Predictions 2023: How’d I Do?
Well that was one hell of a year.
As I do each December, it’s time to grade my own homework. And the past twelve months certainly started out well. But unless a certain fascistic presidential candidate has a change of heart in the next few days (he won’t), I’m afraid I didn’t break .500 this year (last year I was smokin’ hot, I must say).
In any case, let’s get to it. Here’s a review of what I predicted, and my take on how it turned out.
- ChatGPT finds a business model. Remember, this prediction came just six weeks post-launch, and given the company’s servers were melting from demand, there was a lot of speculation around whether OpenAI would or even could stand up reliable revenue streams. The company was being compared to early versions of Netscape, Google, and Facebook — and none of those breakouts had real business models in their first year. But OpenAI did in fact find its business model — selling its services to Microsoft, naturally, but also to millions of individual developers and consumers. While profitability is still over the horizon, OpenAI did become — as far as I can tell — the fastest business to a billion-dollar run rate in the history of the Internet. Not bad for year one, even with the goat rodeo at the end. I’d grade myself as getting this one right.
- Google launches a ChatGPT-inspired search interface. In my post covering this prediction, I wrote that “Google will build a novel conversational interface to its flagship Google search application.” Months later, Bard launched as a sandboxed “experiment” (something I also predicted). This pasts Fall, Google put Bard, now powered by its more advanced Gemini model, into wide release. Many, many questions remain about how Google will manage its customer base in relation to Bard and its core search business, but again, I’d say I got this one right.
- Microsoft launches “Enterprise Explorer.” Here’s what I meant by that: “Built, again, from a mashup of OpenAI technology and Microsoft’s Azure compute platform, E2 would address some of ChatGPT’s most annoying problems — its indifference to truth, for example, or the biases inherent to its Web-scale training corpus. The idea would be this: Train a…