AI productivity is on the upswing, but some estimates may be over the top

Is AI really delivering what the C-suite is expecting?

Not when it comes to productivity. Meta is instructing employees developing the Metaverse to use AI to “go 5x faster,” even though that expectation is unlikely to line up with the technology’s actual capabilities.

That’s been the promise and reality of AI for years. The technology has been hyped, often beyond its actual capacity, and while it can be helpful, productivity gains of around 500% are unlikely—for now.

Propel Software CEO Ross Meyercord said Meta’s demand of its workers is likely more of a “stretch goal” than a realistic expectation.

“You need to think of it as 5x versus how can you go find me 10%—it fundamentally forces you to think about the problem differently,” Meyercord said. “If I need to achieve 5x productivity improvement, I have to fundamentally blow up the existing process and start from scratch.”

Reassessing. Some organizations are realizing that and taking the opposite approach from Meta. CompTIA Chief Research Officer Tim Herbert told IT Brew that there’s been a broader reality check of AI’s ability to effect change, leading to a resetting of investment.

“At the C-suite level, there are expectations that it will boost productivity and by extension, the bottom line,” Herbert said. “But at the same time, there was also a fairly high incidence of companies backtracking, and they cite a number of reasons for that—everything from performance to execution.”

Few disagree that AI can help, said CyberArk CIO Ariel Pisetsky. But the issue is whether or not it’s a productivity boom—and there’s the question of how long it will take those huge gains to arrive. Pisetsky said he’s confident we’ll get there eventually, but for IT teams it might be on the horizon already.

“Something is going to happen,” Pisetsky said. “Probably sooner than later.”

Bull talk. While there’s a lot of talk about hype cycles and bubbles, Pisetsky believes there’s potential for AI to impact productivity. Perhaps things won’t reach the 5-times productivity Meta is demanding immediately, but the consistent improvement of AI systems is a sign that exponential growth is likely to continue.

“I’m not sure this is hype,” Pisetsky said. “It’s not just regular technology, because it’s advancing in leaps and bounds way beyond what we’ve seen other tech do—things that were very hard for computers to do just a very short time ago are now possible.”

Details matter. That may be true, but it depends on what side of the business you’re on. For example, there’s a difference between large language models and agents, as Meyercord said. That’s because agents allow for chaining processes together in a way that could deliver the big productivity projections.

“The world of agents, on the expectation side, will cause those of us in management roles to continue to set our sights higher and demand 5x or maybe even numbers higher than that,” Meyercord said. “It is the technology now which has the potential to do that, because you’re fundamentally able to have the AI do work on your behalf, and do more and more complex tasks—[and] be able to link those tasks together.”