A vice president rolls out a new AI workflow, and six weeks later it is already outdated because a cheaper, faster model just shipped. This is not a one-time headache. According to new research from MIT Sloan Management Review, it is the new normal, and most companies are still managing it like a temporary crisis instead of a permanent condition.
The instinct to move faster comes from a real cautionary tale. Chegg, the online homework help company, saw its stock price crash after admitting that free chatbots like ChatGPT were pulling students away from its paid service. Within a day of that admission, the company lost a billion dollars in market value. Many executives looked at that and concluded the lesson was simple: move fast or get wiped out.
But that lesson misses something important. Past waves of business disruption, like the shift to mobile phones or cloud computing, eventually settled into a new normal that companies could plan around. AI is different because each new model helps build the next one. The distance between one big leap and the next keeps shrinking instead of growing, so there is no calm period to plan toward.
This has real costs on the ground. Deloitte's newest workforce survey found that as employees lean on AI more, burnout, loneliness, and heavier workloads rise right along with it. AI is supposed to remove drudgery, but in practice it often adds new mental demands faster than it removes old ones. Workers are left trying to keep up with a moving target using tools built for problems that have a clear end.
A few companies are building structures instead of relying on urgency. Microsoft turned an internal AI advisory group into a full department after realizing that scattered guidance created duplicated work and inconsistent decisions across teams. Airtable's chief executive split the product organization into two speeds: one team ships new AI features every week, while a separate team handles the slower, more careful infrastructure work that cannot be rushed. Salesforce built continuous, bite-sized AI training directly into the tools employees already use, rather than relying on a single annual course that goes stale within months.
One detail matters more than it might seem: employees who worry AI will replace them have little reason to get good at it. Aon's chief executive has tied the company's AI push to a promise that it will expand what its workforce can do, not shrink the workforce itself, a promise that lands better because he already led Aon through the pandemic without cutting jobs.
The practical takeaway is not to slow down. It is to stop treating each new AI release as a fire to put out and start building something sturdier: a standing team that tracks and translates AI changes, a second track of work that runs at a calmer pace, and training that lives inside daily work rather than sitting in a separate folder nobody opens. Companies that keep sprinting toward a finish line that does not exist will wear out their best people. The ones that build for the long haul will still be standing when the next model shows up, because there will always be a next model.