Lede: An MIT Media Lab experiment reported by BBC Future found students using ChatGPT for essay writing exhibited up to 55% less brain activity in creativity and information processing regions than control groups relying on unaided cognition or web search (BBC Future, 2024).

Paragraph 1: Kosmyna et al split 54 participants into three cohorts writing on open-ended prompts about loyalty, happiness and daily choices. EEG data indicated the no-tool group displayed widespread activation described as "brain on fire," the search-only group retained strong visual cortex engagement, and the LLM cohort showed sharply reduced activation; post-task recall tests confirmed the ChatGPT group could not quote their own output or claim ownership (BBC Future, 2024; Kosmyna, MIT Media Lab).

Paragraph 2: These results extend the Google effect quantified in Sparrow et al. (Science, 2011), which demonstrated participants were less likely to remember facts they believed would remain accessible via search engines. Parallel findings appear in aviation automation studies: Casner and Schooler (Human Factors, 2015) documented measurable pilot skill decay after prolonged autopilot reliance, illustrating repeated cognitive offloading patterns across domains.

Paragraph 3: BBC coverage notes vulnerability in young users but does not connect the MIT data to Goldman Sachs (2023) projections that AI could impact 300 million full-time jobs or to longitudinal observations of declining critical thinking scores in cohorts with early LLM exposure tracked in subsequent education research (Sparrow et al., 2011; Goldman Sachs Global Economics Paper, 2023).