Fearful of disruption and looking for a competitive advantage, boardrooms and CXOs alike have rushed to artificial intelligence (AI) as the next big thing, but before adding intelligent agents, business rules management, machine learning, cognitive systems, or any of the other monikers applied to AI processes in recent years to their own operations, there are seven factors executives should consider, Silicon Valley analyst Ray Wang says.
“The rush to exponential technologies in new business models has placed AI in the forefront of boardroom priorities for 2018,” Wang writes in “Demystifying ArtificialIntelligence: Everything You Wanted to Know About AI but Were Afraid to Ask,” a report published late last year.
“As leaders move beyond the AI hype, the journey toward AI requires both a business mindset and the institutional fortitude to invest in the building blocks for success.”
“With the goal of precision decisions, successful AI projects require more than just great algorithms or access to data scientists. The shift to the cloud has enabled access to compute power, storage of massive data, and the ability to democratize decisions at scale.”
While Constellation expects the AI market – and subsets including machine learning, deep learning, natural language processing, and cognitive computing – to be worth $100 billion USD by 2025, the companies that reap the most value will the ones that possessed, or at least paid close attention to, the following elements:
- Large library of data: “The battle for large data sets has nothing to do with having more data,” Wang writes. “The goal is to build the largest graph that maps the connections to data. More data should improve the precision of insights and allow for more patterns to emerge.”
- Massive computing power: The winners in the coming AI wars – and here he admits a handful of companies, including Amazon, Apple, Facebook, Google, and Microsoft, have an undeniable competitive advantage – will have access to or own cheap computing power. “The ultimate metric for AI rests in pricing not by just compute power but potentially cost per kilowatt hour,” Wang writes. “So the cheapest rate of compute power may determine the cost structure for AI smart services.”
- Compression of time: There is no substitute for time, Wang writes, with AI’s early adopters possessing the advantage of algorithms which have had time to improve. “Data set gathering requires time for better precision,” Wang writes. “More interactions in the network depend on time.” Companies that exponentially reducing the time needed to solve a problem, he says, gain a massive competitive advantage.
- Awesome math talent: Throughout his report, Wang emphasizes that despite the gains in artificial intelligence programming since its infancy in the 1960s, artificial intelligence is still enabled by humans – and both the creation of new algorithms and the ability to apply human intuition to computation requires great math talent. “Algorithms are only as good as the math talent,” he writes. “Success will require the hiring of digital artisans—those who can balance right-brain and left-brain expertise.”
- Industry-specific and domain expertise: As an ever-widening pool of industries, from accounting to advertising, incorporate AI into their operations, paying attention to the user experience will emerge as a key differentiator, Wang writes. “The more advanced and specialized the AI system, the more relevance to the end users,” he writes.
- Natural user interfaces and user experiences: With users expecting to interact with AI in ever-more human ways, businesses should expect to develop their AI systems to accommodate them, Wang writes. “From sensory capabilities to visualization, voice and gesture, the interfaces will improve in human-like and natural capabilities,” he says.
- Contextual decisions: The goal of any company when implementing AI should be the production of more precise decisions, Wang writes. “AI systems augment humanity,” he says. “The recommendation engines that emerge will enable choices, accelerate decision making and ultimately provide filters that deliver situational awareness.”
“The path to full artificial intelligence requires many phases,” Wang writes. “The good news: many organizations have made investments in big data and predictive analytics projects. The foundation of ingesting large quantities of structured and unstructured data for the use of solving business problems provides a solid start.”