5 Myths About Cognitive
A survey of early adopters helps correct some common misconceptions about artificial intelligence.
CIOs looking to gain support for cognitive initiatives may find themselves battling some persistent myths.
Artificial intelligence (AI) is one of the most frequently discussed topics in business today, but even more than most new technologies, its promise is sometimes obscured by a set of lingering myths—particularly among those whose exposure to the technology has been limited.
Professionals with first-hand experience have a different perspective, according to the 2017 Deloitte State of Cognitive Survey, which is based on interviews with 250 business executives who have already begun adopting and using AI and cognitive technologies. The responses of these early adopters shed considerable light on the current state of cognitive technology in organizations. Along the way, they help dispel five of the most persistent myths.
Myth 1: Cognitive is all about automation
It is rare to find a media report about AI that doesn’t speculate about job losses. Much of the reason for that is the commonly held belief that the technology’s primary purpose is automating human work. But that’s hardly the full story—in fact, there are significant uses for AI that do not involve substituting machine labor for human labor.
A Deloitte analysis of hundreds of AI applications in every industry reveals that these applications tend to fall into three categories: product, process, and insight. Product applications embed cognitive technologies into products or services to help provide a better experience for the end user, whether by enabling “intelligent” behavior or a more natural interface or by automating some of the steps a user normally performs. Process applications use cognitive technology to enhance, scale, or automate business processes, while insight applications use AI such as machine learning and computer vision to analyze data to reveal patterns, make predictions, and guide better decisions. In some cases, these technologies can be used to automate human work, but often they are used to do work that no human could have done otherwise.
Survey respondents clearly believe AI is important for more than just automation. While 92 percent say it is important or very important in their internal businesses processes, 87 percent rank it comparably for the products and services they sell. Cutting jobs through automation falls at the bottom of respondents’ list of potential benefits.
Myth 2: Cognitive kills jobs
Hand in hand with the belief that AI is all about automation is the expectation that it will destroy countless jobs. While it’s impossible to know what will happen in the distant future, both the objectives and the predictions of survey respondents suggest that job loss won’t be a major outcome. Only 7 percent of respondents selected “reduce headcount through automation” as their first choice among nine potential benefits of the technology; just 22 percent chose it among their top three.
When asked about the likelihood of job loss in the near future, respondents were similarly upbeat (Figure 1). Just over half expect that augmentation—smart machines and humans working side by side—will be the most likely scenario three years from now. Only 11 percent expect substantial job displacement; a larger percentage expect job gains or no substantial impact on jobs.
Respondents are more likely to be concerned about substantial job loss in the more distant future—22 percent expect it to happen in 10 years—but even then, a larger proportion (28 percent) expect augmentation to be the most likely outcome. The same percentage anticipate brand-new jobs.
Myth 3: The financial benefits are still remote
Many people view AI as a futuristic technology dominated by a handful of tech giants making headlines with high-profile applications. They believe most companies will not be able to achieve real financial benefits anytime soon. There is some truth to this view: The tech giants are indeed at the forefront of AI R&D and have capabilities not available to everyone. On the other hand, there are ordinary companies in every industry that have deployed AI and reaped financial benefits.
While just 12 percent of survey respondents say their companies have invested $10 million or more in AI so far, a quarter have invested between $5 million and $10 million. Most have committed less than $5 million to date, yet a significant majority—83 percent—say their companies have already achieved either moderate (53 percent) or substantial (30 percent) economic benefits from their AI projects. Only 16 percent of respondents say their company has so far failed to realize any economic benefit.
In fact, the survey suggests that the economic benefits of AI may increase with experience: The more AI deployments respondents report, the higher the percentage who say they have realized economic benefits. Among executives whose companies have deployed 11 or more AI projects, 92 percent report economic gains.
Myth 4: AI is overhyped and bound to disappoint
There is no question AI is one of the hottest technology topics today, but is it overhyped? Many survey respondents don’t think so. Just 9 percent say they think the hype is overblown; 10 percent actually say the technology is underhyped. And over a third—37 percent—say AI is fundamentally different from IT, suggesting they believe it’s distinct and worthy of at least some of the excitement surrounding it.
Myth 5: Cognitive technology is just for ‘moonshots’
It’s not uncommon to hear cognitive technology projects equated with “moonshots”—highly ambitious, transformational change initiatives often considered disruptive to companies and industries. Does this mean smaller projects are not worth pursuing?
A substantial percentage of respondents (47 percent) do agree that “it is important to strive for large-scale, transformational change with cognitive technologies.” More than half, however, believe it is better either to “pick the low-hanging fruit” (40 percent) or wait a few years until the technology matures (12 percent).
Any transformational project is likely to face high risks and expense. Multiple less-ambitious projects—particularly when focused on specific business process goals—can be more likely to succeed and may also yield transformational outcomes collectively.
Of course, the fact that survey respondents don’t agree with these myths doesn’t necessarily mean they won’t ever come true. But the fact that 250 early adopters dispute them certainly casts some doubt. At the very least, it makes sense to keep an open mind about the ultimate impact of AI.
—by Thomas H. Davenport, external senior advisor, Deloitte Consulting LLP, and distinguished professor, Babson College; and David Schatsky, managing director, Deloitte LLP