Riikka Tanner // May 21 2018
Should You Have an AI Strategy?
A little less than a year ago I was giving a speech for public sector IT management and CIOs, telling them how AI is gaining traction and becoming inevitable force to be reckoned with. Having said that, I assured the audience that the bigger implications would still be far off but that they would be wise starting to familiarize themselves with the technology and taking steps into the world infused with artificial intelligence and intelligent automation.
Well, truth be told, I knew better. I was just trying to be considerate towards the audience and break the message gently. But today, as I am reading from the new PAC report based on a survey that was conducted earlier this year, that only 11 percent of companies in Europe have an AI strategy in place (17 percent in Finland, respectively) and that only 25 percent of those companies consider AI as strategically important, I can’t make up my mind if this is ludicrous or brilliant.
Let me explain.
Yes, I think it is well advised for a company of a decent size to have an AI strategy. No, I don’t think any business should have an AI strategy per se. The biggest pitfall of implementing AI is to have the technology implementation as an endpoint itself. As a matter of fact, AI should be reviewed and considered just as any other technology out there, put in a business context to help your organization succeed.
In practice it means that first must come business problem, challenge or desire and only then, if plausible, comes AI enhanced technology. We are not investing in AI, instead we are investing in enhanced productivity, raised throughput, improved predictions, outcomes, accuracy or optimizing processes by increasing speed, quality, flexibility and yes, hopefully in the future also enabling better service or discovery of novel solutions and possibilities.
AI is the new lawn mower
So effectively, AI is an enabler. The catch is in the magnitude of this enabling technology. One reputable AI/ML expert, Mr. SK Reddy offers a nice explanation for this in his analogy between AI and a lawn mower. Lawn mower was invented in year 1830 and designated to cut grass of large gardens and sports grounds and thus to reduce manual labor needed – an invention which in itself provided a very real solution to a very real challenge. But little did those people back then know that thanks to the lawn mower, what we have today (as a byproduct), is sports industry worth 630 billion dollars each year.
Today, AI is primarily a tool for process optimization. Before taking AI into strategic level explorations, first focus on understanding what AI can and cannot do today. Start thinking possible use cases and scenarios in your business where you would benefit from AI, you will find there are already many different areas of the business ranging from sales & marketing to finance & accounting, from IT to supply chain, all the way to production, that have already identified several use cases for AI. Taking customer service as an example, Gartner analysts predict that, by 2022, at least 40 percent of employees in customer service or public administration roles will be using AI -powered virtual agents in their daily work for decision-making or process related support.
Targeted tactical planning
The PAC report highlights pretty nicely which use cases in each industry are showing the greatest potential with vast majority of respondents in consensus. This information is both good and bad, depending on where you currently stand with your AI initiatives. These use cases represent in essence the next industry standards. Needless to say, this also means that those companies not tailing their competition yet or taking action on these areas, are seriously putting their future at risk.
Machine learning is getting commoditized a lot faster than expected although the development has been hindered by lack of competent specialists. My humble guess is that machine learning will do for regular enterprise tools the exact same thing cloud did for on-premise solutions. In just a few years from now, there will be little or no software solutions that will not employ some sort of AI technology – ranging from fairly simple machine learning algorithms to deep learning solutions for image processing or natural language processing tools that analyze text or conversational tools for speech recognition. And this is just the beginning.
So how about that answer to the question if you need an AI strategy?
You tell me.
Photo by Daniel Watson