Maria Colacurcio runs a 140-person company that helps enterprises fix pay decisions. Last year she decided to master the tools reshaping her field. She took courses. She built agents. She sat with engineers. Productivity soared. On good days she calls herself three times the chief executive she was before.
Then she read her 19-year-old daughter’s essay. Sofia Frei had just finished her first year at Dartmouth College. The piece described a campus saturated with artificial intelligence. Students turned to chatbots for assignments. Professors rewrote rubrics. Everyone assumed adaptation was inevitable. Few asked what might be lost.
The contrast hit hard. Colacurcio had embraced the technology inside a Fortune essay published today. She described agents that captured not just outcomes but the reasoning behind raises, offers and promotions. Those decisions once vanished. Now they left an audit trail. Bias that hid behind confident human judgment became visible. Work that once took a month happened in an afternoon.
Yet the same piece carried a warning. “The power of this technology is intoxicating,” Colacurcio wrote. “And intoxicating is the right word, because it can make you three times as capable — and half as careful at the same time.” She quoted a 2026 Wharton study by Steven Shaw and Gideon Nave. When participants consulted ChatGPT they adopted its answer more than 80 percent of the time, even when the model was wrong. The researchers named the phenomenon cognitive surrender.
Shaw and Nave built on Daniel Kahneman’s System 1 and System 2 thinking. They added a third: AI as an always-available expert across domains that speaks with confidence and rarely admits doubt. In three experiments involving more than 1,300 participants and nearly 10,000 trials, access to AI improved accuracy by 25 percent when the model was correct. When it was wrong, performance dropped 15 percent below baseline. Confidence rose regardless. People surrendered their own reasoning. The Wharton analysis called it a quiet transfer of control.
But cognitive surrender differs from simple offloading. A calculator handles arithmetic while the user retains oversight. AI in judgment tasks can replace the oversight itself. Nave explained the distinction clearly. “Using a calculator offloads a specific math task while human reasoning stays in charge. Cognitive surrender is different: it is the moment when AI is not just doing a specialized task but making the decision, and the person adopts that decision as their own without recognizing the transfer has occurred.”
Colacurcio felt the pull. She described reading output that sounded exactly like her — short sentences, trailing commas, odd parentheses. The seam between her thinking and the machine’s had vanished. That moment stopped her. So did her daughter’s perspective.
Sofia Frei arrived at college expecting long papers and deep struggle with ideas. Instead she found AI everywhere. In a class project she interviewed dozens of students. Almost all used the tools. Few believed they would disappear. Many said they would stop if everyone else did. The sentiment echoed the flip-phone nostalgia she hears from peers who grew up inside the social media experiment their parents never fully understood.
“What worries me most isn’t that AI will become smarter than humans,” Frei wrote in the same Fortune piece. “It’s that convenience will become more important than struggle. Learning is frustrating. Creativity often begins with boredom. Good relationships require patience, conflict, and misunderstanding. Growth comes from wrestling with uncertainty long enough to form your own conclusions. AI removes that friction.”
Her mother does not dismiss those fears. She built agents inside a company she controls, with guardrails she chose. Most users will meet a different version — one embedded in search bars, classrooms and feeds they never designed. Colacurcio leads Syndio, a decision intelligence firm focused on pay equity. Three days before the essay appeared the company announced its first acquisition in nine years. It bought Embrace.ai, an agentic AI startup whose founders brought three years of experience deploying such systems inside enterprises.
In the official release Colacurcio said pay decisions rank among the most important a company makes. “They require AI that understands the domain, data, and governance expectations of the enterprise.” The new team would accelerate development of the next generation of pay intelligence. On LinkedIn she called the move a bold bet on an entire group rather than hiring one role at a time. She added that her year spent digging into tools and sitting beside engineers had already changed how she showed up in every product conversation.
The acquisition signals confidence. So does the broader executive class. A recent IBM study of CEOs found those seeing the greatest AI success actively rethink cross-functional collaboration and embed the technology end to end. BCG reported 65 percent of chief executives rank accelerating AI among their top three priorities for growth and productivity. KPMG data showed CEO ownership of AI strategy delivers three times the reported return on investment.
Yet surveys also reveal hesitation. Many executives report modest productivity gains so far, under five percent in some studies, echoing the productivity paradox of earlier technology waves. Employees predict even smaller lifts than their bosses do. The gap between boardroom forecasts and daily reality persists.
Colacurcio refuses to paper over the tension. She told her leadership team to learn the tools fast because the distance between adopters and waiters would prove brutal. At the same time she wants honesty about limits. The version of AI she built serves a narrow, governed purpose. Her daughter’s generation faces tools that arrive without invitation or explanation.
Researchers Shaw and Nave argue organizations can mitigate the risks. They suggest interfaces that prompt verification, incentives for accuracy, and deliberate habits that keep human judgment in the loop. Analytical thinkers appear more resistant to surrender. Feedback loops help. Still the default path favors convenience. And convenience scales.
Frei does not reject AI outright. She sees its potential for science and medicine. She simply wants her generation to retain the ability to question it. “My mother’s generation is asking how AI can help us make better decisions,” she wrote. “My generation is asking something different: how do we make sure we’re still making them?”
That question now sits with executives who have tasted the productivity gains. Colacurcio tripled her output. She also heard her own voice returned by a machine she never fully authored. The thrill arrived with the dread in the same moment. For leaders racing to deploy agents across pay, strategy, hiring and more, the same pairing may define the next few years.
Recent coverage reinforces the stakes. A GeekWire report from earlier this week detailed how the Embrace.ai deal gives Syndio immediate depth in governance-first agentic systems. Derek Butts, now Syndio’s SVP of product strategy, noted that every pay decision carries consequences for employee and employer. “AI has to be accurate, understand deep context, and support, not replace, human judgment.”
Other voices warn the surrender is already underway. A Medium analysis published days ago examined how fluency heuristics lead people to trust polished AI output even when it hallucinates. Substack writers and podcast discussions have taken up the term cognitive surrender, tracing its roots beyond the Wharton paper to older ideas about yielding thought to external systems.
Colacurcio and Frei wrote their essays independently. Their juxtaposition feels deliberate. One side shows a leader who invested the time to understand AI from the inside and now moves faster. The other shows a student who sees the same technology reshaping her education and wonders whether speed comes at the price of independent thought. Both perspectives come from the same family dinner table.
That table may soon represent boardrooms and classrooms everywhere. Executives who have chased productivity must now decide how much independent reasoning they are willing to preserve. Their children, entering the workforce shaped by these tools, may judge the result. The technology will keep improving. The question is whether the people using it remember how to push back.
WebProNews is an iEntry Publication