The endgame: exploring the future of AI

There's no earthly way of knowing

Which direction we are going

There's no knowing where we're rowing

Or which way the river's flowing

Is it raining, is it snowing?

Is a hurricane a-blowing?

- Roald Dahl, Charlie and the Chocolate Factory

Nothing in the future is really set. To all but a very select few, AI at its root level is a complete mystery. Even to those few in the drivers’ seats, there are a huge range of environmental factors that could set the overall direction of AI in many different ways. Today, let’s explore some of them.

Oh, and these 🎵 are links to songs I’ve paired.

Stagnation of AI

Technological stagnation: bleak but not terrible (🎵)

Over a year from the release of GPT-4, many new features have been announced, but there’s slowdown in the development of the intelligence part of Artificial Intelligence. The closing lead of OpenAI seems to indicate a ceiling. Based on Open AI’s own model evaluations, the differences between Chat GPT, Gemini and Claude Opus are maybe not neck-and-neck, but they are in the same ballpark.

The rockstars of the industry, leading researchers such as Andrej Karpathy and the legendary Ilya Sutskever, recently left OpenAI. That might be just me looking for a human narrative in a much broader game, but it feels like it matters.

It’s not so bad though. There are massive benefits even if progress stopped today. Businesses have only started to apply AI in any meaningful way, and just with current compute, the applications are vast, with slower progress smoothing out change and adoption.

Regression: a reversal of fortune (🎵)

We may just be in a kind of wild west phase of AI, where it is fairly unregulated, and nobody really knows what they’re doing with it. It’s possible that a culmination of factors starts to drag it back.

AI has trawled most of the internet by this point, and so the idea of it being trained on even more data rings a little like those sub-prime mortgages layered in before the GFC. Perhaps nobody really figures out synthetic data in a meaningful way (data for AI to train on created by AI). Perhaps the internet becomes filled with images and text to jailbreak AI, or just corrupt it.

Beyond this, regulations could, rightly or not, hamper further progress in AI, or even revert it. AI companies could be required to remove specific data sets, weakening their models.

In any case, let’s not take for granted what we have now.

Bubble: like stagnation, but worse (🎵)

Markets are chaotic and irrational. Right now, big tech is betting hard on AI, such as Microsoft, with their planned $100 billion USD data center. But will it pay off?

Gartner indicates that Gen AI is at the peak of inflated expectations before the trough of disillusionment. Maybe just due to its inherit nature it won’t integrate into businesses to the extent the market is betting it will.

Sam Altman (Open AI CEO) recently said “Does the global economy feel any different to you now or materially different to you now than it did before we launched GPT-4? I think you would say no.”

Maybe that will remain the case, and investors will chase the next trend, ridding AI of the massive, sustained costs it needs to run. I just don’t think this is how it will end, though.

Rapid ascension: that’s not an endgame, this is an endgame (🎵)

(or “we were so preoccupied with whether we could, we never stopped to think about whether we should”)

Not sure if this is the best end, but you are kidding yourself if you don’t think there’s some part of you that wants to see it.

Amidst all of the possible futures, we can see this one thread where AI goes beyond everything, ridiculously fast.

This is the future where AI drives scientific progress, medical miracles and art that speaks to us on a level we didn’t even know we had.

It’s a future where we are challenged by a rapidly outdated economic and social structure, and by questions of purpose in a world where we are fundamentally obsolete.

But it’s a different world. A new world full of possibility.

This is all blind speculation; I have no articles to link here. It’s just phenomenal that the advent of AI has enabled us to take a peek through this doorway, if only just a little.

Previous
Previous

Exploring the business applications of multi-modality

Next
Next

Designing future business processes: shifting to an AI-driven approach