By Aashish Kalra
The convergence of Big Data and Cloud, powered by Artificial Intelligence (AI) and Machine Learning (ML), will be the next game changers. AI is transforming the way businesses are defined, designed and delivered and enterprise value is captured. AI will be integral to every disruptive business to gain customer and competitor insights. According to Gartner, by 2020, customers will be able to manage 85 per cent of their relationship with enterprises without human intervention.
With AI taking a center stage in operations of enterprise, it is imperative for businesses to blend it with their existing infrastructure to analyze business patterns, derive meaningful insights and solve problems in a faster way. Businesses should realize that success will increasingly depend on people and machines collaborating with each other. This will not only drive efficiencies, but also create new forms of growth and innovation. According to a report by Accenture, investment in AI and human-machine collaboration could boost revenues of organizations by 38 per cent by 2022.
Data is the next natural resource, like air, oil and water. An AI model must be trained on a huge and comprehensive dataset to perform the required task. The probability of error is higher when the application has compromised with datasets.
AI solutions are best when coherently used with other supporting and well-integrated architecture. Hybrid platforms are in abundance to find, which when interconnected with AI delivers faster data processing, deeper business insights and smarter business decisions. Businesses need to identify the right kind of IT platforms that brings AI together with their existing business model as a package and helps in improving the core functions.
Bias and variance are two important notions in AI and ML. They are indicators one should always keep a close watch on when training and remodeling is done. One of the biggest goals of AI is to train a model that can be generalized to new data. If the model is incapable of correctly predicting on new data, then the training is less effective. Businesses should consider adopting different multi-fold cross validations which will make their models capable enough to handle all internal and external discrepancies and fetch accurate and real-time results without bias and variance.
Enterprise initiatives are mostly dominated by two modes of AI/ML. One is Active AI/ML in which people directly determine the role of AI/ML to get the necessary job done. The other is Passive AI/ML, in which the algorithms largely determine people’s parameters and processes for getting the job done.
The humans are in charge in the former while machines are in the later. In organizations that possess more data resources, the active AI/ML model there needs more oversight and disciplined supervision. In passive AI/ML, the subversive risk is that it is too rooted in human compliance, adherence and obedience. In other words, the machines tell the humans what to do.
Organizations need to enact inter-related initiatives to mitigate the risks and strike a healthy balance between the risks and opportunities as there is no solution to the challenges which include making a declaration of machine intelligence, employing radical repository transparency and creating a trade-off road map as they would suggest where management believes active AI/ML investments will be more valuable than passive ones.
With the convergence of big data and cloud, AI and ML has got on the radars of businesses that would like to innovative and stay ahead of competitors. These technologies are driving innovation and transformation across industries and there are no arguments about their influence and power in the succeeding decades. According to a Narrative Science survey, 38 per cent of enterprises are already using AI technologies, and 62 per cent will implement them by 2018. Analyst firms such as Gartner and IDC predict that 2018 will see a sharp rise in the number of organizations deploying AI technologies. This is not to say that businesses should rush into a complex and confusing market. However, organizations should start experimenting with small scale applications now considering the aspects discussed so that they can ramp up with these emerging technologies when they become the new norm.
(The author is the Chairman of Cambridge Technology Enterprises)