(As Published in Indo French Chamber of Commerce Journal)

A narrative on AI – the economic implications, myths, and way forward

For sure, Google Maps helps you reach a destination far more accurately than it used to five years back. I remember a nightmare with Google Maps in 2011, when I had to find an exam center in Coimbatore, India. I was lucky enough though, to have started my journey, well in advance. I narrowly managed to get to my center on time; a place that was 47 kms north of where Google Maps landed me up. I took a pledge back then, to never rely on the App, only to be broken by myself in 2015. Today, I rely on it for almost everything, from navigation to finding a restaurant.

Well, this is AI in real for you. With more data about places, routes, vehicles, traffic, users and routines, the AI layer on the tech stack of Google Maps – a probability theory based algorithm that processes information to users – is just getting better by “learning” every day.

Is AI new, or the “AI buzz”?

Relying on computer models to predict an outcome, or perform an action has been very much in practical application for years in weather monitoring and auto-pilot functions. Hence, it wouldn’t be inappropriate to describe that the buzz around AI is what is new. However, the buzz is real and has merit, unlike other bogus trends that come and go. The topic of AI has gained much importance over the last five years. AI has, in fact, become a field of study in itself, given the numerous use cases and applications that it is likely to exploit. 

But why should you not ignore the buzz; let’s examine. The real impact of AI could be much larger than what earlier breakthroughs such as computers and internet have created. This argument has been repeatedly endorsed and subscribed by CEOs of world leading tech companies such as Microsoft and Google. To add, there was a recent proposal made to the Chief Minister of the State of Kerala, India to setup a dedicated Ministry for Artificial Intelligence just like the ones for Agriculture, Electricity, Fisheries, Industries and Information Technology.

But to understand AI in its complete sense, it is significant to have an over-arching understanding about the non-tech quotient of the subject.

Understanding the layer of AI

According to data science theory, all work, decision making or process flow is either deterministic or random in nature. However, in practical sense, a lot of the work we do, decisions we make, or processes we design, fall somewhere in between the two extremes, and force us to use probability to perform them. However, emotions creep into our thinking and influence the results of such decision making. Interestingly, we are often unaware of these calculations, just for the fact that they do not explicitly demand one using a pen, calculator or computer. Driving a car could be a good example to start with. Human drivers do not use calculators or computers to evaluate the degree of turn they have to provide to a steering wheel before negotiating a curve. The process of maneuvering a curve appears so deterministic that one might think about a mechanized robot to perform the task with much ease. But in reality, there are a number of calculations that the robot would have to make before performing the move, not only for the first time, but every time. A variety of variables such as obstacles, speed of the vehicle, condition of the road, and even weather make a difference and influence the calculations. Hence, even such a simple task cannot be automated without providing room for being judgmental. This calls for intelligence, and when a machine is asked to perform such a task, we call it “Artificial Intelligence”.

But the fact of the day remains that AI has become increasingly mis-quoted in cases where no such application of probability is required. Generic automation or process automation is often portrayed or conceived as AI. Process automation is use of technology or engineering to remove the human component out of a repetitive, highly deterministic and mundane task. This could be as simple as asking your mail client to send an automated “out-of-office” reply mail, to large complex robots that fix parts in an automobile assembly line. The code of work here is structured, defined and programmed in advance, and the machine does not “think” before performing the task. Thus, robotic process automation, RPA as it is known, is not necessarily AI.

In application, AI is a set of algorithms that will “teach” a computer or machine to “think” on a task in real-time. The AI algorithm will stop with providing thought invokers for the computer on how to use available data and make judgments. It is just like how teachers in primary school taught us about addition and subtraction; we used them lessons while settling change with the grocer. 

Is AI only for the Tech?

The notion of viewing AI as new tech should probably be the last thing to do. AI is a field of study, based on math, more specifically on probability theory. A person good at math is more likely to fare better as an AI scientist, than someone who knows to code. Going forward, we will have more and more computer applications that will enable non-coders to write AI algorithms with ease. To draw a parallel to this, let us consider an example. I’m a management graduate, which means by default, there are chances of me using MS Excel, Word or PowerPoint in a better way than one who is coding software at Amazon or Google! I apply my domain knowledge in business and finance on different computer applications that help me perform certain tasks. No one calls me a techie though I spend 95% of work time in front of a computer!

What are AI Companies

To understand the nature of AI companies and the value they bring, it is important to learn the evolution of companies from revolutions in the past. The agriculture revolution called for innovations in food processing, textiles and garments, packaging, storage, transportation etc. The discovery of engines paved way for the machine tool industry to develop. Machine tools were capable to make other machines that found applications across different existing industries. The invention of electricity and its transportation accelerated the appliances industry – smaller machines for the home, and sowed the seeds for the electronics revolution. The electronics industry hit pay dirt with computers on one side, and communication systems on the other. The merger of these two technologies gave life to the internet. The internet came up with a number of new possibilities, and perhaps today, internet companies dominate the world in market cap. 

In the chain of events, the opportunities that every significant invention threw were taken up by others, and not the ones who made those inventions themselves. The internet for instance, saw the dominance of Amazon, Google and Facebook, and not Ericson, AT&T, Cisco, HP or Dell. While the latter pioneered in the underlying invention and worked towards making that better, the former created application layers on top of the underlying technology to create enormous impact and value. The fortunate ones: Apple, IBM and Microsoft, stayed abreast by constantly evolving their business models to the internet era. However, even today, they are not in the checklist of internet brands, though their businesses are largely based on internet based products and services.

