How to make Machine Learning & Artificial Intelligence Platforms Relevant - Identify the USE CASE first

The gold rush of Machine learning and Artificial Intelligence is going on at a maddening speed. Every Industry and every big tech giant is talking about AI and ML. But the big question is - What is the best way to utilize AI & ML and what is its future?

               First – As with every idea and technology; the most important thing for a successful idea to revenue conversion is the USE CASE. If there is a need and ROI to justify the investments in AI & ML then it makes every sense for being future ready. But just because everyone is talking about AI& ML; it doesn’t mean that all the other technologies/platforms/applications will suddenly become useless.

               Second – ML & AI has to ensure that the link between a client’s Business needs and customized solution design is still the recipe for revenues. Businesses often imagine that they can develop a ML & AI platform which can handle multiple clients’ needs like an assembly line factory production. This might work for very selected and repetitive tasks but then all businesses are not the same.

               Third, the black box approach of ML & AI means that there will always be mistrust when it comes to data security and the actual logic being implemented. Just because it is high tech doesn’t mean that it can be explained and adapted by all.  For example, Credit risk still uses Logistic regression for scoring and though a lot of efforts are being made to implement ML & AI in credit risk; it is still not there. This is because lenders need to explain to regulatory bodies and to certain extent to the customers about why and how they are assessed for giving credit.

               However; there is no denying the fact that ML & AI will rule the roost in future primarily because now the implementation of these technologies will be more and more across the board and it will address the above mentioned blocks. Secondly, the big plus about AI & ML is in utilizing the UNSTRUCTURED data(images, audio, video) which has not being tackled by traditional computing methods that much. Techniques like Image processing by using Convolutional Neural Networks; and self learning algorithms using Reinforcement learning techniques and other similar advances will open new windows to insights which we just can’t imagine to develop using standard computing.

               At the end; we must realize that even ML & AI programs have to be developed, tested and improved for each and every new problem and the problem has to be framed with reference to the business and the end objective. It cannot be one size fits all approach and it has to focus on new horizons which are novel and thus relate to tangible business and customer benefits.