Artificial Intelligence: CEO's Top 5 questions!

If one doesn’t come across some news about AI on a daily basis, then one must be living blissfully on a remote island disconnected from the world. Alas, no such luck unless you are Richard Branson and can go off on your own island and disconnect!

AI has been traversing a lot of bits of the Internet lately. But AI is not new and has been in the making for decades through the dedication of many academics and more recently R&D groups at major technology companies and a number of startups. Of course, as with any exciting technology, there is a good deal of hype with AI as well. So how do CEOs determine what’s in AI for their companies? How can AI meaningfully help transform and grow their businesses?

To answer that, first, let’s deconstruct the definition of AI as offered by Wikipedia.

What the definition implies is that with advances in technologies machines can be made to: “perceive” its environment; adapt decisions based on available data or environmental input; and choose actions that are most likely to be considered a success as humans try to do. In essence mimic “cognitive” functions of humans.

Two key themes are central to the definition. First mimicking humans in perception and decisions and second do that using available data. With these two central themes in mind, I believe CEOs should ask following questions to his / her CXO team. Objective answers to these questions will help determine the best course of action for the company interested in AI.

1.) Does our company / our industry have high-quality data? Or can we get it?

As data is central to AI it is essential that data is high quality. Not just Big Data but Smart Data. Many companies produce terabytes or more worth of data routinely but often it requires tremendous effort to clean it up, to make it consistent across systems and applications, and to make it easily accessible to run AI experiments. If the answer to this question is yes, then great. If it’s a weak yes or no then it is imperative to have a thoughtful effort that addresses the data need.

2.) To start with, what’s the specific domain of our business need we can experiment with AI?

It is important, especially in midst of all the hype, that your team can articulate specific domain where it believes AI can help. Customer Service / Engagement or operational efficiency are some examples domains of company’s business within which specific use cases can be thought of that can benefit from AI. For example, insurance company AI agent can continuously learn from insurance quotes requests, claims processing (filed, accepted, fraudulent, rejected etc.) and other relevant data to recommend most optimum policy coverage to customers while optimizing its own risk with respect to future potential payouts. The resulting system can be made to learn with more and new data that produces better results which in turn leads to better data to feed back to the system.

3.) Who are the 5-7 AI startups that are addressing a domain most relevant to our company / industry and how can we collaborate with them?

More often than not innovations based on emerging or complex technologies (and indeed new technology building blocks) come from startups. If a company doesn’t proactively and with all sincerity engage with startups than it will not benefit from these innovations. In AI space it is important to identify key 5-7 startups that can help in a specific domain. For example, here is a starting point to look at most recent AI centric startups in the field of Healthcare. And here are AI startups in retail and e-commerce space. Source for both is CB Insight.  They have a similar list for other industries as well. If your team isn’t looking at this and planning active exploration with appropriate Startup, then ask why not?

4.) What are specific AI use cases our competition or industry peers exploring?

There are always learnings from closely observing what the competition or adjacent industry peers are doing. This is helpful to determine if you have an opportunity for first movers advantage or be informed fast follower and presumably avoid early adopter missteps. At the very least good answer will show your team is on its toes and informed.

5.) What 2-3 key milestones can the team work towards and in what timeframe?

As with any new technology driven experiments, there needs to be a thoughtful balance between rigid milestones and very open-ended effort. For example, in the insurance company example above first reasonable milestones could be 1.) understanding of what’s high-quality data available and how it can be used 2.) definition of one or two specific uses cases and 3.) what would be a minimum viable product.

I hope this framework / set of questions can help shift through some of the hype and lead to a thoughtful and objective approach to AI for companies across industries.