The fastest worker is not always the one who gives AI the most work. Sometimes the win is a clean handoff. Let the machine sort records, draft options, or check a long file. Sometimes the win is a slow talk. Ask another question. Test the answer. Add context. Learn while you build.
That choice can save a company time and money. It can also protect a person’s energy, skill, and peace. New research suggests this choice, not simple AI use, may be the human edge that matters next.
The latest data breaks the automation myth
This month, OpenAI released GPT-5.6. Its flagship model finished one broad test in 61% less time and at about half the estimated cost of a leading rival. The machine side of work keeps getting faster and cheaper.
But a July 14 study in Frontiers in Artificial Intelligence found something more important. It studied millions of Claude conversations tied to 3,365 job tasks. People who refined answers, checked results, and used AI to learn showed deeper use. The study found two kinds of trust. One is delegation: let AI reason alone. The other is collaboration: build the answer through a back-and-forth talk.
Neither mode won every time. Rich dialogue did not simply lead people to turn on more autonomous reasoning. The best fit changed with the task. The mature user was not the person who used every feature. It was the person who knew which mode the moment needed.
The business gap is a human gear shift
Business leaders can see the future, but most have not built for it. JLL’s new survey of more than 2,200 leaders in 21 countries found that 60% expect AI to reinvent human roles, not replace them. Yet only 15% said their company had reached the stage where it was truly improving AI use.
of leaders expect AI to reinvent roles, but only 15% say their organization is optimizing AI use, according to JLL’s July 14, 2026 survey release.
That is not just a tech gap. It is a role gap. Most jobs were built like one fixed gear. You do the task, start to finish. AI changes the job into a small set of choices: What should I own? What should I hand off? What must I check? What can I learn? What needs a human talk?
Psychology tells us why this matters. People need some control to stay engaged. They need to see that their work helps someone. They need challenge to keep growing. Full automation can save minutes while taking away mastery and meaning. Full human control can protect pride while wasting time. The right mix gives back both output and ownership.
Microsoft’s July Work Trend Index results from Switzerland show the pattern. Eighty-four percent of AI users there treat the output as a starting point, not the final answer. They keep judgment and duty with the person. That is not fear of AI. It is good command.
Return on Individual is the new control panel
This is where ROI must mean Return on Individual. Do not only ask how many hours AI saved. Ask what those hours became. More profit? Better focus? More energy at home? A stronger customer partnership? More passion, peace, and purpose?
Then ask what the person lost. If AI does every hard part, skill can fade. If it does every first draft, original thought can fade. If it watches every move, trust can fade. Cheap output becomes costly when people stop growing or caring.
The answer is not less AI. It is stronger Actual Intelligence. Character keeps the handoff honest. Critical thinking checks the work. Emotional intelligence knows when a person needs a talk, not a bot. Grit stays with the hard part. Elastic intelligence shifts gears as the task changes.
That is Josh’s trillion-dollar EQ bet. AI made information free, so information is not the moat. The moat is the human operating system that turns cheap intelligence into wise action. Companies will not win by removing the individual from work. They will win by raising the return on that individual.
Bottom line: automate the task, not the human. Measure your own operating system with the free Actual Intelligence diagnostic, then measure the team with the free Culture Test at actualintelligenceos.com.