Technologies like large language models (LLMs) and other foundation models present new opportunities for forecasting. Combined with a ‘Superforecasting’ approach which purports that non-experts can yield stronger forecasts than experts by using regularly updated, structured analyses, we believe that LLMs can fundamentally change the way we forecast the future.
Imagine the possibilities of an AI-powered "analyst" with all the world's information at its 'finger tips’. LLMs are capable of modeling an abundance of timely, structured and unstructured data at scale. They can also account for highly complex interdependencies. For instance, the price of oil is dependent on countless variables — the value of the US dollar, OPEC price commitments, demand, reserves, and more. Chaining forecasts at scale becomes possible with LLMs.
LLMs amplify human forecasting capabilities. Many forecasts are qualitative in nature, including those that are predicting a numerical variable. Even numerical forecasts are likely dependent on numerous qualitative or unstructured inputs such as disruptive policy decisions, elections, military activity, and more. This makes it nearly impossible to have accurate forecasts without any sort of qualitative input.
We launched this forecasting community to encourage others to engage in LLM powered forecasting. We are excited to be on this journey together!
Fill out this form or reach out at hello --at-- phaseai --dot-- com