Pick your potential

Things look different now than they ever did before.

I’ve been thinking a lot about the trajectory of AI over the coming years.

Actually, it’s daunting enough thinking about it over the coming weeks and months.

A lot of people have a lot of different things to say about what’s possible, what kind of future we have to look forward to – or dread. And everybody’s got an excellent reason for it.

My thinking tends to be more basic, less tied into the constantly shifting landscape of features and functionality and funding, less a tune to regulations and jogging for political position, and more about the basic state of things.

At the risk of hopelessly oversimplifying everything, and in the spirit of calling something out that really affects all of us in ways that we’ve never been affected before…There’s this image I have of the reality versus the promises around AI. See above.

Oh look! let me pause for a minute, because I think I see a Northern Harrier flying over the marsh grasses in the distance. I’ll be back in a minute. Where are my binoculars?

…

Oh, I think I missed it. But I’m sure it will be back, because the marshes I can see from my front window are full of life, and harriers come to them pretty frequently.

OK, back to the AI landscape. The image above illustrates why I think that making any predictions at all about AI bubbles, AI features, AI whatever can’t possibly be accurate all across the board. When people make these predictions about “AI“ it seems like they’re making these blanket predictions about a unified, homogenized industry.

But that industry doesn’t actually exist in the way that we were thinking about it a couple of years ago. If you look at the divergence of all the different models that have come out, as well as all of the standalone or self hosted or private models that are out there, you get a very different view then what we had not so very long ago.

I’m not sure it’s even accurate to use the term “AI“ in the singular, because there are so many different features and functionality that are possible with it, there are so many different implementations, there are so many different applications and possibilities. And what we have now – what we have had emerging over the last few years – is an increasingly crowded field of options, contenders, possibilities that really have varying degrees of promise and reality rolled up into them. They all give us spikes of hope, and then come crashing down to reality at different intervals. They never failed to inspire, and they never failed to disappoint. And yet, we keep going.

So for anyone to make a monolithic prediction about what will happen with AI by 2027 or 2030, or beyond that… seems a really simplistic and more like Clickbait than serious thought.

Or maybe there’s something I’m completely missing and I need to be enlightened. It’s not the first time that’s happened, and it won’t be the last.

I guess my point is that if we’re going to talk about AI, we have to wrap our heads around exponentially increasing levels of complexity, variation, and innovation. And we also need to understand that every single option we have is a mix of empty promises and exuberant optimism and the harsh reality of what is possible now, compared to what we think is possible… As well as what may be possible in the future, where I really feel like anything could happen.