New Answers Need New Questions

Business leaders, even CIOs, are locked into old ways of thinking, shaped by what their systems are capable of. In this blog post, we’ll help readers ‘think the unthinkable’ to get the most out of AI-enabled data.

Unthinkable thoughts

When I look at where we are with AI, I’m struck by how often the word ‘unthinkable’ comes up. From the second-order effects we can’t imagine, to the patterns and connections we can’t see (which are revealed to us by AI/ML), and even as far as innovations like object detection in physical implants might one day reliably let people see for the first time. The whole world of AI is (very excitingly, in my opinion) filled with things that are beyond our cognition—and yet they’re too important not to think about.

 

The funny thing is that AI—supposedly a big new force in tech—is anything but new. When I was at college in [redacted], we had machine learning and neural networks—though admittedly a lot less complex than we have today. What has changed is availability of compute power, which now allows for models with millions of layers, and trillions of hyperparameters. Add on increasing global collaboration, free online learning (both turbo-charged by real-time AI translation, which has had a transformative effect in widening participation around the world) and open-source tools/code, and we find ourselves at a unique moment in time. The processing has caught up with the theory, and things are getting very exciting.

AI is already all around us. The biggest shock I see when talking to customers interested in exploring AI is just how much it already inhabits daily life. If you’re using Microsoft Office, Uber, an airline, or a bank you are already using AI-based capabilities..

Maybe that’s because we don’t call it ‘AI’ when we encounter it as end users. We call it ‘avoid traffic?’, or ‘background blur’, or ‘recommended for you’. But those the algorithms behind consumer AI power other uses. The shape recognition that enables background blurring on a smartphone photo app is the same tool that recognises a tumour during ML-enabled surgery.

That photo background isn’t the only thing that’s blurred. AI is increasingly part of our personal and professional lives.

Some businesses are transformed by AI, while some are built on it. Uber, for example. Or StitchFix, a pioneer in online personal styling that has leveraged machine learning to disrupt fashion retail in several ways. StitchFix can recommend outfits based on stated preferences, as well as creating recommendations based on social media posts, previous purchases and returns, and buying patterns. A (human) personal stylist then makes the final selection: a great example of hybrid AI enhancing a human’s work, not replacing it.

But they have many other uses for AI: what to stock in each warehouse, which warehouses to associate with specific customers, and even which stylists to assign to who. All that means StitchFix now has a significant amount of data on style preferences and buying history, sliced by demographics and behaviours. Which means they can now mine that data, understand trends and become a source of insight for design houses and retailers. It's a data company as much as a fashion outlet.

Other industries and processes exist without AI, but can be hugely enhanced. Pharmaceutical drug discovery, for example. AIs can check compound interactions at a scale and consistency that humans can’t get near. You could put 1,000 scientists in a room, working in parallel, and they still wouldn’t get close to what AI can manage—we’re talking billions of checks in hours. But this doesn’t mean replacing scientists – this AI analysis means human researchers are investigating compounds already shown to have a high chance of effective and efficient outcomes.

So ask yourself, ‘what data do we have? Can we do something new with it?’ Perhaps you have massive amounts of customer data you could be slicing in new ways—across preferences, geographies and more—that could be informing everything from forecasting and logistics to product development and advertising.

Done right, AI can help you think previously unthinkable thoughts about your business. Efficiency and automation are great, but we’re talking about finding a new competitive edge and even, as in the case of StitchFix, a new business model.

And you know what? That’s fine. I’m human. Which means I’m impulsive, sometimes irrational and maybe a little superstitious at times. They way my brain works is (currently, anyway) ‘unthinkable’ to AI. AIs are rational, consistent, and definitely don’t love the Beatles like I do.

If we thought like AI, we wouldn’t have to invent it.