Cracking the code
Abhijit Akerkar, Head of Applied Sciences Business Integration,Lloyds Banking Group (UK)
Conquer those AI riches by understanding data-to-value
Over $35 billion has been invested into AI by companies around the world in the last year alone. But two in every three of those companies are yet to generate any tangible value. And of those companies that have turned the corner, almost half are still at the sub-scale level. The good news is that the value is not elusive. Companies such as Google and Amazon generate over 50 per cent of their profits using AI. How can your company crack the code to conquer those AI riches?
The answer starts with understanding the data-to-value chain. The biggest mistake companies make is to have a single-minded focus. Building a data science unit that can build models. What they forget is that the machine learning model is not equal to value. A model is only one component of that value chain.
Let us assume that your data science unit has built a model to predict the call demand into your call centre. Your model is highly accurate. Would that result in lower wait times for your customers? No. Unless you have thought through the full picture. Let us assume that your model predicts 40 per cent more calls during 2-4pm on coming Tuesday. You will need 40 per cent more agents to cater to the surge in demand. Does your employment contract allow you to ask agents to go to the office for two hours? If yes, do you have 40 per cent more seats in your call centre? If not, does your system allow your agents to take calls from homes? This is a classic example of the last mile challenge.
At the other end of the spectrum is the first-mile challenge. The value chain starts with data. This is where many projects get knocked. Building the data foundation is like moving a big mountain. You will have to find the right data sources and get hold of relevant experts. You will have to get policy approvals to transfer data from those sources. You will have to overcome the hump of poor data quality. And more importantly, you will have to weave together the data innovatively from different data sources to make the impact.
The critical thing to bear in mind is that the quantum of value you will realise will be determined by the weakest link in the chain. For example, if your talent or engineering link is weak, you having a fantastic last-mile capability will not help you in realising the full value.
How high can your upside be? That will be determined by your business strategy and portfolio selection of use cases. Around 80 per cent of use cases that companies are pursuing are focused on cost reduction. Those use cases will not deliver the most significant returns. The sweet spot lies in using AI as the evolution of decision making.
Finally, insurance is a must as you venture into your new-found adventure called AI. The best downside protection will be the risk governance framework. These machine learning models come with new risks. Your organisation will not be prepared to cope with them. Mitigating those risks will help you keep the value you would have realised.
- The views expressed above are personal.
The first step is to understand the data-value chain to help generate profits for the company.