The same is true for AI. It has the potential to disrupt existing norms in ways that we could never have imagined. Companies like IBM and Google, who are pioneering an AI-first strategy, will essentially provide the foundation for newer start-ups to break out. IBM’s cognitive solutions clearly provide an indication for such a transition; and today, a majority of the “real AI” start-ups rely heavily on IBM’s platform and services. A new breed of companies may emerge from the AI revolution and create the next phase of dominance.

Like the internet, AI will have different business models emerge in the near future. In the internet economy, we have companies like Google and Facebook that make money on the internet itself, while companies such as Amazon, Uber and Netflix that use the internet to deliver products and services. (Amazon Web Services (AWS) being an exception though). In AI too, a set of companies will sell AI solutions (eg. Google Maps, Chatbots), others will use AI [as an enabler] to deliver other products and services (eg. autonomous cars).

AI Deception

There seems to be a lot of undeserving attention given to two fallacies around the topic of AI. One is the myth of a jobless future, and the other is false AI claims made by start-ups.

Myth of Joblessness

Since time immemorial, the future of anything is uncertain. Hence, it wouldn’t be appropriate to be judgmental on the fate of jobs on account of AI. In fact, arguments of a jobless economy are speculative deterrents that need to be shied away. The ancestors of these speculators have seen their names go to the graves when more jobs were born out of every revolution, in contrast to their predictions. It is rather a lazy excuse to keep one from not exploring the opportunities around. Unfortunately, the momentum of such erroneous claims induces a climate of fear and resistance. The appeal to public at large should be to be receptive of the new wave, explore the opportunities, and build the necessary capabilities for an AI-dependent future.

False AI Claims

A recent study revealed that 40% of AI start-ups in Europe never deployed AI algorithms. With the precedence of Theranos, a once acclaimed bio-tech start-up, the cited report on AI startups could easily be interpreted in the global context as well. AI is being increasingly quoted by Companies, regardless of their orientation, size and sector, making people to guess the role of AI. On one hand, large utility companies are using AI based solutions for optimizing their load factors. We can also see large EPC companies like SNC Lavalin, mention AI in their context on their website. A number of startups are adding to the noise. Most of these claims are as equal to the claims made by Companies who only had a website, but yet called themselves as internet companies during the dot com buzz. Without doubt, it is evident that there isn’t much clarity around the subject of AI. Companies make wild and bogus claims to seduce investors and customers, who are themselves just about learning the subject and its potential. 

AI today – where are we?

AI has made rapid progress in the last two years. However, given its wide canvas, there are miles to go forward. AI has started to crawl into our daily lives with simple toolkits such as the new Gmail which suggests content for your mail, the Alexa personal assistant and the improved Google Maps that matches a newly opened restaurant based on your preference. AI has also made significant advances in health-care, where health-tech companies are able to predict the chances of cancer, heart disease, diabetes, or obesity, well in advance than with conventional diagnostics. We also have a number of startups working on cyber security, digital identification and forensics that use AI to predict fraudulent transactions. Without your knowledge, Banks and Insurance companies are moving the trust protocol to an all-digital environment with the help of AI based technologies. Autonomous cars may roll out for public use anytime soon. Luxury segment cars already deploy a number of Advanced Driver Assistance Systems (ADAS) which use simple AI applications to keep the driver engaged and content.

Important caveats

Today AI is a naïve field of study. It comes with pre-loaded conditions for it to work. 

At the crux of the subject is the belief that algorithms will out-perform the human brain’s intelligence; as they will (1) be developed on collective and comprehensive data from varied and diverse sources than any one single brain could have ever gathered, and (2) have removed the emotional bias from the equation of decision making. However, this belief could still be grossly wrong, and have been stemmed out of an underestimation of the human brain, and in particular, the human mind. 

Another important idea that forms the foundation of AI is the belief in majority. AI works on probability theory which in-turn relies on data distribution. For AI, the majority is always right. In fact, this is how even we work in the real world. However, the larger question here is whether the minority was wrong in the first place. In our day-to-day context, we sometimes ridicule the majority by using phrases such as herd mentality. So is it still right to go by this theory?

The last, and probably the most important of all, is around the authenticity of data itself. Data is probably the most important yet vulnerable asset of our times. Even as companies are in war against data manipulation, the trust quotient is still struggling to find its place.

Way Forward

The boundless opportunities of an AI future outweigh the fewer challenges that come with it. But the way forward should take holistic cognizance of our migration to AI, unlike the rollout of the internet. 

The pace of the internet revolution not only left behind a digitally excluded community, but lawmakers and governments too. It took more than twenty five years for a General Data Protection Regulation to take shape. US Congressmen are still figuring out how tech companies such as Google and Facebook operate. The Indian government introduced e-Commerce rules in 2019, ten years after the birth of Flipkart. And, the digital divide is real, and a cause of serious concern. 

AI has the potential to probably change the narrative of our lives, far more than what internet has done to us. We will have harder problems, tougher questions and conflicting stakes that will emerge in the rollout, or perhaps even later. Governments should be capable, agile and open to explore, understand, and govern the different aspects in which AI will influence individuals, communities and nations.