AI and Lessons from Past Hype Cycles

Do I think that AI is the most disruptive force on humanity since the Agricultural Revolution? Yes, organised farming rewired the fundamentals of human life as we then knew it and I believe AI will have a similar profound impact. That doesn’t mean I am not sick of reading about it where every punter thinks they are an AI expert. I spend a lot of time advising senior executives about disruption and growth, and I can tell you right now that I don’t know shit. Neither do you. 

The human race is notoriously crap at predicting the future. Perhaps the one thing we can do with some level of certainty is to re-visit the past and learn from previous technology fuelled hype cycles. Please don’t take this as a slight on AI. Something can be both the biggest thing since sliced bread (and organised farming), while at the same time be in the middle of a hype cycle. Ok, maybe I am being optimistic when saying “with some level of certainty” … we are also pretty damn poor at learning from the past. 

Here are 4 lessons from the past that might help us contextualise how to approach what is now confronting the human race, and what the implications might be for business.  

Long-term impact is more profound than we think. Short-term impact is less profound than we think. 

The human brain is poorly adapted to long-term thinking. We therefore tend to underestimate just how much things are going to change when faced with new technology. When the internet landed for most consumers, with slow dial up speeds coupled with a sense of wonder around being able to access information from brochure style websites, very few were able to foresee just how ubiquitous online connectivity would become in every aspect of our lives.  

That said, we worried that physical bricks and mortar retail would die and that supermarkets would cease to exist. We predicted that kids would struggle to make friends and that a lack of social connection could even impact reproduction, putting the human race at risk. Of course, none of this came to pass, and in similar fashion, AI will change us forever but in ways that we simply can’t fathom. The legion of AI gurus confidently predict how the world is changing in front of our very eyes, we then become sceptical when things don’t materialise overnight, and ultimately fall asleep at the wheel while glacial forces carve mountains.  

It’s not about the tech. We need to think value creation which requires a whole of business model perspective. 

I really don’t care which LLM is better than others. I also don’t care about the agentic capabilities of one platform compared to another. Previous hype cycles show us that the technology converges and that competitive advantage comes from how it is used to create value. The usual suspects when it comes to innovation case studies illustrate this perfectly. Content consumption (movies and music), photography, home brewed coffee, taxis, newspaper classifieds… all of these industry transformations highlight that business models generate value rather than the underlying tech. And this means that perhaps our efforts should be focused on business model reinvention – something that has been discussed for years (think The Invincible Company co-authored by my friend and colleague Frederic Etiemble) or even decades (think The Ambidextrous Organisation by O’Reilly & Tushman from the Harvard Business Review) rather than scrambling to develop an AI strategy. If previous hype cycles tell us anything, then developing an AI Strategy is a complete waste of time. The consultants who are cashing in might disagree, but rethinking your entire business model is a far better use of time.  

Demos and flashy applications are fun, but scale is hard. Quick wins need to be balanced with ecosystem and infrastructure development.

There are so many self-proclaimed experts running around and wowing organisations with the mind-boggling capabilities of agentic workflows, autonomous decisioning or intelligence intersecting with ambient computing (e.g. fall detection or movement pattern detection in aged care settings). I too have showcased AI to curious clients – my favourite being the unbelievably natural and human-like voice agent Maya from Sesame which if truth be told scares me just a little. And yet, some of the companies I work with are struggling to get their workforce to start using Copilot. Or worse, they are still debating whether to make it available to their staff.  

Where in technology ecosystems will agents live? How will security be considered? How will organisational norms evolve around usage? These types of questions are as important, if not more, than how will we reduce head count in our call centre by deploying voice agents. Did data lakes and more powerful CRM systems deliver the utopian promise of ‘the segment of one’? Has the internet of things delivered the fully connected homes or smart cities that we were promised? I have no doubt that these will come, but the impediments to progress relate to scale and organisational infrastructure, and AI will probably be no different.  

Speed matters, but not for the reasons we might think. We don’t have to be the first movers and sit on the right of the adoption curve, but we do need to sit further along the adaptability curve. 

I would encourage everyone to be curious about AI and start experimenting now, but not because they have to be first. I recall working as a summer clerk in a commercial law firm as they rolled out email to professional staff via green screen Wang computers (yes, I’m older than I look). It didn’t matter whether they did this before or after competitors. Speed won’t drive first mover advantage any more than embracing email, internet browsing or mobile phones did for any organisation. However, what speed does provide is accelerated learnings about your ability to adapt. The winners will be those that can roll with the punches and learn how to learn.  

Of course, none of this is new thinking. That is why this is a lesson from the past – we have seen it time and time again that adaptive organisations outperform. And yet, in the excitement surrounding AI, we are once again forgetting that the winners will emerge from a small pool where culture, process, governance and operating models are all tuned for rapid iteration where exploring new territory can co-exist with core business excellence. If an organisation asks me to help with an AI strategy, I talk to them about operating model. 15 years of leading cough cough “digital transformations” has shown me that envelope funding, organisation structure and board member mindsets will be more important topics of conversation than where the AI use cases are. And, the only way to design and test these boring vanilla topics (which are a hell of a lot less sexy than building a cool agent) is with speed, because it will take time and repetition for the organisational bureaucrats to get comfortable with new ways of doing things.  

The world has forever changed. Just like my kids can’t fathom life without the internet and mobile phones, their kids won’t know a world without AI. The impacts will be more foundational than anything I can possibly imagine. Therefore, rather than trying to crystal ball and gamble on any one of an infinite number of future possibilities, organisations should look to the past and extract the learnings from previous hype cycles. Yes, AI is different, but the lessons of yesterday may just help them survive and thrive. I am happy to admit that I don’t know the answers for tomorrow. Instead of trying to see over the next hill, I am instead trying to improve my fitness for whenever I happen to be standing at that summit. Maybe our leading institutions should be trying to do the same.