Large language models like ChatGPT may not be great at picking stocks, but that doesn’t mean they can’t be useful in the world of investing. Greg Jensen, the co-CIO of Bridgewater, one of the world’s largest hedge funds, recently shared his perspective on the role of AI in investment decisions.
Jensen acknowledges that these intelligent language tools are awful at reading the market and making accurate predictions. However, Bridgewater is taking a different approach by using these models to generate hypotheses that can be tested through rigorous statistical analysis.
Using large language models to directly pick stocks is seen as a hopeless endeavor by Jensen. Instead, Bridgewater combines these models, which lack precision, with statistical models that excel at analyzing past data but struggle to predict the future. By combining the strengths of both approaches, the hedge fund aims to build an ecosystem that can achieve impressive results at a larger scale.
Jensen’s perspective aligns with the general sentiment surrounding large language models like ChatGPT. While these AI-powered tools may occasionally make small mistakes or even come up with completely new answers (referred to as hallucinations), they have the advantage of producing vast amounts of text rapidly.
Bridgewater sees this ability to rapidly generate text as valuable. Jensen describes large language models as already possessing similar capabilities to a top 20% human analyst, but at a much faster pace. Leveraging these models can provide Bridgewater with millions of investment associates that can be controlled and guided through rigorous statistical analysis, allowing them to accomplish a significant amount of work in a short time.
It’s important to note that large language models are not flawless; they have limitations. Jensen acknowledges their flaws but believes that, with the right statistical backdrop, they can be immensely powerful in generating investment strategies. While these models may not replace human analysts entirely, they can enhance and accelerate the decision-making process.
In conclusion, Bridgewater is employing AI language models like ChatGPT to complement their investment strategies. While these models may not be perfect at picking stocks, their ability to generate massive amounts of text quickly presents valuable opportunities for analysis. By combining these models with statistical analysis, Bridgewater aims to leverage the strengths of each approach and achieve impressive results at a larger scale.
