Why You Don’t Need an AI Strategy

If you’re in business and not investing in – or somehow using – AI, you’re missing out, losing touch with reality – according to thousands of headlines around the world – every week. Seriously – let’s do the math: If one (or even 10) percent are betting on AI and the rest is not, which group represents reality?

Of course AI will change the world. The lack of intelligence is a non-issue at this point, AI is just a name – and is changing the world as we speak. Lots of important changes that affect every one of us – sooner or later, probably sooner. But what’s the rush? OK, things are changing faster than ever before and the AI wave is beginning to feel like a digital tsunami. But while ‘digital’ is the name of the game and the enabler behind most big changes these days, the world is still analog. 

People, buildings, cars, products, animals, farms, and so on – forever. Digital has ‘infected’ everything, mostly in good ways, but things in the real world still take time. And we – in particular those of us on the tech side, myself included – tend to underestimate the intertia, the ‘slowness of things’ as one of my professors used to say. 10 years ago I predicted – with great confidence – that autonomous cars were 3 years away, scoffing at the 30 years Moody’s (the credit-rating/analytics company) predicted. Today – 10 years later – the vehicles are here, but in a small and relatively insignificant way. My timescale was far off.

Why does it take so long? ‘The technology wasn’t ready’ is the easy way out. It’s also untrue. The technology was ready then and has improved dramatically since, but no explosion. The thing is, we tend to forget or ignore ‘the human factor’ – the one brilliantly pictured in Sully, the great 2016 movie about the damaged plane miraculously landed on the Hudson River (well worth revisiting). While decisions in the digital world are typically made in split seconds, people need time to think. And (re)act. Minutes, months, sometimes years.

Which is what happened with autonomous cars. We – the users/customers/decision makers – took time to think, and decided we didn’t want them (check out The Autonomous Nightmare). A scenario the pundits and technologists hadn’t taken into account in their (our) predictions. Our expectations were simply out of sync with reality. 

Does that ring a bell? It should. Because that’s exactly what’s happening with AI these days. The pundits, vendors and countless sales people claim that most of the world is falling behind. It’s not. It’s a race – and a big one, but for the few – for now.

They are probably right about the effects and importance of AI, but their time scale is completely off. Nvidia’s incredible rise on the stock exchange is real enough, but it cannot last because the underlying numbers don’t support it. In fact, there is no magic in their technology. It’s not even well suited, it’s simply better than anything else available right now. Nvidia just happened to be in the right place at the right time – and they had the ability to take advantage of it. Good for them, good for their customers, but it doesn’t change the world overnight (see AI isn’t Magic, it’s Just a Computer). And it will not last.

We’ve been here before. Many times, and the similarities with the microprocessor ‘revolution’ (late seventies/early eighties) are particularly strong, with Intel in the role currently occupied by OpenAI. The catalyst. Too far back for most people to remember, and admittedly a different world, but not all that different if you adjust the metrics. Around 1980, new microprocessors were popping up left and right and threatened to turn all kinds of jobs upside down. 8 bits, 16 bits, 32 bits, megahertz and megabytes, falling tech-prices, Silicon Valley exploded etc. Another goldrush, everyone wanted a part of the action.

What happened? The world changed completely, just as predicted, but slower. 2 years became 10 or more and we, you and I, adapted to new opportunities, challenges, expectations. The digital age started, enthusiasm and scepticism flourished, we got PCs, supercomputers, ballistic missiles, smart bombs, smart watches and streaming. And AI. Not overnight, but slowly, surely. 

What’s particularly interesting given today’s AI-frenzy is why it took so long. From the discussion above we could assume that the ‘human factor’ was to blame. And it was, but only partly so. The big brake was businesses, politicians etc. Established regimes in which so many are heavily invested that they will resist (any) change at any cost. ‘The machine’ just couldn’t be accellerated.

Another example: I’ve been preaching – in books, posts, analyses, lectures, discussions and more – digital transformation for 15 years, conveying what I thought was obvious: The ‘big switch’, how to approach it and the consequences of not doing so. Plus the urgency. The result? Well received, seemingly well understood, but few took action with the prescribed urgency. And – contrary to my predictions – most of them are still around, still ‘transforming’ but not rushing. What happened?

Same story: My urgency ignored the human factor and ‘the business inertia’. I tried to get them ahead of the pack and forgot that most organizations don’t want to or cannot be ahead. They are – and want to – be average.

Again, this is reality and it applies just as much to AI as it did to digital transformation and computerization before that. By the way, again going back to the microprocessor ‘revolution’, it’s interesting to notice which companies turned out to be the biggest ‘brakers’, working hard to slow down progress: Microsoft and Intel. A story for a different time and date, but Microsoft effectively prevented the world from taking advantage of 32 bit microprocessors for almost 10 years. Quite an achievement. What happened was the usual – that the early winners turned around and spent more energy blocking/fighting competition than moving technology ahead. It was wild for a while – as documented in numerous fascinating books, such as Breaking Windows (David Bank), The Microsoft File (Wendy Goldman Rohm), The Software Conspiracy (Mark Minasi), Inside Intel (Tim Jackson). EWntertaining and educating today, frustrating back then.

Here we are, watching another season of history repeating itself, this time with AI at center stage. And again it’s a combination of hardware and software, inflated expectations, big money, looming fights (wars) and lost opportunities. How to deal with it? Unsurprisingly, what worked well through previous ‘revolutions’ may well work this time too, like carefully hitting the brakes, cut through the hype and ask ‘why’ or ‘does this make sense’, ‘when’? 

I read the other day that Klarna, the payment processing company, is using AI (presumably chatbots) in customer service and has effectively replaced 700 heads while reducing the duration off service calls from 11 to 2 minutes. Impressive, isn’t it? Except we know nothing about the metrics. Such as how much of the former 11 minutes was wait-in-line time and how many of the new 2-minute calls are short because the customer just give up talking to the stupid bot at the other end? Was Klarna’s customer service bad or good to begin with (it was actually very bad) and has it actually improved?

That said, there are plenty encouraging stories out there about successful applications of AI. Unsurprisingly, they all have something in common: Realistic expectations, relevant, controlled and curated data, narrow target (check out the post Why Humanizing AI is, well, Stupid …). That’s where the ChatGPTs of the world are destroying the market: Creating unreal expectations, delivering unreliable (often fictitious) results and trying to cover everything from world history to quantum mechanics and psychology. Of course it has to crash – and the real winners aren’t the big companies with huge budgets and datacenters, but smaller shops that understand the technology, the challenge and the opportunities. They don’t create or sell AI, they deliver solutions.

This is where most of us should be looking. We don’t need an AI strategy for that, but we do need a good understanding of the (changing) opportunities. Which is business as usual – and more than challenging enough. We do need to move faster than yesterday, but what else is new?

Well worth repeating, the key is understanding – not the technology, leave that to the experts, but the concepts and the key ingredients, such as such as the importance of data – quality, availability, curation, balance etc. And remember, “knowledge is vital, but knowledge without understanding is useless” (Thucydides).

Boottom line: When the AI sales person – internal or external – comes along, ask for the business case, not the market stats (or market hype). AI is a tool – think of it as a car. The fuel is data, the drivers are (your) people who understands your business, the road is your business strategy. 

Do you still need an AI strategy? Maybe – at some point. For now a plan is good. A plan in which section one is ‘Why?’, section two is ‘How’, section 3 is ‘When’. Keep it short – adapt regularly. Sounds familiar? It should …