Learning with AI without the Brain Drain
The 20-Hour Edge
Let’s face it: Chatbots and LLMs can speed up extracting and synthesising information a lot. It’s like when Wikipedia just came out and was widely discouraged as an authoritative source for research. Today, we are slowing coming to terms with what AI can do for us, and we are finding out what it’s good for and what it’s not.
There’s a problem however when using LLMs. While LLMs can accelerate some tasks, they introduce a hidden cognitive cost. People who rely on LLMs show weaker neural connectivity in language and executive networks, and perform worse on recall tests compared to people who do the work themselves. If you use AI to do the thinking, you lose the knowledge. Research by MIT shows a 40% drop in retention for chatbot users (MIT Media Lab, 2025).
The brain prioritises energy conservation (often referred to as “use it or lose it”). When an external tool (AI) performs the synthesis, the brain offloads the cognitive effort (cognitive offloading). Since the brain did not perform the heavy lifting of neural encoding, the memory trace is weak or non-existent.
When the tool does the thinking, you do not learn. You are relegating yourself to a passive role, where you have to check a lot of text and lose the skill to create it. On the flip side, metacognitive prompts increase activity in executive control regions, so it’s important how you use AI. Cognitive scaffolding is fine, offloading is not.
A particularly interesting use of AI is skill or knowledge acquisition. As Josh Kaufman argued and showed in his book, you can master the basics of a new skill in 20 hours by following the Pareto rule of 80/20. Let’s see how you can use the AI as a coach instead of a generator.
We can use a four-part framework to accelerate this:
Deconstruct the Skill: Ask the AI to find the 20% of sub-skills that provide 80% of the results. Ignore the rest.
Tighten the Loop: Use the AI as a sparring partner. It should give you hints and error logs rather than full solutions.
Remove Admin Work: Let the AI handle the schedule and resource finding. Keep your energy for the hard mental work.
Simulate Reality: Run high-pressure role-plays. You must pull answers from memory under pressure to make them stick.
The rule is simple. If the AI is doing the thinking, your brain is resting. If your brain is resting, you are not learning.
Use AI to increase your effort on the right things.
Let us know if you are interested in rules of engagement in the age of AI. If you are interested in cognitive engagement, retention, and prompt-based instruction design, please get in touch. We have you covered with specifically tailored courses at Chelsea AI Ventures.