FutureGPT is an experimental online service that shows how large language models (LLMs) can be used to generate forecasts about future events. It’s positioned as a proof of concept: the goal isn’t to provide flawless predictions, but to illustrate both the potential and the limits of AI in “predicting” outcomes.
How it works
You enter a question about the future or a hypothetical scenario, and FutureGPT generates a prediction using a language model. For higher-quality answers, you can enable GPT-4; in that mode, the service uses paid predictions. The interface also lets you request new forecasts and adjust model settings.
What it’s for
FutureGPT is best suited for research and learning, including:
Testing hypotheses and “what-if” scenarios
Demonstrating how LLMs produce probabilistic reasoning
Discussing uncertainty, assumptions, and model limitations
Important limitations
The creators emphasize that FutureGPT is not financial, legal, or professional advice. Outputs are experimental and shouldn’t be the sole basis for important decisions.
