AI: the most over-hyped tech in history?

The relentless onslaught of AI hype has started to stumble in recent weeks. Sam Altman (OpenAI’s CEO) has taken back his predictions that the technology will replace all knowledge work, while Dario Amodei (at Anthropic) keeps amending his own timeline and the percentage of roles he thinks will be affected. As the industry’s leaders climb down from their soapboxes, it forces the question: how much of what we’ve been told about AI is real, and how much is the biggest hype cycle in history?

To be clear, there’s no doubt in my mind that when we talk about AI we’re discussing a transformational technology. Since the public launch of ChatGPT in November 2022, the progress has been remarkable. Over the space of just three and a half years trillions of dollars have been raised, hundreds of millions of people have become users and thousands of companies have been launched. AI has already impacted healthcare, how people buy things, access to information, education and entrepreneurship, and has levelled the playing field for those in less developed economies.

However, at the same time, what at first seemed like magic is starting to show cracks and gaps in capability as our own level of competence as users increases. We’re also beginning to understand the real constraints we face. Yes, we can know accelerate the writing of code 100 or 1,000 times, and non-technical people can build software, but we’re starting to see that this wasn’t actually the bottleneck. Understanding the problem we’re solving, making key design and architecture decisions and other subjective elements that lead to great products are much more challenging, and yet to be solved. 

Much of the over-hyping of AI has been driven by the fact that it’s a great hook for marketers. Those targeting businesses have long pushed “productivity gains” as a benefit - for the first time, the AI narrative has allowed companies like Artisan to say publicly what they’ve said privately for decades: “Don’t hire more people. Buy our stuff instead.” Fear sells and drives engagement - the currency of the 2020s - and few fears are more visceral than super-intelligent robots taking our jobs and way of life. Naturally, encouraged by their AI assistants, marketers have leaned into the trend.

It’s also marketing that drove the initial claims of Altman, Amodei and the rest. The CEOs of these now-giant companies are great storytellers, and their stories (more than actual results) drive the valuations of their businesses, enabling them to raise more and more money. Being correct in their predictions was a problem for later (although, as they are discovering, later eventually becomes today).

While I do still have concerns about the overall lack of preparedness for the future of work, I’m increasingly confident that AI will not, in fact, take all the jobs. Jobs will change, and we need to anticipate and train people for the shifts that will occur, but the only way we’d see an employment apocalypse is if we run out of problems to solve. That’s clearly not where we are as a species. There are diseases to cure, lives to extend, poverty and hunger to alleviate, galaxies to explore, and we need more energy to enable each of these endeavours. Every piece of science-fiction ever written shows humanity as a space-faring race - imagine how much needs to be done in order for that to happen. As long as there are issues to fix, ideas to bring to life and people to entertain, we’ll find new ways to do it, which will lead to employment. AI and other technologies will change the nature of work, but not its existence.

We absolutely need to talk more about and prioritise safety in the development of AI. The danger of an extinction-level event, while unlikely, is too great a risk to leave to chance, and AI Safety is a topic we’ve neglected in the arms race to develop the best LLM. Alignment is a problem - how do we ensure that the machines we build don’t develop objectives at odds with our own? Bad actors, too - putting advanced technology in the hands of those who’d use it for evil is a terrible idea and we’re not paying enough attention to preventing this from happening. Simple human error could be amplified exponentially as well. I’ve said this before, but the problem with the AI version of Chernobyl is that there’s a good chance none of us makes it through. Regulation, global co-operation and transparency are key. The threats are real. But they are not inevitable.

The truth is that today, adoption is still much lower than the media would have us believe. As I said in a recent blog, You’re not as far behind as you think, only 16% of people have used a free chatbot, and around 0.3% actually pay for subscriptions to AI tools. The technology hasn’t yet crossed the chasm, although it’s getting there, and if you’re doing any serious work with AI now, you’re well ahead of the curve.

Even among those using AI professionally and starting companies based on the technology, most use cases are trivial. Having spent the last 18 months in the AI start-up world through the work we’re doing at Versapia, my overall response to most of the products I see is either, “so what?” or, “oh, another one of those.” The fact that we can now build things so easily is exposing the lack of imagination and vision in the tech ecosystem. (There are exceptions. I’ve seen some platforms and applications that are genuinely astonishing - they’re just few and far between).

I remain bullish on AI, but we are in an unprecedented hype cycle. Things will change rapidly, for sure, and the near future (say on a 10 year horizon) will be very different from today. We need to prepare for that, and we need to guard against the dangers this new world brings. If you’re a business leader, you should absolutely be exploring how to use the technology available to you in your work - but that’s always true. Beware of the hype, though. It’s designed to scare you, sweep you up and lead you into short-term decisions that put money, power and influence into the hands of others.

(P.S. If you know someone who needs to read this today, send it to them and encourage them to subscribe to the Versapiens blog. If you haven’t subscribed yet, come join us on our journey through the intersection between culture, technology and business.)