The gap between the hype and the reality of artificial intelligence is putting intense pressure on executives and employees, leading to unrealistic expectations in the short term and increased burnout.
While 96% of c-suite execs expect AI tools to increase productivity, 77% of employees say these tools have actually decreased their productivity and added to their workload, according to new survey data published today by Upwork’s research institute.
That’s not an unexpected result, for anyone who’s studied the J-curve of technology adoption: there’s an investment required up front in learning, experimentation, and support that results in a downturn in productivity before a longer-term upswing. (The initial dip in productivity followed by the eventual upswing gives the “J” shape to the curve.)
Unfortunately, that reality is considerably different from expectations of many leaders who are being pressured to “do more with less” while also hearing near-relentless hype about the potential of AI to transform organizations. That contributes to the hasty conclusion that “The AI Revolution Is Already Losing Steam,” as a May Wall Street Journal column put it.
But the playbook is pretty clear for how to approach AI in ways that improve business performance and don’t increase executive or worker burnout. Too many organizations are increasing short-term expectations on workers, without making the investments that are required for any successful change to occur. Successful programs focus on potential, not productivity: experimentation and iteration with the tools at the team level that allow people to make work better. That requires investments of time and energy, training, and support—investments that executives need to be realistic about, including frank conversations at the board level.
“You have to take some of your best people out of the business and help them focus their time on building models for your business that create those impacts. And there is certainly a tax to doing that,” says Dan Priest, chief AI officer at PwC. “But, you get on the other side of it and you see the upside of the J-curve kicking in. It’s pretty significant.” Priest says that in some areas of its own business, PwC has seen 20% to 30% efficiency gains thanks to AI investments.
The new Upwork survey of over 1,250 c-suite executives and 1,250 non-executive employees provides a stark picture of the pressures and expectations around AI:
- 81% of leaders say they’ve increased demands on workers in the past year.
- 85% of companies are either mandating or encouraging the use of AI tools to improve productivity.
- 47% of employees who are using AI don’t know how to use it to be more productive.
- 77% say the tools have thus far decreased productivity.
The result? Some 71% of full-time employees report being burned out and 65% are struggling to keep up with increased demands for productivity. That’s undoubtedly impacted by the fact that 74% of employees think their organization’s approach to measuring productivity needs an overhaul.
Other recent research from firms including BCG and Slack points to similar concerns. Slack found that 32% of workers overall have experimented with the tools, and just 16% use them regularly. In a study of software engineers, BCG found that among the 30% of firms that have formally adopted AI coding tools for developers, 76% say that less than half of developers are using them.
What’s holding this back are the investments people need from leadership to make effective use of the tools. I’ve heard from more than one company that they know they need to pause to develop a more detailed plan for how AI tools can contribute to business performance, but they’re too busy with delivering quarterly results to stop and develop one.
What’s the strategy and where and how are we investing?
BCG’s research found that less than half (47%) of firms had a plan for what they would do with increased efficiency from using AI. Upwork found similar results, in that only 13% of firms had a well-implemented strategy for AI. That lack of foresight undoubtedly impacts the odds of AI tools contributing to an organization’s top-level productivity in any meaningful way. Adding to the complexity, Nickle LaMoreaux, IBM’s chief human resources officer, has noted that the benefits of AI at this stage are “fractional,” impacting a small portion of an individual worker’s tasks.
If AI saves a worker an hour, do they know what to do with that hour that will contribute to the company’s performance? Determining that is “a very intentional piece of work that companies have to do by job role,” LaMoreaux told Charter earlier this year. “You need to make sure you direct where that time is now going.”
The investment in direction requires clarity in goals up and down the organization. Startups that I’ve worked with don’t face the issues around lack of clarity of what to do with found fractions of time: important work to be done far exceeds available time and staffing. Teams that have clear roadmaps and priorities allow leaders to point toward opportunities for growth, not just spur worries that efficiencies will lead to job cuts.
Invest in training at the team as well as individual level.
Slack’s research shows that those who are given training and time to learn are 19 times more likely to report that AI is improving their productivity. Training is essential, as only 15% to 17% of employees feel they have the skills needed, compared with 37% of executives who think their employees do have the skills. Research done in Denmark suggests that training is especially important for women to feel confident in using AI and reduce the current AI gender gap.
Yet only 26% of organizations offer AI training, according to Upwork’s research. As BCG points out, the needs of individuals are highly variable. Engineers, for example, require different AI training and support based on their experience level, the languages they speak and code in, and the applications they use.
One of the most effective programs I’ve seen to date is run by AlmostTechnical, an offshoot of Women Defining AI, that puts together cohorts of people who do daily hands-on exercises in bite-sized chunks and, more importantly, talk about what they’re learning and where they’re stuck.
Done within teams, it’s a huge opportunity to help people work together, build trust, and think about how they can leverage technology in ways that drive better outcomes for them and their team. A recent Charter research report documents other ways that organizations have been effective in introducing AI into their workplaces, such as hackathons and dedicated Slack channels.
Focus on reducing toil.
One of my favorite pieces of research in the last year comes from BCG’s Debbie Lovich who shows that the toil of work—the “administrivia,” the paperwork—is what drags us down and leads to lower worker engagement and higher turnover. Too often, the focus of AI efforts is on aspects of an individual’s core, creative work instead of using those tools to improve the systems and workflows around them.
Teams are actually great at themselves identifying what makes work harder: the repetitive tasks, the paperwork, and processes that are cumbersome. Supported by modest internal tech resources, they’re best positioned to co-design and then leverage toil-reducing tools.
For example, Slack’s Sales team built a tool that automated the creation of customer presentations, the task of pulling data that used to take hours trimmed to minutes. BCG worked with administrative assistants to create an AI calendaring tool that reduced the laborious task of trying to triage schedules. Some 79% of admins said the tool made their jobs more enjoyable—and 86% said it made them more effective.
All of this takes investments—it doesn’t come for free. Ongoing improvements will make the gains easier to come by. But in the meantime, as in every past technology revolution, some firms will feel burnt and walk away. Then there will be organizations that pay the price early, invest in their people first alongside the technology and thrive long term—and watch their competitors in the rear-view mirror a few years down the road.
Charter’s Kevin J. Delaney contributed reporting to this piece.