Three Major Observations In AI adoption

Along with the hype of AI and machine learning there is a lot of promise


Business leaders across every organization type talk about using artificial intelligence. CEO’s are claiming that if you don’t have an AI strategy your business will not be around long. AI is even being compared to electricity!

All of this euphoria may seem overwhelming to an organization that is already overwhelmed by the basics of using technology infrastructure and IT. As CEO of an AI startup, I have spent the last year and a half looking at industries that AI is impacting. It is interesting to note that while adoption is definitely low across every industry, those taking advantage of it are winning in a major way. Here are my observations that others may find useful.

AI as an augmentor

First, let's look at an industry that I usually stayed away from because of all the regulatory requirements and the approval processes involved: Healthcare. In every analytics and big data company I have been a part of, healthcare is not the largest revenue generating vertical.

While healthcare accounts for a major portion of the US GDP, it is typically known to be a slow industry in terms of technology adoption. The promise of AI seems to be overcoming this in addition to other regulatory changes. Physicians are the first it seems to grasp the promise of AI. In the US alone there will be a shortage of as much as 100,000 physicians by 2030.

Before people worry about doctors being replaced, AI should be looked on as a way to augment what physicians do. In diagnostics, AI is being touted to improve cancer detection, and now in the new world of value-based healthcare models, using AI to improve efficiencies across the board will become critical. In other industries like e-commerce, retail, etc, AI becomes central to demand planning and replenishment. In manufacturing, you use it for predictive maintenance and in real estate, people are predicting true value of a property and rent. You name the industry, innovators are at work using AI to get a better answer than they have had in the past.

Translating the unstuctured

Unstructured data is the crux of all data being generated. 90% of the world's data is unstructured. Living in the world of nicely formatted and organized table structures is over. The schema needs to account for non-numeric data types that need to be interpreted and then structured. Language understanding is at the forefront of AI technology.

Personal assistant technology like Alexa, Siri, and Cortana are good interfaces that are attracting everyday usage, but understanding unstructured data goes way beyond these devices. 36 of the 40 zettabytes of data we are expected to amass by 2020 will be unstructured. Getting value out of this data will become even more critical as we move forward. Within the unstructured world, some data is harder to deal with than others, language is actually the hardest. While a picture is worth a thousand words, interpreting one word can be more complex than the image itself.

Don't be scared of AI taking your job!

As a kid when I started getting interested in computers, I remember one of my favorite computer users would tell me that computer software is nothing more than a way to put data into or out of a database. So how is AI dramatically different from what the world has been doing in its adoption of technology for the last two decades?

When you run a Machine learning algorithm, you allow a computer to look at a multitude of possible permutations, many more than a human could perform. This is where cognitive AI does its magic, whereas in the past we created these models manually – When dealing with analytics, there are just too many data points to consider. The obvious dip in the line chart as our eyes see it does not factor in the other 20 variables.

The machine, however, is able to keep track of significantly more variables. Hence, why would you not use this power to get better insight than what we can eyeball?

This is an exciting time for AI, data usage, and optimization and they are only gaining value. Industries will take time in adopting, but those who do will enjoy significant gains and advantages. 


Read next:

Why We Need Data Visualization To Understand Unstructured Data