Companies face uneven progress as AI promises outpace rollout

Artificial intelligence is moving deeper into corporate planning, but companies are finding that adoption is slower and messier than the forecasts suggest. Gartner predicted that automation will take over 69% of the work managers now do within four years, while a Plutoshift survey found many industrial firms remain far from full AI deployment.

Gartner, the technology research firm, said the shift would require companies to rethink the manager’s job. It also forecast that 47% of learning and development spending could be wasted as AI removes about two-thirds of what Gartner described as task-based learning that happens on the job.

The findings point to pressure on employers to retrain workers while the content of many jobs is changing. Gartner’s forecast also suggests that traditional workplace learning may become less useful if software absorbs routine tasks before employees can practice them.

Plutoshift, a Silicon Valley company that sells data and predictive analytics software to industrial businesses, reported a more difficult picture on the ground. In a survey of 250 industrial firms, the company found that more than 72% said building the data collection processes needed for machine learning took longer than expected.

Only 17% of the companies surveyed by Plutoshift said they had reached full AI implementation. About 70% said they were still reviewing needed resources, weighing possible business uses or running limited pilot projects.

Prateek Joshi, Plutoshift’s founder and chief executive, said many companies working through AI projects lack the technology and data systems needed to make them work. “In the end, these implementations can fail to meet expectations,” Joshi said.

Plutoshift also found that almost 20% of surveyed companies said “peer pressure” pushed them to start AI projects. That figure adds to concerns that some businesses may be buying or building AI tools before they have a clear operational need or the infrastructure to support them.

Questions about AI marketing have also sharpened around facial recognition company Clearview. ZDNet reported that Clearview faces a class-action lawsuit in Illinois accusing it of violating the state’s biometric privacy law, which bars use of residents’ biometric data without consent.

BuzzFeed News reported that the New York Police Department disputed Clearview marketing claims tied to the August 2019 arrest of a man accused of placing rice cookers made to look like improvised explosive devices around New York City. “The NYPD did not use Clearview technology to identify the suspect in the August 16th rice cooker incident,” a department spokesperson told BuzzFeed News.

The NYPD also told BuzzFeed News that it had “no institutional relationship” with Clearview, while Clearview founder Hoan Ton-Thot said the department was testing the company’s technology. Separately, London’s Metropolitan Police said it would begin using a live facial recognition system from NEC Corp. in parts of the city where intelligence suggested wanted suspects were likely to be found; privacy advocates said they would challenge the use in court.

IBM has called for a risk-based approach to AI oversight. In a policy paper, the company said AI regulation should be guided by accountability, transparency, fairness and security, and it urged companies to name an AI ethics lead, assess potential harms, disclose AI use, use explainable systems and test for bias.

This story draws on original reporting from Fortune.