Approaching AI As Enterprise Initiative
Headquartered in Mumbai, MullenLowe Lintas Group is one of India’s largest and most storied communication groups. Apart from his senior level professional engagement, Pravin has been actively engaged with academia like Oxford & MIT Sloans in areas like Blockchain and Artificial Intelligence’.
Artificial Intelligence(AI) is amongst a top interest area (if not a priority one)for most CXOs today. As such, AI can imply multiple things, but to put it simply it’s mainly about the collective intelligence of humans and machines.
Well, is collective intelligence something unheard of before? Not exactly! It’s been around in different forms and practises. Take an example of different groups of individuals working collectively to strategize an army move. We have many stories to relate to from the major wars across the globe where the success has been a result of such collective intelligence. More recent and more technical example is Wikipedia- which is result of collective intelligence or knowledge created by people across the globe using a technology platform.
So, AI is about finding out how can people and computers be connected so that, collectively, they act more intelligently.
• By connecting people to each other in new ways, so they can act more intelligently as a group.
• By connecting people to computers that have more artificial intelligence, more AI.
• Collections of people and computers will be smarter than the computers and smarter than the people.
At this point, it’s worth while to understand about human intelligence and machine intelligence. Human intelligence can be of various types such as Visual, Statistical and Analytical. Over a period, machines can be trained in these areas to gain
expertise. A good part about machines is that as they handle more data, their self-learning capabilities mean that they get more and more intelligence at a very fast rate.
So, when we consider the combination, what tasks should computers do, and what tasks should people do? Obviously, let machines do the things they do better than people, and people do the things they do better than machines. For instance, machines are much better than people at remembering and processing huge amounts of information. People are usually better than machines at interacting flexibly with other people.
We should also be trying to figure out how to let the human-computer systems do things better than if done individually before. As an example, let’s look at a cybersecurity system where people and machines work together to detect cyber attacks. The machines are much better than the people at detecting unusual activity on the network. But the people are better than the machines at recognizing which kinds of unusual activity are just random and which are due to malicious intent.
So, it turns out, in this case, that the combination of people and computers together recognizes more incidences of actual malicious attacks than the machine alone. We’ll need to think of the combination of people and computers as a continuously evolving system, learning from experience to be better and better over time.
• Programmers improve the machines. The Google programmers are constantly tweaking and improving the Google search algorithms, same with the Facebook programmers with many of the Facebook algorithms.
• The machines learn from their own experience with various kinds of machine learning. For instance, Microsoft’s Azure has the custom vision API which can be trained to recognise brand logos, celebrities and emotions. It just gets better with very image it handles. In the research lab are developments like events, routing and isochrones, which makes it an interesting area to keep watch on.
As we consider deploying the combination of human expertise and machine capabilities, it will be important to identify specific implications, advantages as well as inherent limitations if any. The opportunities could be around:
• Driving cost advantages running across processes, products or services. E.g. repetitive process automation for repetitive tasks, usage of robotics in supply chain.
• Creating service differentiation- usage of machine learning capabilities, providing chat bots in customer engagement
• New Business Model as a combination of these capabilities
For an enterprise adopting AI, it will be critical to create a knowledge map of this aspect to take things forward. Additionally, while we have the different types of AI- machine learning, natural language processing and robotics to consider, it can also be a combination of AI, IoT and even blockchain process which can throw open different avenues. To summarize, while it’s very much a strategic area and while it does have the CXO mindshare, it ought to follow with a structured roadmap as an Enterprise initiative.