While AI is seen in various arenas, its promise has threats, doubts, and successes
Artificial Intelligence (AI) is surrounding us in many ways. We will look at some of the uses, with extra focus on ‘Business’. But first, let us consider the industry evangelists like Bill Gates, Jeff Bezos, Elon Musk who intrigued the world by saying that it poses a risk in terms of what it can do in warfare. A 2018 report by AI and security technology experts, says that digital, physical and political attacks using AI are likely to increase. They all talk about pre-emptive regulatory control and a set of best practices to mitigate these risks. We will leave the question of progress made on these for others to discuss. They all also talk of the good AI can actually do, like in social policy, education, healthcare, etc.; we will not explore here as to what this could mean in the hands of authoritarian regimes.
What AI does is play the ‘Volume’ game, because irrationality and accidents are exceptions
A significant set of AI applications is in industrial automation and warehouse management being a classic example where shipping is almost error-free. Similarly, there is medical work like surgery, where the error rate is close to zero because the robots’ fingers are at no risk of shaking. These kinds of applications need one of two things. The first is handling inanimate objects; more the standardization therein, the easier it gets. The second is expert knowledge, so that technology knows how to perform surgery. There has been significant success in such automation through AI, and we have only just begun.
The next big advantage of AI talked about is for the business world. But that is questionable in itself; and here’s why. Firstly, AI is (still) based on logical and data-driven decisions. The data problem has been addressed by this writer in an earlier column, “The King Has No Clothes”.
The other, real issue is it (still) cannot replicate emotions or multi-generational memory. It cannot tell me why I should marry my high school sweat-heart. It cannot tell me why Mona Lisa is what it is to everyone who sees it. It cannot tell me whether the next streaming serial will be a hit with viewers.
It can however tell me what kind of children my wife and I will give birth to; and what percentage of the next million visitors will like Mona Lisa; and what kind of content and promotions will attract higher viewership. It can also break some traditional beliefs (like the best first-5 chess moves). But it cannot predict what I will do next; and that is one promise AI cannot (yet) keep for business. This is important because even though the business is a rational entity, its final consumers are predictably irrational. There is nothing to stop a consumer from switching from white-only shirts to pastel shades (maybe because his teenage daughter suggested so).
It is this irrationality that will be difficult to replicate. And it is this irrationality that makes for bold leaps in human progress. We should also not forget the contribution of accidents to human progress; another thing AI cannot replicate.
So what AI does is play the ‘Volume’ game, because irrationality and accidents are exceptions (this in turn implies that AI needs large volumes of data inputs). And so AI discounts these exceptions. That leaves with majority being analyzed on data that is many times questionable, and the minority irrationality being ignored.
Lastly, there are the applications of “discovery”, or of unknown information that AI spits out by making connections that are not obvious. For example, a very high-end vacation provider (private jets and islands, etc.) would run a marketing campaign at the start of the year for the peak summer season. It produced 1-2% success rate. So they worked with a large processor, who ran its AI engine and came back with an interesting fact. The laundry bill shoots up in October for people who go on these vacations the following summer. Till now, the reasoning is not clear; but the results speak for themselves, 10-15% success rate.
So while, AI is seen in various arenas, its promise has threats, doubts, and successes.
The author managed large IT organizations for global players like MasterCard and Reliance, as well as lean IT organizations for startups, with experience in financial and retail technologies