From Switchboards to AI: Why the Real Disruption Is Always Slower Than You Think

This week marks the 150th anniversary of the first telephone call — Alexander Graham Bell to Thomas Watson: “Mr. Watson, come here, I want to see you.” It’s a moment worth pausing on, not just as a piece of trivia, but as a reminder that the conversation about technology changing work is as old as technology itself.

A recent piece in the Semafor Technology Briefing highlighted something easy to overlook: automated telephone switching technology was invented in 1892, but human switchboard operators weren’t fully phased out until decades later. By the 1950s, roughly 340,000 Americans were still working those boards, long after the machines were technically capable of doing the job.

Why? Because those operators weren’t just connecting calls. They knew their customers. They passed along messages. They made judgment calls. They filled in the human gaps that pure technology couldn’t. Their value wasn’t in the task itself, it was in everything wrapped around it.

That gap between technical capability and actual replacement is something leaders need to sit with right now, because we’re living it again.

We know AI will displace some jobs and, fingers crossed, create new ones. But the harder question is the lag: how long does the friction between those two cycles actually last? History suggests longer than most organisations have patience for.

There’s also a useful concept worth keeping in mind here: Moravec’s Paradox. AI tends to excel at tasks humans find cognitively difficult, analysis, synthesis, pattern recognition, while still struggling with things we do naturally and intuitively. That means the disruption won’t always land where we expect it to.

Remember when rewriting emails was the go-to beginner use case for AI? The obvious stuff gets automated first. The subtle, judgment-heavy work tends to stick around.

There is more powerful technology available to organisations right now than at any point in history. New models, tools, and platforms are appearing almost weekly. But adoption isn’t really limited by technology anymore.

It’s limited by people, specifically, by how well organisations help their people understand where AI fits in their work, how leaders align around its use, and whether the right systems and guardrails are actually in place.

Technology capability moves fast. Work systems change slowly. That tension isn’t new, and it doesn’t resolve itself.

At Twenty44, this is exactly the problem we focus on: helping organisations get honest about whether their expectations around AI adoption are realistic. In most cases, they aren’t — either the organisation is moving slower than it thinks, or it’s underestimating how much structural change is needed to make the technology actually stick.

The telephone took 60 years to fully reshape the workforce, even after the switch was thrown. That doesn’t mean AI will take that long, but it does mean that capability alone was never the point. The organisations that get ahead won’t just be the ones with the best tools. They’ll be the ones that figured out how to change the way work works.