156 - Cautionary Tale

Try to name two AI-centric "Unicorns" today who had more than virtually or literally nothing in terms of demonstrated cutting-edge technology before receiving $30m+ investments.

OpenAI had nothing before their first $30m (a shady investment between literal roommates), and virtually nothing to justify the $1 billion investment from Elon and Microsoft.

Anthropic also had nothing prior to their first $124 million investment and promised something that is very literally impossible, which they of course didn't deliver. They still had nothing when they raised another $580m, $500m of which was money laundering from Sam Bankman-Fried, who went to prison later that same year. It remains a mystery why Anthropic's founders aren't his cellmates.

Mistral raised over $100m shortly after being founded when they had precisely nothing to show for themselves, and the founder's primary claim to fame was a paper that failed replication.

Some of the other names, like Google and Meta, come from companies with billions in revenue, so investment wasn't required for them. Though there are many more startups, most are only vaguely related to AI, companies offering some niche product that most people have never heard of.

Understanding this, a competent investor should ask:

• Compared to leading companies when they were pre-funding, how does this opportunity stack up?

• Compared to companies that failed post-funding, what can be learned and avoided?

Publications in prestigious journals, benchmarks, and former employment with major tech companies are among the worst possible forms of Substitution Bias that may be applied to vet opportunities, as they've already been systematically gamed by bad actors. In contrast, methods of comparison, as noted by Kahneman, are a valid way of reducing cognitive bias.

...investors should never catch themselves asking for:

• A pre-funding startup to beat post-funding startups with years of post-funding operation, on their terms, particularly when they don't directly compete.

• Nor should they require a level of technology to be demonstrated, which if demonstrated would immediately send the investment opportunity far outside of their price range.

I've personally seen investors make each of these critical mistakes. Yes, people are pitching trash "AI Startup" ideas like crazy, because year after year people invested heavily, and often primarily, in pure snake oil. So long as you continue investing in snake oil, the volume of trash pitches will continue to grow.