Google's data: despite the hype, workers aren't automating themselves away

Ars Technica examined Google's data on how people actually use AI tools at work. The conclusion cuts against the alarmist agenda: the overwhelming majority of scenarios are assistance with a task where the human stays at the wheel, not a full handover to the machine. People ask AI to explain, improve, check, draft — and make the decisions themselves.

This matches other research this year: turnkey automation is rare, while AI amplification of humans has gone mainstream. The fear that "employees will train the AI and become redundant" is not supported by the data — value shifts to those who can work in tandem with a model.

For a leader in Kazakhstan this hints at how to frame adoption: not "we'll replace people" but "we'll give everyone an amplifier" — the results come sooner and the resistance is smaller. Start by giving employees a legitimate, company-approved way to ask and verify: at 1Cifer that looks like an agent answering a worker in plain language from their company's data — from stock levels to an order's status — so people spend the freed attention on decisions rather than hunting for numbers.