When OpenAI struck gold with ChatGPT 4 it was a breakthrough in the truest sense. That sudden jump in its ability to generate sensible language was completely unexpected and powerful enough to be instantly recognizable as a breakthrough.
That’s because big discoveries can often come out of left field. Viagra, for example, was supposed to be a blood pressure medicine before it turned out to be the “cure” for a different problem entirely.
With generative AI promised as the solution for so many different problems it’s easy to forget that it wasn’t invented to fulfill an actual need. Instead we simply opened up an single unexplored vein in the entire possibility of intelligence – artificial or otherwise. What we’ve discovered was never about making something that “thinks like a human” but an actual deep dive into the abilities and possibilities of intelligence itself.
On a purely business level that lack of a clear direction makes AI more difficult to exploit. Part of that’s because, as developers, we aren’t fully sure that our AI implementations will actually do what it is we want them to do. The other reason is that by the time we have fully implemented something as a usable technology, new discoveries may have either changed how we are implementing things or fully supplanted the need for our product at all. Especially when it comes to consumer-focused implementations.
Just as Einstein’s relativity theory utterly undid our assumptions about both the foundations and the limitations of the universe, AI is upending the fundamental assumptions we make about what intelligence is and by extension what it means to be intelligent.
It also means that while our personal models of reality may get us through the day, they are going to be relatively meaningless when it comes to creating things with these new technologies. The human-centric frameworks we rely on fail to encompass AI’s radically different manifestations of intelligence.
To some degree that’s already true. To really be able to use stable diffusion or midjourney for example it’s necessary to let go of traditional ideas of what and how images are created for one that is built on layers of perception, relationship, chaos, and noise. We must embrace new modes of understanding to harness AI creatively.
To experience these new processes of alternate intelligence for yourself can be both startling and humbling. But the fundamentals are woven into reality itself. Speeding up videos of vines and slime molds reveals that intelligence doesn’t need to be organized behind a consciousness to be effective. Nature finds a way.
Since the 80s there’s been an AI game called “Life”. It’s a simple evolutionary simulation that reveals that the laws of intelligence are as tightly wound into the bedrock of the universe as gravity and light. And it gets just as weird as quantum math if you dig down deep enough into its causes and effects.
Almost a century out from the first atomic bomb we had begun to believe that we had reached a point where the pace of technological discovery was not only stable but harnessed. As always, that complacency was a mistake.
I think it’s fair to say that the discoveries that are being made today will easily be as consequential as those that allowed us to split the atom. Once again the limiting rails of human belief, inspiration, and direction have been lifted. We already have AI giving us new hyper optimized strategies for everything from battlefield planning to playing a game of GO that are unlike anything that has been realized in the 4,000 years since people first started playing the game.
Fortunately for us the technologies made possible by AI go far beyond just strategic simulation. Our discoveries in physics have brought us technology from the atomic bomb to computers and flat screen TVs. Similarly, our newfound grasp of intelligence’s depths will surface possibilities and perils we can’t yet fathom.
But just as new discoveries are often unexpected, the distance between the latest breakthrough and the next may be farther than we imagine. We thought self-driving cars would be commonplace by now. But it turned out the timing of innovations remains unpredictable. When you’re this far off the map the only way to know for sure where the next breakthrough lies is to keep exploring in the direction that got you where you are today. AI’s journey has only just begun.