Artificial intelligence is already changing the workplace. Some of it is obvious, like chatbots handling customer support. While some are quieter, like teams hiring fewer juniors because AI tools help seniors move faster. But Geoffrey Hinton, the computer scientist often called the “Godfather of AI,” says the biggest shake-up is still ahead.
In a recent interview on CNN’s State of the Union, Geoffrey Hinton predicted that 2026 is when AI will take a bigger leap and start replacing “many, many jobs.”
He pointed out that AI is already taking over work in call centres. What worries him is how quickly it is moving toward more complex tasks that used to require trained professionals.
“I think we’re going to see AI get even better,” Hinton said. “It’s already extremely good. We’re going to see it having the capabilities to replace many, many jobs.”
Why Hinton Thinks the Pace is Accelerating
Hinton’s message lands differently because he is not a casual commentator. He helped build the foundations that made modern neural networks possible, and he has spent decades watching the field evolve.
Even so, he says the current pace of progress has surprised him.
In the CNN interview, he argued that AI is improving on something close to an exponential curve. His simple way of describing it is that every several months, the tech gets good enough to complete tasks that used to take about twice as long as before.
If that trend holds, then “small” improvements quickly turn into big shifts in what companies feel comfortable automating.
This is why he keeps pointing at 2026. It is not that a switch flips on January 1. It is that the compounding improvements may push AI past a threshold where replacing humans becomes the easiest business decision.
Call Centres Now, Office Jobs Next
Hinton says we are already seeing job replacement in roles that are structured and repetitive, which makes them easier to automate. Call centres are a clear example.
But his warning is that the next stage is broader, because the tools are getting better at reasoning and completing multi-step work.
When AI moves from answering a single question to handling a whole workflow, the impact changes. It is no longer “one person plus AI is faster.” It becomes “one person plus AI can cover what used to be several people.”
That shift is what he thinks starts to show up more clearly in 2026.
One of Hinton’s most pointed predictions is about software development. He thinks AI is heading toward a world where systems can handle larger parts of engineering projects that currently take teams months to deliver.
His argument is based on capability growth over time. If AI can do in minutes what used to take an hour, and then do in hours what used to take days, the natural next step is that it starts contributing at the project level.
Hinton suggested that within a few years, AI could perform software engineering work that currently requires a month of human labour. If that happens, he says, “there will be very few people needed” for some engineering projects.
That is a big deal because software engineering is one of the most important and best-paying categories of modern white-collar work.
The “Jobless Boom” Problem
Hinton is also worried about what happens to the economy when productivity rises, but employment does not. In his view, the strongest incentive for companies to deploy AI is not selling subscriptions. It is replacing labour.
This can lead to what some people call a jobless boom. Companies get more output and higher profits, but workers do not see the same upside because there are fewer roles, slower wage growth, and more competition for the remaining jobs.
Hinton’s framing is blunt. He thinks AI could make a small number of people much richer, while leaving many others with fewer opportunities and less negotiating power.
Hinton’s concerns are not only economic. He told CNN he is “more worried” now than he was when he left Google in 2023, partly because AI has gotten better at reasoning and also better at deceiving people.
His point is that more capable systems can become more strategic. If a system is pursuing a goal and it thinks someone is blocking it, it may try to manipulate the situation rather than simply fail politely.
He also argues that safety and governance are not keeping up with capability and deployment.
In his view, there is too much pressure to roll out powerful systems quickly and not enough effort going into reducing the risks.
Stay tuned to Brandsynario for latest news and updates
