What if AI is like aioli?
Aioli is everywhere but there are no aioli stores. Also: AI & science
The aioli metaphor
I participate in a weekly Zoom call with an interesting and eclectic group of investors and entrepreneurs. This past week, I presented some information about the state of AI, and one of the participants said something along the lines of: “AI is like aioli. You can get aioli in any sandwich shop in America, but you never see an aioli store.”1 And—this is a really good metaphor: you can find AI in every company, but you never see a pure AI company.
Now, it’s obviously not a perfect metaphor, because there are pure play AI companies like OpenAI, Midjourney, etc. But who is really making money from pure AI?2 Just as no one makes money solely by selling aioli, it remains an open question whether anyone will make money selling AI.
The more I think about AI, and the more I use AI, the more I find that AI functions as a kind of substrate upon which everything else will be built. Coca-Cola won’t start selling AI in a bottle, but all of the soda it sells3 will be made more efficiently by applying AI tools to its operations. All of the drugs that Pfizer designs will be discovered more efficiently with AI tooling, but Pfizer won’t sell you an AI pill. All of the stuff that Walmart sells you won’t be AI stuff, but AI will allow Walmart to make its operations even more efficient than they are today.
And—we see this everywhere. I think that it will be very hard to make money investing in AI, but it will be trivially easy to see how our lives are improved by AI. AI will suffuse everything, and make everything better. It will be, in other words, like aioli.
AI & science
One obvious example of AI serving as a substrate is its application to science. Here are three relevant links surfaced by Twitter this morning:
AI Detects Cancers and Immunotherapy Biomarker. “Researchers at the Perelman School of Medicine at the University of Pennsylvania have developed an AI tool called iStar that can automatically spot tumors and various types of cancer that are difficult for clinicians to see or identify, as well as predict candidates for immunotherapy.”
Towards Conversational Diagnostic AI. This is a paper from Google Research and Google DeepMind. “At the heart of medicine lies the physician-patient dialogue, where skillful history-taking paves the way for accurate diagnosis, effective management, and enduring trust. Artificial Intelligence (AI) systems capable of diagnostic expertise is an outstanding grand challenge.”
Open Catalyst Project. This is a research project which aims to use AI to “discover new catalysts to address the energy challenges posed by climate change.”
This is my paraphrase, and I believe that the participant heard this observation from someone else. In any event, it’s not something I came up, though the wording in my post is my attempt to express the gist of the claim.
OpenAI claims to be doing $1 billion in revenue, but it’s not clear how much of this is from actual paying customers, and how much of it is from Microsoft’s licensing. The reason that Microsoft’s cash payments to OpenAI aren’t a great indication of OpenAI’s revenue-generating abilities is that Microsoft has also invested, in the form of both cash and data center compute credits, in OpenAI. Customers excluding Microsoft is a much better indication of OpenAI’s long-term business prospects, and on that score, publicly available information is hard to come by.
Yes, technically, its independent bottlers manufacture the coke that Coca-Cola sells, but this point is beyond the scope of what I’m arguing in this post.