If you listen around, artificial intelligence seems to be like the eighth wonder of the world. If you fall for this line and happen to pour some cold, hard cash into the technology, you may see your cutting-edge investment vanish into thin air.
Product recommendations in online shops, photo searches on Google and language recognition on smartphones – machine learning is already a permanent part of our lives. It also comes into play in such areas as predictive maintenance, customer support in companies and autonomous driving. Unfortunately, the fascinating possibilities of the technology have also given life to unrealistic expectations.
One such delusion is that machine learning can come up with the “right” answer without even being asked the question. Such visions seemed like tall tales to analysts at the research and advisory firm Gartner. In a paper titled “Hype hurts: Steering clear of dangerous AI myths,” the analysts took on the job of debunking many misconceptions about the technology.
Myth 1: Buy an AI to solve your problems.
Companies do not primarily need an AI. They need solutions for their business. AI technologies may play a role in these solutions. And, don’t forget, THE “artificial intelligence” does not exist outside the world of marketing.
Myth 2: Everyone needs an AI strategy or a chief AI officer.
The truth of the matter is much different: Every company needs a business strategy. Back in the 1990s, nobody was tossing around the idea of introducing a strategy for graphical user interfaces (GUI) and hiring a chief GUI officer even though everybody was talking GUI at the time. But that’s not half of it: Such a sweeping technological revolution as artificial intelligence will brush aside all C-level positions as well.
Myth 3: AI has human characteristics.
Clever people can whip up a combination of highly developed data analytics software, special algorithms and mountains of data to create the illusion of a human counterpart. This boosts the value above those things that the technology can actually deliver. As a result, investments will quickly produce disappointing results and, if worst comes to the worst, cause some serious career setbacks.
Myth 4: AI learns on its own.
Data scientists have to do a huge amount of work to develop AI applications and keep them running. Any company that happens to underestimate the costs related to retraining will be in for some unpleasant surprises.
Myth 5: It’s easy to train applications that combine deep neural networks (DNNs) and natural-language processing (NLP).
The job of training an AI-based “customer assistant” requires massive investments of time and money – during both the launch and operation phases. The AI staffer not only has to grasp the information provided by the customer, but also has to set up a separate workflow for each possible customer need and then continuously manage it. What’s more, no one can predict just how well the system will actually perform in the real world of business.
Myth 6: Maximize investment in leading-edge AI technologies.
This is, without a doubt, the most risky idea. It is much better to follow in the footsteps of leading companies. Gartner advises companies to apply the best-of-breed approach, that is, to use the best individual option in the market and to shy away from a complete solution. Companies should also not forget about “aging” innovations whose own waves of hype have already crested. Systems of experts or more simple forms of machine learning may be able to handle the job at much lower costs and risks.
Myth 7: AI will transform your industry — take the lead!
This rallying cry will apply to only a very small group of companies. Sure, AI can automate all sorts of tasks, and its abilities are indeed quite tempting. But the costs and risks associated with it will certainly keep you up at night as well. The best approach for most companies is: Let other companies blaze the trail and then learn from their experiences. Gartner recommends that companies should initially develop the necessary expertise within a limited department in the business.