AI in your enterprise

The term ‘Artificial Intelligence’ (AI) was coined by John McCarthy in 1956 and has had a complicated journey to arrive at its current status as an independent branch of study and innovation. The concepts associated with AI today were developed by personalities as diverse as the sci-fi novelist Isaac Asimov, mathematician Alan Turing, cognitive scientist Marvin Minsky, and many others. Punctuating the path to the present were advancements like the SHRDLU natural language parser developed by Terry Winograd at MIT, the pioneering Shakey the Robot by SRI International, and clinical expert system MYCIN by Edward Shortliffe at Stanford University, among others. While all these milestones were major developments, AI nevertheless was for long stuck in still-not-quite-there territory until IBM’s Deep Blue created a sensation by beating chess superstar Garry Kasparov in 1997. This event marked a steep upward turning point in AI’s trajectory.

In the present stage of its accelerating evolution, AI is all about learning algorithms: supervised, unsupervised, reinforcement, and deep learning. In supervised learning, regression algorithms and their kin are commonly deployed, along with decision trees and random forests. Supervised learning employs historical or training data to help make predictions. Unsupervised learning lets AI machines learn from raw, unlabeled data, often unstructured, using methods such as cluster analysis. These techniques help with tasks such as customer segmentation or sentiment analysis. These techniques help with tasks such as customer segmentation.

Methods like Q-learning often support reinforcement learning to calculate expected future rewards for each action. A rather famous example involves the AI program created by Google DeepMind to play Atari games. Over time, the algorithm understands the implications of its actions better via a ‘reward signal,’ and improves itself. This makes reinforcement learning very similar to how humans learn. Deep learning, based on artificial neural networks, has in recent years become quite a buzzword because of its widespread conspicuous applications. These include autonomous vehicles, fraud detection, image and speech recognition, and language translation, among others. Deep learning algorithms’ performance and ability to solve complex problems increase with data availability – not infrequently to the stunning standards of science fiction, justifying the buzz.

Everything we’ve discussed so far has focused on the ‘what.’ However, AI-driven cognitive solutions are all about the ‘why.’ The problem will drive the algorithm and, therefore, the solution. But that doesn’t diminish the importance of data quantity and quality, which plays a very significant role, as always.

Unquestionably, organisations that use information effectively have a significant competitive advantage over those that don’t. And it’s all about focusing on growth, and not just cost savings. By concentrating on only furthering automation to reduce costs, we miss hidden opportunities for greater personalisation and market differentiation. While the old cliché about not using technology for technology’s sake holds true as ever, in the age of wow-factor AI it can be hard to resist. But even for organisations that yield to this temptation, visionary CEOs and CIOs can steer their AI adoption well to avoid their competitors leaving them behind.

Read the full insight to find out more.

About the author

Nikita Dhami is a highly experienced, reliable, and efficient leader of dynamic teams, including running marketing automation and AI functions, operating in fast-paced environments. She has a background in leading and implementing end-to-end analytic solutions for businesses, which in turn, deepened her understanding of and thought leadership on data, AI and machine learning, and marketing applications.