Write to us
Contributors
Writers
Designer
Artists
Concept and Direction
Enquiry
According to Gartner’s Hype Cycle for Emerging Technologies, 2017, Artificial Intelligence will be everywhere. It is being seen as the most disruptive class of technologies over the next 10 years due to radical computational power, near-endless amounts of data, and unprecedented advances in deep neural networks. In the coming years, organizations with AI technologies will be able to harness data in order to adapt to new situations and solve previously unknown problems for business competitiveness. Gartner says that enterprises looking at AI should consider technologies like Deep Learning, Deep Reinforcement Learning, Artificial General Intelligence, Cognitive Computing, and Conversational User Interfaces to be relevant in the coming years.
Arup Roy, VP, Analyst, Gartner India, tells us how AI can be relevant currently and how the organizations should go about deploying this and what to watch out for:
Interviewed by Priyanka Bhattacharya
Artificial Intelligence (AI) in terms of usage in business cases is still in the early stages. There’s still a lot left to be desired. Homogenous adoption is still way off since organizations are in the process of evaluating the technology in real life scenarios. Unlike many other hyped-up technologies of the past, we do not see the hype about AI dying down. In fact it is not a hype anymore; it is slowly moving mainstream to becoming a way of doing business.
It is now in the 3rd or 4th era of consuming IT. The world has moved from client-server era to internet based computing to app-based economy. In the fourth generation it is about AI, where the whole world is squeezed into chat platforms powered by intelligence. Intelligent apps that will help you in your purchase, offer you usable advice on stocks etc. The possibility will be immense. The human-machine interaction will change. It will be more life-like, as in a human agent.
In the enterprise today, in terms of AI adoption, processes like help desk or service desk can look at AI enabled solutions. AI apps today are mature enough to help in an organization’s business processes. Other than that, conversational agents are also becoming a mature solution that can be deployed within an organization.
AI will have different flavours – it will have distinct usage for the enterprise and for the consumer. In an enterprise set up, a virtual bot can be used to manage internal corporate needs like IT services. In a B2B scenario where B2B clients or end users are serviced or connected, we could see conversational bots. For example, take the scenario of a bank where an associate joins the team, and a virtual assistant can be used for on-boarding the new employee and training on all the bank’s processes. In a B2B scenario, the same bank can use virtual assistants and AI enabled chatbots to service corporate customers or even retail customers. In fact, the bank can use this AI-enabled solution to even keep track of its external vendors and ensure smooth processes with the company’s systems architecture.
My first advice is that if you are starting out with the thought process of ‘how can I leverage AI?’, then forfeit it. That should never be an organization’s way to deploying AI related solutions. The central focus should be on digital transformation. It should be about what relevant solutions are needed to be digitally ahead. The CXOs need to flesh out services and processes they need to remain competitive. They need to address the question of which direction the business should take and how best to service the customers. Reinventing the business model is more important. Only when that has been worked out should the technologies that enable it to be looked at.
There are a few things that you need to take note of when working out your AI-enabled digital transformation strategy. You should look at the financial impact, customer impact, business impact, the way you can execute the transformation. If that’s in your digital transformation strategy, then explore how AI can impact the points mentioned. You need to draw a business case – look at how AI can speed up your processes, help in transactions, benefit in operational terms, and the cost of acquiring the AI-based solutions. When the benefits outweigh the cost and the risk, then you should evaluate and deploy.
However quite contrary to this, the CIOs today are often mandated with deploying AI-enabled solutions without really understanding the business needs. The request is often “We need chatbots, so please implement them,” and the entire department gets going on deploying solutions to enable chatbots within the processes, without a serious thought about whether they need it or not, and are these chatbots really helping in the overall scheme of things.
So, I would say that instead of being aware of AI and looking at deploying the technology, evaluate whether you have a proper roadmap drawn up for digital transformation. Do you have an enterprise-wide thought process? If you are comfortable with the answers, then look at the solutions.
In your evaluation process, to avoid AI washing, look at proof of concept from the vendors. See how well it is automated and how robust is the solution, whether it is scalable and flexible, how much effort is required to deploy it, the complexity of the solution, whether it has tools to self-learn, the concept of the solution, the cost of maintenance, and how effective are the results. Then you need to draw up the governance of this tool with the other tools within the organization. This is the framework which you should use to ensure real AI deployment in your processes.
Deep learning, AI-based solutions, advanced statistics, linear regression, branches of machine learning, ability of the technology to deliver the right results… these are some of the things that CCXOs should start thinking about.
Though right now one of the reasons which can hold back the deployment is the expensive deployment cost – the cost of hardware required is high, but the compute is getting cheaper. Also, another issue is availability of the right skillset among the engineers, and the solution providers. The understanding of these technologies is still very nascent, so many are unable to make the solutions truly effective or usable. There needs to be a good understanding of neural networks and how it impacts the overall AI technology. It is a challenge and this resource-related challenge can slow down deployment. We still have to work on that.