marty chavez

The connectionist/neural network approach to AI, once dismissed as 'bonkers,' proved massively scalable after 30 years of Moore's Law doubling compute every 18 months, reaching the second half of the chessboard where scaling laws make large language models fruitful — a realization that only became obvious planet-wide after ChatGPT in late 2022.

Building digital twins of trading businesses (SecDB at Goldman, now at Sixth Street) enables risk simulation in software before reality, turning risk management from a reactive spreadsheet exercise into a proactive computational advantage — a pattern repeating in private credit as the 'last frontier' for digitization.

Alternative assets/private credit remain modeled in disconnected spreadsheets with complex waterfall structures, creating a massive opportunity for digital platforms to replace Excel as the state-of-the-art — Sixth Street is building this infrastructure as AUM scales past $100B.

Algorithmic trading systems ('algos') were early AI agents autonomously placing orders; Chavez's three-strategy framework — tell computers what to do, collaborate with computer-users, or resist and lose — shows that human-AI collaboration (strategy 2) is the only viable path, as resistance accelerates career obsolescence.

A plausible future scenario has AI driving productivity growth to 20% (vs 3% today), meeting material needs offscreen while humans focus on uniquely human activities — analogous to Bridgerton's offscreen income — making this a structural tailwind for AI investment if the probability is non-zero.