizational structures emerge.
• •Fully Automated Firm as the Real Threshold: Productivity gains and genuine labor displacement from AI will not come from drop-in AI workers filling human-shaped roles. The real inflection point arrives when AI enables wholly new organizational structures — what researchers call the fully automated firm — where workflows are rebuilt around AI capabilities rather than adapted from human ones. Current productivity data reflects the pre-paradigm-shift phase.