Do you ever worry that someone – or some – is pulling the strings on the tech giants and investors today? Perhaps an intelligent machine engineering its own world domination?

The money pouring into development of artificial intelligence might suggest such a plot. Gartner research indicates that revenue from data science and machine-learning platforms grew to $2.4 billion in 2016 – growth that continues to be driven today by organizations’ desire to use advanced analytics to improve decision-making. IDC estimates that spending on AI and machine learning will grow from $12 billion in 2017 to more than $57.6 billion in 2021.

Clearly, something is afoot. The question is: should you care?

Nearly every technology vendor out there is touting an AI solution. Marketers claim AI can do everything from approve loans to diagnose disease, improve email marketing and prevent crime. Though the AI hype hasn’t even peaked yet – and many companies feel they must adopt it in their own environments somewhere – the reality is that commercial applications on the market today vary widely in their ability to deliver and are useful only in narrow contexts. In Gartner’s 2018 CIO Survey, only four percent of respondents indicated they had invested in and implemented an AI solution in their operations. If AI solutions have such great promise, why have they not reached broader adoption?

Sure, as consumers, we’re quick to fall in love with the stuff AI-powered apps claim they can do for us. But, for most organizations, the business value of the AI offerings on the market today is still an open question. Even the most well-known AI solutions struggle to meet the mark. Alexa can call up my favorite songs, but it will likely take a few tries; and the messages Siri completes to my wife are sensible only a fraction of the time. For an enterprise to find AI relevant or useful, its leaders must first find a real sense of AI purpose – and then a gainful business case.

The truth is, most companies lack clarity about where or how they can use AI. Many must first reckon with a dearth of skills required for training the technology or face the daunting task of modernizing their data management platform. Companies that think they can “set it and forget it” – which may be tempting with a solution that seemingly stacks neatly into an IT environment – will not create the outcomes they are after. It’s one thing to be foiled by a $100 smart speaker on your kitchen counter that politely persists it doesn’t understand your question, and quite another to be frustrated with an underperforming enterprise-grade AI that requires a significant monthly investment, especially when that investment – if done right – has the potential to make an order-of-magnitude difference to both top and bottom lines.

The media swirling around AI is dizzying. Here’s what really matters:

  • AI requires hard work. It involves training software to perform tasks based on (lots of) examples rather than cut-and-dry programming. In fact, the training is what makes it “intelligent.” There are no short cuts.
  • AI can liberate insights from big data – but it first needs access to big data (and then it needs lots of training).
  • AI needs to be customized to your business context. Simply downloading and applying open-source software to your data will produce thoroughly disappointing results.
  • AI can accelerate business decision-making by helping close the gap from insights to action, but you must first understand what AI can do and how it fits into your strategy.

It ends up most companies that invest in AI today find it harder than they imagined to get the value they hoped for. Training an AI application so it is relevant and continually smarter is more expensive than they predicted – and ROI numbers are difficult to formulate and then even harder to hit. What’s more, AI – like robotic process automation, natural language processing and other emerging technologies – needs to be orchestrated in concert with other relevant systems. To be able to do its thing – which is, simply put, to be able to map a set of inputs to a set of outputs to detect fraud, say, or identify an outage on a power grid – AI solutions must touch the right data sets, expose outputs to the right networks and interface with humans or automated resources at the right moments. Companies that don’t bother to take the time needed to meticulously integrate their AI solutions for full impact shouldn’t bother with AI at all.

It’s easy to become disoriented by the AI noise. Some have let it seep in and take over their thoughts; others are simply tuning it out. Neither of these impulses will serve you well. Act with cautious optimism but do not act slowly. If you wait, AI will infiltrate via other channels, and then you’ll have no choice but to give a $!@%. 

CAI helps companies assess, pilot and orchestrate AI solutions to address specific business needs. How? First, we ensure AI investments align strategically with the business and then we help companies give AI solutions their due – training them, exposing them and continuously realigning them to address business objectives, even as they change.

Contact us to find out how we can help you realize the value and vision of AI-enabled operations.