“How Do I Start With AI?”: Answering the Multi-Billion Dollar Question
“How Do I Start With AI?”: Answering the Multi-Billion Dollar Question
This article gives an answer to the question of how to start with AI. Also look at employing AI.
Join the DZone community and get the full member experience.Join For Free
About a year ago, I was convinced that the key to succeeding with Artificial Intelligence (AI) was to take a platform approach. In other words, the synergies that accrue from appropriately bringing together the range of technologies that are making AI a reality for enterprises was, I believed, the way to go. I still firmly believe that.
In fact, having personally met over 200 executives (business and technology) since then, from around the world, who seek to find relief and new value from AI, I am convinced that opting for best of breed capabilities from a variety of vendors is not necessarily going to work out in practice. For one, despite claims of using only open standards in building these offerings, deploying the offerings from a variety of vendors in an integrated manner is a challenge. Further, the business and operational challenges that naturally occur in such situations with multiple providers are deterrents too.
In my conversations with the aforementioned executives, it also became abundantly clear that adoption of AI, while desirable for many, is still a daunting proposition for a great number of them. This made it clear to me that something beyond a mere platform approach is needed. After all, no one wakes up in the morning going, “Today, I must go get myself an AI platform.”
Path to AI, a recent survey conducted by Infosys, reveals that while many recognize the gains that AI can bring to their industry and, by extension, to their company, only about 20% have created a strategy to take their organization from point A (automation) to point B (AI adoption). This is despite decision-makers (CEOs, heads of innovation hubs/centers of excellence, CTO/CIOs, operations heads, and IT personnel) recognizing the fact that even at the outset, AI can bring improvements. The research shows that only 18% have a complete AI strategy to manage their transition from intelligent automation to AI. A staggering 55% state that they still don’t have a strategy in place despite planning to make the shift.
Perhaps instead of approaching AI as an organization-wide solution, they ought to have the option of starting their journey with a manageable chunk of effort that brings quick success. What experience has taught us is that when attempting to introduce new technology it is often better to do it in small chunks with a greater likelihood of success.
One approach could be to start with robotic process automation (RPA) and realizing the benefits that come with it, and subsequently making the transition to more cognitive automation that leverages AI capabilities more directly and deeply. Perhaps, yet another approach could be to use bite-sized non-disruptive applications that solve unique pain points by leveraging AI capabilities. Let us consider each one.
Let us take, for instance, a large telecom group managing multiple companies in mobile, IP TV, broadband etc. faced with challenges related to customer service. In this situation, it is required of their agents to coordinate effectively with field technicians, such that issues can be resolved in real time. Using an automation platform, they can integrate various applications and data across companies enabling agents to access the right expertise to address customer concerns in real time. Such a solution can deliver a significant increase in customer service with a dramatic reduction in query handling time, thus resulting in substantial cost savings.
Having started this, the organization will have gained the ability to manage its vast treasure trove of data in a disciplined manner and learn the discipline of working with data for greater process efficiency. This prepares the organization to become more successful with some of the more involved AI technologies, such as Machine Learning, which by its nature must employ large amounts of data and varied data sets.
If that is not of interest to an enterprise, then they might consider an easy-to-deploy AI-powered app that delivers better time-to-value. For instance, a credit card issuer that’s facing issues with cardholder churn, high operational costs, and revenue leaks, can use an AI-enabled business application that is geared for fraud detection. It can unobtrusively work with the company’s existing systems to provide alerts every time a fraudulent transaction takes place. Or, for instance, in the matter of sourcing and procurement, an organization can benefit a great deal by automating functions like procurement data ingestion and data management activities, eventually classifying spends correctly with AI. It will help the business develop top-notch procurement practices that will help them decrease spend and enhance risk management.
The key here is to understand how organizations will best consume AI. In recent months, it has become abundantly clear to me that customers across different industries and geographies are almost universally inclined to accept and adapt faster if they could do so in small bite-sized portions, which helps them get started. Of course, this depends on where an organization is in their own AI journey — a continuum which ranges from needing basic automation or even simple applications to mid-level automation to cognitive automation to needing more evolved AI.
From the aforementioned research, it is clear that while 75% of the respondents are seeking an enhancement of productivity and efficiency, 52% are looking for cost optimization, 29% for scalability in business, 18% for a head-start in the AI journey, another 18% for employee reskilling, and 10% for various other goals like speed and accuracy, compliance, better customer support, automotive safety enhancements and better business sustainability.
What this shows is that the decision to undertake the AI journey irrespective of where it starts from is clearly a function of the business situation an organization finds itself in and its place in its overall lifecycle.
So where do we go from this realization?
Netting It Out
Employing AI to stay competitive and realize new value is likely the going forward trend, not because it is a fad but because we can actually make a palpable difference with technology now. The best way to succeed in this matter is to not employ a traditional best of breed approach when it comes to selecting capabilities — a well-integrated suite (preferably also quite modular) is the way to go in order to future-proof the enterprise to the extent possible. More importantly, a clear business decision has to be made about whether RPA will be the sharp tip of the spear penetrating the fabric of more traditional organizations, or will the cause be better served by an AI-powered application such as procurement or fraud detection and management.
With the first steps taken, the organization should go boldly where it hasn’t before — deploy more solutions that take advantage of the underlying platform’s capabilities.
If one is to succeed thus, then an organization must start by eschewing the notion of working with different vendors for each little matter and seek out a partner that can provide such an entry point with a platform to back it so that the necessary growth is unimpeded by integration and unnecessary data sanctity issues. Perhaps even more importantly, this approach gives them the ability to taste the benefits of AI in little chunks before they embark on something more ambitious; building on those small initial successes can go a long way in gaining organizational buy-in (perhaps a subject of another post).
Are you ready to take on your AI challenge?
Opinions expressed by DZone contributors are their own.