AI Pioneer Who Built First Hedge Fund Won't Trust ChatGPT With His Money
Vasant Dhar helped bring AI to Wall Street in 1994. Decades later, he has pointed warnings about trusting modern AI with investment decisions.
Long before generative AI became a boardroom buzzword, Vasant Dhar was quietly wiring machine intelligence into the plumbing of financial markets. In 1994, he built one of the earliest AI-driven hedge funds, a milestone that places him in a rare category of technologists who have watched the full arc of artificial intelligence's evolution on Wall Street — from narrow, rules-based systems to the sprawling large language models that now command public fascination.
That historical vantage point is precisely what makes Dhar's current skepticism worth taking seriously. Despite his pioneering role, he has made clear that he does not trust tools like ChatGPT to manage his personal finances. This is not the reflexive caution of someone unfamiliar with the technology; it is the considered judgment of a practitioner who understands both what AI can optimize and where its confidence becomes dangerous.
Read more Apple Stock Nears All-Time High as Options Signal Unusual Activity →
The distinction Dhar implicitly draws is one that sophisticated investors would do well to internalize. Early quantitative AI systems were designed to find statistical patterns in structured data — price histories, volume signals, earnings trends. They operated within tightly defined boundaries and were evaluated relentlessly against hard performance benchmarks. Generative AI, by contrast, is trained to produce plausible-sounding language, a capability that can create the illusion of financial insight without the underlying rigor that markets ultimately demand.
For retail investors newly tempted to outsource portfolio decisions to a chatbot, Dhar's position serves as a calibrating signal. The fact that someone who literally built the technology three decades ago maintains a clear boundary between AI as an analytical aid and AI as a financial decision-maker suggests those boundaries are not arbitrary — they reflect something real about how these systems fail under pressure, in novel conditions, or when the cost of a confident wrong answer is measured in dollars rather than prose.
The broader lesson may be about epistemic humility in a moment of enormous AI hype. Wall Street's most rigorous quantitative shops have spent years learning what machine learning can and cannot do with money. That institutional knowledge is not yet embedded in general-purpose AI assistants, and Dhar's decades of experience underscore why that gap matters. Continue reading at MarketWatch.com
