Why I bet Inside Out 2 will reach $85 million domestic opening weekend, as predicted by AI

🌈 Abstract

The article discusses how Large Language Models (LLMs) can be used to predict box office success and other future events by tapping into the collective opinions and sentiment of the public. It presents research from Baylor University showing that LLMs can make surprisingly accurate forecasts by generating "future narratives" that bypass their usual restrictions. The author then applies this technique to predict the opening weekend box office performance of the upcoming film "Inside Out 2", and is willing to bet the earnings from this article on the accuracy of the AI's prediction.

🙋 Q&A

[01] Large Language Models and Predictive Analytics

1. What are the key findings from the Baylor University research on using LLMs for predictive analytics?

• The researchers found that prompting LLMs like ChatGPT to generate "future narratives" - stories set in the future with hypothetical events - can yield surprisingly accurate forecasts, such as correctly predicting the 2022 Oscar winners.
• This method allows the LLM to bypass some of its usual restrictions and speculate on future events, while still leveraging its ability to generate text procedurally based on patterns in its training data.

2. How does the author replicate and validate the Baylor University findings?

• The author replicates the experiment with the Trump hush-money verdict, which was an unfolding event that both ChatGPT and the experimenter would be blind to. The author finds that ChatGPT accurately predicted Trump would be found guilty on all counts.
• This helps rule out the possibility that the LLM was accessing new information outside its training data via the Bing browser, without the knowledge of the original researchers.

3. What are the potential advantages of using LLMs for predictive analytics compared to traditional methods?

• LLMs may be better suited for predictive analytics where language and sentiment are the key factors, as they can tap into the collective cultural consciousness and discern the consensus of public opinion.
• The author suggests LLMs can act as a "conduit for collective intelligence", reading the collective sentiment from their extensive training data and anticipating where something might fit in the wider narrative of cultural and social discourse.

[02] Predicting Box Office Success with LLMs

1. What is the author's specific prediction regarding the box office performance of "Inside Out 2"?

• The author predicts that "Inside Out 2" will be the first film of 2024 to reach $85 million on its opening weekend, outperforming other major 2024 releases like "Dune Part Two" and "Godzilla X Kong: The New Empire".

2. Why does the author feel confident in this prediction?

• The author believes that the inherent "boringness" and predictability of AI-generated text is actually an asset when it comes to forecasting events shaped by collective opinions, emotions, and attitudes.
• The author argues that LLMs can tap into the "zeitgeist" and discern which stories are likely to resonate with audiences, making them well-suited for predicting box office success.

3. What is the author's wager and proposed charity donation related to the "Inside Out 2" prediction?

• The author is willing to bet all the first week earnings of this article (up to $850) and donate it to a charity for children's mental wellbeing if the AI's prediction is wrong.
• The author also commits to donating any amount the article earns above $300 before the film's release, regardless of the outcome, to further support the chosen charity.