Let's start with the premise that change is hard for everyone. In large organizations, change is even harder. Over the past 15 years, I've watched large organizations try to embrace mobile, big data, cloud, and general digital transformation, only to see many of them struggle to implement these technologies time and time again. Today, it's AI that is forcing change on companies and their employees, whether they like it or not.
Part of the problem is technical debt, the notion that an organization's tech stack needs to evolve to make the most of new technologies, rather than using a set of technical capabilities designed for an earlier era. It's not easy to try to change something fundamental to how a business runs without risking ruining what already works. Not many managers are willing to fully embrace such change. Substantial change comes with big risks and big possibilities.
Another aspect of the problem is organizational inertia. It's just hard to change the way people do things. Let me tell you a story from years ago when I was a technical writer. We were implementing a computer system for a small town's register of deeds. The town register was on paper, stored in cabinets. It was manual and cumbersome, and tracking the register could take weeks because people had to manually dig through piles of paper.
The computer system was clearly superior, but the front desk clerks who dealt with the public were not convinced. Part of their job was to rubber stamp completed documents before they were sent off for filing, and they were very happy to do it. To these clerks, who had been working behind the counter for 20 or 30 years, the rubber stamp represented their identity and their sense of power. They didn't want to give it up.
In the end, the system architects simply caved and let them keep the stamps, and they accepted the change, even though the online system no longer required them.
Which brings us to the biggest problem of all: change management. The hardest part about adopting any new technology isn't researching, buying, testing, and deploying it. It's getting people to use it. And, more often than not, you have to get people to keep using it. Failing to do so can undermine even the best intentions of the team implementing the solution.
If we think about all this, and also consider the level of change that AI will bring, we see that a much more fundamental adjustment in the way we work is on the horizon. The stamp holders are feeling their power slipping away, and we need to be careful not to alienate them, or we risk wasting our money.
At the end of the day, organizations are people, and people are complex, so you need to look beyond technology to the end goal of implementing new software that can transform your business.
AI is a whole new way of working
Large-scale technological change within organizations is not new — the arrival of the PC in the 1980s and the rise of spreadsheets and word processors were one such moment, as was the Internet and the World Wide Web — but AI has the potential to be even bigger than these past waves of change.
“The internet era lowered the cost of information delivery, and CIOs seized the opportunity to bring digital technologies into their organizations. But AI is a distinctly different type of technology because it lowers the cost of expertise,” Karim Lakhani, dean of Harvard University's Lab for Digital Data Design, told TechCrunch.
Organizational change is difficult and requires top-down approval.
Image credit: andrewgenn / Getty Images
Box CEO Aaron Levy went a step further, saying that this is the first time that computers will do the tasks people previously did, rather than helping them do them more efficiently. “It's a new relationship with computers because the computers are making decisions. They're evaluating information. They're processing data just like humans do,” Levy said, saying companies need to rethink the role of computing in their organizations.
“As a result of AI being able to run inside the enterprise, a whole new set of frameworks and paradigms have to evolve,” he said. That means they need to start thinking about how the technology will impact their entire organization, looking at issues like accuracy of answers, data leakage, and the data used to train models.
Of course, Levy believes his company's platform was built to address these issues and help customers solve them, but companies tend to work with multiple vendors telling the same story, making it difficult to find one that can truly help and add value.
Is this working?
One of the big problems facing organizations is discerning whether generative AI truly delivers on its promise of productivity gains. Currently, there is no good way to directly link GenAI capabilities to productivity gains, which makes it a hard sell internally to skeptical employees who may be concerned about their future when it comes to adopting AI.
Meanwhile, some employees will demand these new tools, and that tension could create further organizational stress as managers try to implement AI across the company amid differing views on how AI will impact their jobs.
Some, like Altimeter Capital partner Jamin Ball, have written that the technology is so transformative that companies must leap forward whether they see immediate benefits or not. “The world is evolving right now. AI is a massive platform shift. Not embracing AI or not investing in it means risking losing market share and slowly becoming irrelevant,” Ball wrote in his July Clouded Judgement newsletter.
Gartner analyst Rita Salam says that looking back to the days of the first word processors, the value proposition was never about cutting costs by eliminating secretarials — it was about helping to create new ways of working, and AI brings a similar value proposition.
“Reducing secretarial staff probably wasn't worth the cost, but when you think about removing the physical constraints of idea generation and writing ideas down, iterating on them, and making them available to everyone in the organization, I think it's opened up an era of potential, though unproven, innovation and the ability for people to organize their thoughts in a completely different way,” she said. These changes are hard to measure, but the benefits are huge nonetheless.
Getting executive buy-in has always been a critical piece of the digital transformation puzzle. Like the PC before it, the cloud has transformed the way companies do business.
Lakhani said AI is different from the cloud because CEOs can use it to make it work for them. It doesn't take a technical explanation to understand the power of AI, and it can help drive change within organizations. “My sense is that AI is different from the cloud because CEOs, executives, and people who influence corporate strategy in Davos now have access to AI tools and can see some of their problems being solved by AI, and I think that's accelerating the conversation,” he said.
But it's not enough for vendors to just pour money into organizations and sell them a solution — they have to find a way to show value. “Hyperscalers and vendors need to do a better job of showing how organizations can actually adopt these things,” he said.
But overcoming the people problem will be an even bigger hurdle. Lakhani says there are three truisms that organizations can follow as they tackle this challenge. First, “machines will never replace humans, but humans with machines will replace humans without machines.” Second, “AI will fail at the front lines unless you have a top-down mandate for change and create incentives for the 'stamp makers' to really adopt it and feel good about what they're doing.” Trying to force it will fail, he says, so you need to define how and why you're changing for everyone and avoid a “because I said so” approach.
No one said this would be easy. Organizations are at different levels of maturity and technology readiness. But people are people, and substantive change doesn't happen easily in large enterprises. AI will test organizations' flexibility more than any technology has before. It's no exaggeration to say that some companies will live or die on how well they handle AI.