In 2011, IBM Watson, competed in the game show Jeopardy! against some of its most successful players, and won handily. It was a watershed moment not because the machine beat humans at their own game, but because the possibilities opened our eyes. What followed was a series of striking breakthroughs in AI -- image recognition, speech recognition, and many more -- all possible through a technique known as deep learning.
As we step into a new decade, an entirely new picture of the future of AI emerges. Heading into a new decade, AI looks more like “AI-as-a-service,” embedded into seemingly everything and almost invisible. We are already in the era where we “AI-ify” to augment human capabilities and in some cases introduce autonomous AI to eliminate human tasks altogether. So, what is propelling this massive growth of AI? Three technology breakthroughs act as catalysts:
1. Massively parallel computation: Thinking or reasoning is a massively parallel process of billions of neurons, passing signals to other neurons through networks to deliver the outcome of judgment. Similarly, the graphics processing unit (GPU) unlocks new possibilities where neural networks loosely based on the way neurons work in our brain, can facilitate hundreds of millions of connections between the nodes at almost at a sub-second processing time.
2. Big data: With digitization and proliferation of smart phones, we have access to massive amounts of real-world data. Our ability to collect, clean, standardize, and store this real-world data provides us with an enormous training ground for AI. The result? We are beginning to see intelligence infused into almost everything, consequently transforming everything into a “smart” thing.
3. Better algorithms: Deep learning algorithms help to reason and interpret the data, churning out patterns and recommendations. We can now collect lots of data and apply sophisticated algorithms to arrive at predictions.
Four core roles of AI
Demystifying AI, we can broadly define AI maturity by understanding organization’s strategy for using it in these ways:
- AI as strategic advisors: AI strategists are responsible for understanding problems, strategy development, and defining action/execution plans. Algorithms deliver data-driven insights and recommend actions to be taken. Human intervention is to identify which decisions are deferred to the algorithms and how the decisions are implemented.
- AI as task executioners: AI algorithms are responsible for analyzing process, creating clear task descriptions and goals, and defining detailed service-level agreements and KPIs. Humans responsible for operational task execution are free to perform higher-order tasks such as reviewing the outcomes of algorithms.
- AI as a virtual assistant to employees: AI algorithms are responsible for effectively working together with humans as intelligent agents.
- AI as an autonomous organization: Algorithms are responsible for substantial portion of decision-making process. Companies allow AI full autonomy in steering an organization to new levels of risk, profitability and innovation.
AI applications are data hungry
For effectiveness of AI, we need lots of data and robust data management practices. There’s need to identify data sources, build data pipelines, clean and prepare data, identify potential signals in your data, and measure your results. Organizations that are serious about AI have historically been proficient at acquiring and managing data as a strategic asset. Data-driven, a stepping-stone toward AI, really means that all decisions and actions taken by the enterprise and employees are done by using the most factual and accurate data and there is a well-defined method of applying scientific analysis to this data to arrive at decisions. It is about being conscious about how you are using (or not using) the tools (data and algorithms) at your disposal. You must ask questions and not maintain the status quo; let’s call it the “data-driven” strategy, which is a prerequisite to an effective AI strategy. This pervasiveness of AI triggers another interesting phenomenon -- the more we use it, the smarter it becomes.
An AI-first approach to everything has more implications. Organizations need to assess business and technical landscape and determine the need for AI-led interventions. You don’t need to blindly follow what magic AI has done elsewhere. When doing a lift-and-shift and appling it to your business scenarios as this approach may do more harm. You need your AI applications to be relevant to your business, take advantage of your data, and learn about and improve your past performance. And along the way, if you happen to generate new ideas that result in unique value propositions, new products, and new offerings, it is great.
AI technology is transformational and will require new leadership skills to evangelize within the enterprise. Modern organizations value empowered AI as much as they value empowered people. AI, in many ways, is pervasive and provocative as we begin to outsource everything to algorithms. Automation is truly improving the quality of life, but we should be mindful of tasks to delegate or demarcate between humans and machines. For example, CEOs must make it clear when smart algorithms, rather than human associates, are to be consulted. This can be difficult. Some of the most important decisions regarding machine learning are usually about the extent of authority the AI agents should have.
Sachin Vyas is Vice President of Data, AI, and Automation Solutions at LTI, a leading provider of technology solutions for many of the world’s largest organizations.
Soumendra Mohanty is the author of two books on artificial intelligence and has more than 20 years’ experience in data and analytics.