I think the AI bubble is almost certain (~95% likely) to pop within the next ~7 years.^[1] I'm looking for anyone in the effective altruism community or AI safety community who is willing to bet against me. Consider betting against me if you believe with high confidence there isn't an AI bubble and that there won't be any AI bubble pop within the next ~7 years.

There is obviously a close connection between the probability of an AI bubble and the trajectory of general capabilities improvement in AI. I have argued that the EA 2.0 community and the AI safety community are vastly overestimating how much general AI capabilities will improve within the next ~5-10 years. A bet on whether there is an AI bubble that will pop sometime within the next ~7 years is an indirect bet on how close we are to artificial general intelligence, transformative AI, existentially dangerous AI, and so on.

I've played around with formulating this bet a few different ways. Here is a formulation that marries elegant simplicity with absolutely unambiguous resolution criteria. None of this is set in stone whatsoever. If you have another idea about how you would like to formulate an AI bubble bet, please make me a counteroffer.

Stakes

$20 goes to the charity of the winner’s choice. 

Resolution criteria

By January 1, 2034, all of the following news outlets will report — as fact, not opinion or editorial — that the AI bubble has popped:

• The New York Times
• The Washington Post
• The Wall Street Journal
• Bloomberg
• The Financial Times
• The Economist
• The BBC
• The CBC
• Le Monde

To repeat, all of these news outlets have to report it. And they have to report it as news. Opinion columns and editorials don’t count.

Explanation

Here's how I landed on the timeframe, stakes, and resolution criteria:

• The timeframe is ~7 years because I read that many AI investments are amortized over 5 to 7 years.
• The stakes are low because this bet is mainly just for fun and also to make a point. If you want to bet big money, use the financial markets. Bets are not legally binding, but securities transactions are!
• The bet defers to news outlets as “judges” because operationalizing the exact conditions under which an AI bubble has or hasn't popped is too complex and too easy to get wrong.^[2] Deferring the call to nine highly reputable news outlets in four different countries (the U.S., the UK, Canada, and France) is a good idea because the chance that all the news outlets will get it wrong is minimal.

Again, none of this is set in stone by any means, so please counteroffer me an alternative bet if you've got something better. 

If you'd like to accept this bet or propose something a bit different, feel free to message me here on the EA Forum or on Substack. You can also try my public email address if you prefer, but your email might get stuck in the spam filter. 

1. ^{^}

   Wikipedia definition of a bubble:

   An economic bubble (also called a speculative bubble, asset bubble, or simply financial bubble) is a period when current asset prices greatly exceed their intrinsic valuation, being the valuation that the underlying long-term fundamentals justify.

2. ^{^}

   For instance, what if the valuations of OpenAI and Anthropic get decimated but AI datacentre investment continues to grow because open source LLMs take off? 

   Or what if that happens and AI datacentre investment also plummets because open source LLMs become way more efficient but businesses become much more productive and profitable by using them? 

   What if you choose to bet on Nvidia's market cap, using GPU sales a proxy for AI, and then there's a huge resurgence of cryptocurrency mining?