Many companies have seen success when implementing artificial intelligence. Data and consulting firm Kantar, for one, recently got the green light to further integrate AI into several new services the company said will help bands, advertising agencies and media companies more effectively build and measure campaigns.
Not all companies have seen that success. A new whitepaper shows 85% of artificial intelligence projects fail, and 77% of top tech leaders surveyed cite say that barriers to entry are pushback from senior management, failure to impress CIOs, and not seeing the value in the project. Management often doesn’t want to make the investment, according to a study.
The study was commissioned by Pactera Technologies, a global technology company, which worked with Nimdzi Insights, a market research and consulting firm to determine the impact of AI on company strategies.
Many times the projects look good on paper, but when it comes to actually adopting the strategy and providing results, such as the return on investment, the projects fall short of delivering meaningful results.
The whitepaper cites two specific examples. IBM’s Watson AI Health used for oncology was canceled after doctors were not impressed and the company spent $62 million. Some claimed it gave the wrong recommendations for cancer treatment, according to the whitepaper.
The other example points to Uber after a self-driving car killed a cyclist on the road in Arizona. Multiple sensors on the vehicle didn’t pick up a human ahead, and the driver wasn’t fully attentive to the road. The car never made an effort to slow down.
Some major problems start with the lack of focus on business goals and lack of quality data, according to the whitepaper. AI projects often start by collecting all the available data and then looking for a way to use it. At the launch of the project there typically is too much general data from public sources, from parsing websites on the internet, to collecting data from the company.
Some of the success factors include the ability to communicate and set expectations, strong focus on realistic goals and use case, multilingual data equality and commitment to maintenance, and outsource where the company lacks expertise.