One Year of ChatGPT: These Six Tendencies Have Been Quietly Shaping the Future of AI
ChatGPT marked a singular point in the history of technology — and the world
The most important second-order effect of AI’s success and rapid advancement after OpenAI released ChatGPT one year ago is that it’s easier than ever to miss the forest for the trees, in a good way. We are focused on the countless releases and announcements but greater things are brewing in the cauldrons of history.
ChatGPT is turning one. As Sam Altman said yesterday, “What a year it’s been.” He has more reasons than anyone to say that, but it applies to AI enthusiasts like you and me. For an entire year, the AI community didn’t stop to catch its breath for a single day. Every week we thought it was the craziest only for the following week to exceed our expectations. A few days off enjoying a peaceful vacation meant losing track of dozens of news and events. But in trying to catch up with papers, products, tools, ideas, and concepts, we have been inadvertently spending all our energy and attention analyzing the parts without looking up to see the whole.
I bet you could easily enumerate twenty-five or fifty AI models that came out after ChatGPT. But what does it all mean? Where are we going? How do these changes contrast with what AI was before ChatGPT? How would they impact the future of the field? The pieces are useful in themselves but together, they have been shaping up a puzzle. Today I want to shed some light on that puzzle.
I won’t enumerate all the new open-source models or bootstrapped startups that have emerged during this short — at least it feels that way — period. Throughout the year I’ve used The Algorithmic Bridge to approach AI from a different perspective; while keeping up on the daily affairs, I have tried to climb the mountains of information to offer you a cleaner bird’s-eye view of the AI landscape and help us find the patterns, connect the dots, and discern the broader trends that are easy to overlook without the much-needed reflection that most of you don’t have time for.
That’s what I want to do with this article, in a condensed way. I should clarify that I have no secret insight or privileged information. I’m just an outsider, like (most of) you. What’s written below will become evident in retrospect (some of it will be evident right away). The only added value I provide is picturing how the forest has grown, in which directions, and how tall are the trees, instead of studying each of them individually.
Without further ado, here are the trends and tendencies that have accelerated (or emerged) since ChatGPT was released one year ago. Not every one of them is a direct consequence of ChatGPT but all are indirect effects of its existence. The short-term future of AI will be built on top of these paradigm shifts. I’ve divided them into two parts to make the article more readable and feel less burdensome. Will publish the first part today and the second tomorrow.
First part
The unification of AI: Previously scattered areas are converging around one goal.
The industrialization of AI: Production dominates research and development.
The centralization of AI: How a few big companies are reaping the profits.
Second part
The democratization of AI: Open-source initiatives are catching up.
The de-escalation of AI: Small and cheap models work just fine.
The recognition of AI: How ChatGPT got AI into the collective awareness.