Last week, at a farewell party for a data scientist friend (who is about to ship out from Seattle to Palo Alto to work for a certain social media network based there), I had an interesting exchange with another friend who runs a self funded AI-based startup. Our conversation turned to wondering about whether we’re in the middle of an AI bubble (remember the dotcom bubble?). He asked whether I thought there would be any winners if the AI bubble bursts, and my answer was as follows.
Let’s set a floor on defining “winners” by looking at the table stakes for innovation in general. Venture capitalists (who ought to know these things) already expect maybe 8 out of 10 of the startups in their portfolios will fail, no matter how heavily vetted they are, whether those startups are in AI or some other flavor of “new”. So even if AI isn’t in a bubble that’s about to burst, we were already expecting to see a lot of good AI ideas die, including ideas originated in startups as well as R&D by established players. Maybe those ideas are just ahead of their time; maybe their execution will be flawed; maybe they’ll just have bad luck, and run out of time. A bubble collapse would just accelerate the day of reckoning. We’ll still be left with a number of strong new solutions.
As for other lasting benefits of the current AI hype cycle, even if an AI bubble bursts there are at least three major legacies that would be bequeathed to us.
First, infrastructure. The current massive investment in AI has been equipping us with a permanently enlarged capacity for handling data and AI more efficiently (just as the dot com bubble left us with massive amounts of dark fiber that later proved valuable, among other gifts). Examples of these efficiencies include (but are not limited to) recent giant steps in chip technology related to data and AI (like the continuing evolution of FGPUs) and a massive increase in the number of people trained to work productively with data and AI. We’ve also achieved huge leaps in the way that AI models are deployed and managed. By way of example for this I’ll give a shout-out to Algorithmia, which has evolved from a B2B marketplace for connecting creators and users of AI models into an Enterprise platform for encapsulating AI models, giving users the ability to simply call AWS Lambda functions with a few lines of code to spin up and utilize, at scale, any model from an approved library of models.
Second, incremental gains for existing solutions. The big software companies who are selling broad product suites aimed at both businesses and consumers have been slowly and steadily incorporating AI elements into existing feature sets and adding new features based on AI. Here I call out Salesforce as an example, with their “Einstein” brand (that they apply to AI features or components across their many offerings). For instance, Salesforce’s call center functionality now has AI capabilities for things like recommending offers for a particular customer based on a comparison of that customer’s history to the history of similar customers. When I spoke with a Salesforce product manager at one of their roadshows earlier this year I learned that while they aren’t leading the pack on AI innovation they are steadily developing and rolling out a range of cutting edge AI offerings (like chatbots). In the long term I assume such AI features being developed by Salesforce, Microsoft, and other big players will have staying power, AI bubble or no, because the big players have both the resources to develop viable niche AI solutions and the customer base to promote and perfect these solutions at scale over the long haul.
Third, and finally, a new normal. There are incredibly successful AI solutions that we’re already so married to that we don’t even think of them as AI innovations anymore (“the AI effect”). Among these are Google and Microsoft translation products, now capable of mind-blowing feats like mobile real-time translation via a smart phone. We’re not exactly going to lose these capabilities if (when?) an AI bubble bursts.
So AI has already transformed our lives—even if self-driving cars take a little longer to arrive than most currently expect.