“If Claude doesn’t come up soon, this project is toast, and so might be this contract.”
Anthropic’s Claude was down for around 3 hours on a Monday morning, and many users weren’t just disappointed; they were frozen. The cracks of efficiency began to widen as time dragged on, and many with no contingency in place were just…unable to move.
Technology is always going to have points of failure. But with AI, it’s a bit different. It isn’t just a gap in the availability of the technology; it’s an actual loss of skills, and that is something many organizations need to sit with in the push for adoption.
Cognitive offloading is a real phenomenon, and emerging studies are showing just how impactful it is. It isn’t just that you use AI, it’s how you are using AI.
Studies are converging on the realities of cognitive offloading: Anthropic released a study that found that software engineers who used AI coding assistance scored 17% lower on concept mastery than those who coded by hand. Productivity went up, and understanding went down.
A study in the journal Societies suggests that frequent reliance on artificial intelligence tools may negatively affect critical thinking skills. People who used AI tools more frequently demonstrated weaker critical thinking abilities, largely due to cognitive offloading.
MIT Media Lab study reported that “excessive reliance on AI-driven solutions ”may contribute” to “cognitive atrophy” and shrinking of critical thinking abilities.
Cornell's Metacognition Lab tracked 670 students over four years and found AI-assisted cohorts showed 27% fewer original solutions to novel problems and 44% reduced tolerance for ambiguity
As organizations, we want to be efficient, and we want the highest ROI, but when the workforce trades functional skills for technological skills, that creates a new risk proposition that isn’t easy to mitigate. As the use of AI increases, we must retain core learning and development strategies so that a three-hour outage doesn't become a three-hour standstill.