Being able to predict better and better is the key to AI success

Publish date:

When we talk about artificial intelligence (AI) or augmented intelligence we are referring to an “artificial brain” that, via some strings of logic, behaves or reacts intelligently.

It’s not about making complex calculations fast, or faster, than the human brain. It’s about taking actions that seem very insightful and in context.

Now when you take a deeper look, it is not about intelligence actually, rather, a key component of intelligence, predictability. The all-important asset here is data and how intelligently the data is leveraged to predict (as accurately as possible) any future behavior of the subject – human being, machine, etc. It is also of paramount importance for the underlying algorithm to take insight from the outcome and learn to be able to predict even better.

There has been a lot of buzz around self-driving cars over the last few years. Although they’ve been around for some time, they operated in a controlled environment. Putting the artificial brain to work on its own in an uncontrolled environment (i.e. the streets) is a different can of worms altogether.

So, for an artificial brain (AB) to thrive in a real-life situation (without a definitive number of possibilities) the approach adopted is to make it learn human behavior, as in, “what would a human do in this situation?”

Let’s take an example:

We often like to explore different eateries near our places of work. Let’s say a group of colleagues go out for lunch every day and usually settle over 3–4 eateries in the neighborhood. Over a period of time, their culinary preferences, arrival time, and other behaviors form a pattern. In response, one of the restaurant owners takes the extra step of starting to prepare food proactively according to these customers’ observed preferences so that he can serve them what they want to eat as they arrive. A few things happen as a result. First, the customers are pleasantly surprised that the owner is giving them a personalized experience. Second, even those who weren’t planning on ordering a certain additional side dish that day are doing so since it was proactively suggested to them. All of this increases the owner’s sales, of course. So, what is he doing if not predicting by learning their behavior and identifying some patterns over a period of time?

Now, this is a very simplified example but I will try to compare it with a deeply impactful and transformational business model that has completely changed the “hired ride services” industry – Uber. Every day, Uber accumulates vast quantities customer data – what time of day there is a surge in demand in a specific neighborhood, what type of rides (personal, pool, etc …) is preferred over another, etc. When you consider this individually, a pattern can be developed with respect to the time of day a given person orders a car. Let’s assume that X always orders an Uber Monday through Thursday, between 7:30 and 8:00 in the morning for a pick-up from his residence and a drop off at a specific location downtown. He doesn’t order the ride on Fridays. This information can be interpreted to predict that X takes the ride to his office but opts for alternative transportation on Fridays when Uber prices  spike. This information and the interpretation can then be used to predict many things, and targeted propositions/offers can be formulated. Let’s say that Y takes an Uber from the same neighborhood on Fridays at a similar time. If Uber suggests that X and Y share the ride (and the cost) this would be a completely different model. In essence, instead of the customers ordering cars, Uber would offer them a ride on the assumption that they will accept it. It may or may not work, but, either way, the brain behind it learns from the response for making better predictions in the future. My point here is not that Uber will or should adopt this model. But the insight that enables such predictions may have a significant impact on the company’s business strategy, adding another business model to its bucket of services.

We live at a time when it is not too far-fetched to expect radical change in how businesses run, and institutions function. It is interesting to see how that impacts our society and economy. As Stephen Hawking once said, “success in creating AI would be the biggest event in human history.” Let’s fasten our seatbelts and prepare for a journey that promises endless possibilities, much of which will likely be driven by practical advances in the use of AI. This may all sound like science fiction today, but that is likely to change tomorrow.

Practitioners, should focus on developing algorithms for our clients’ AI enablement that not only deliver superior, insightful, and personalized services through advanced ability to predict, but also inculcate the ability to learn from responses quickly and make that second offer count even more.

I would like to hear more on this from you and it will be great to learn other views and experiences.

Related Posts

Capgemini Invent

How to approach carbon offsetting on your path to net zero

Date icon September 22, 2021

Over the past couple of years, and even more noticeably since the beginning of the global...

Capgemini Invent

Why is a renewable energy sourcing strategy essential on your path towards net zero?

Date icon September 22, 2021

It is time for corporates to accelerate their uptake of renewable energy sources.

Capgemini Invent

3 Steps to building a net zero workforce with a sustainability academy

Date icon September 22, 2021

Net zero transition requires adapted business strategies, new business models, sustainable...