← Home
This Week in Startups · July 31, 2026 · 01:15:09

Why AI has no taste and how to fix it (w/ Thais Castello Branco) | E2319

Thais Castello Branco, founder of Taste Labs, joins Jason and Lon to discuss why even the most advanced AI models produce bland, homogeneous output on subjective tasks—a phenomenon she calls ‘slop.’ She explains how her company raised $18.5M in seed funding to teach frontier models about aesthetics by using paid ‘TasteMakers’ as human feedback curators. The conversation also covers whether AI accelerates cultural trend cycles and erodes the half-life of ‘what’s cool.’ The show opens with news about Leopold Aschenbrenner’s AI hedge fund liquidating its public stock portfolio, Google Earth becoming an AI image misinformation factory, and LinkedIn and Substack joining the crackdown on AI-generated slop.

This summary was generated from show notes and public descriptions, not from a full transcript review. Details may contain inaccuracies.

Novel

TasteMakers: Curated Human Feedback to Improve AI Output Quality
Paying human taste experts to fine-tune models can break the cycle of bland AI output, according to the founder of Taste Labs.

Highlights

AI Models Regress to the Mean on Subjective Tasks
Even the most advanced AI produces homogeneous, uninspired content on subjective tasks because it's trained for average consensus, not aesthetic excellence.
AI Accelerates the Half-Life of Trends and ‘Cool’
Thais warned that AI’s ability to generate endless variations on cultural artifacts may compress trend cycles, making styles appear and disappear faster than ever.

Editorial

Leopold Aschenbrenner’s ‘Situational Awareness’ Hedge Fund Liquidated
The AI-themed hedge fund run by 25-year-old Leopold Aschenbrenner sold off its public stock portfolio amid losses, but the hosts cautioned against counting him out entirely.
Google Earth’s AI Image Generation Creates a Misinformation Factory
A new integration of AI imagery with Google Earth allows users to generate photorealistic but fake scenes over real locations, raising concerns about misinformation.

Misc

Leopold Aschenbrenner’s Situational Awareness AI hedge fund liquidated its public stock portfolio, prompting Jason and Lon to debate whether it’s a sign of an AI bubble popping or just an early stumble.
Jason and Lon wondered aloud why so many venture capitalists seem to have an opinion on the Ceuta border crisis.
Was this useful?