L’Royal Greenwich Observatorythrough the voice of its director Paddy Rodgers and relayed by the BBCwarns that the growing reliance on instant responses from AI risks eroding our curiosity, critical thinking and capacity for innovation.
This “cognitive externalization” would deprive us of unexpected discoveries that arise from manual research and persistent questioning. Our dependence on AI tools, like chatbots or Google’s new AI Overviews, could well make us “less intelligent.”
By short-circuiting research and verification effortwe risk losing the fundamental skills that have built centuries of scientific progress: curiosity, critical evaluation of sources, and the ability to follow unforeseen leads.
Why is the Royal Observatory concerned about the impact of AI?
The Observatory’s main fear is the gradual disappearance of habits of thought which build expertise and innovation. Paddy Rodgers points out that relying solely on direct machine responses causes us to lose habits of questioning and evaluation.
It’s a bit like using a calculator for simple additions: over time, the mental calculation muscle atrophies. Here it is critical thinking muscle which is threatened.
The history of astronomy, which the Observatory knows well, is full of examples where data collected “unnecessarily” by humans proved crucial decades later.
A machine, optimized for efficiency, would never have done this collection work without an immediate goal. These side paths, these fertile errors and these unexpected discoveries are the fruit of a curious human mind and not of an algorithm.
By depriving ourselves of this process, we distance ourselves from verifiable information and fundamental knowledge.
Is AI only a threat to our intelligence?
If it provides problematic shortcuts for the mind, artificial intelligence is also a formidable tool for science. The case of Demis Hassabis, boss of DeepMind (Google’s AI entity), is striking proof of this. His work on AlphaFold2who predicted the structure of almost all known proteins, won a Nobel Prize.
Nuance is therefore essential. The debate is not about rejecting the tool, but about its use. AI can be a intellectual partneran accelerator of discoveries, provided that it does not become a systematic crutch.
It can analyze massive data sets, far beyond human capabilities, to identify patterns and predict protein structures.
The danger arises when it is no longer an assistant but a substitute for thought.
What is “cognitive externalization” and why is it dangerous?
“Cognitive externalization” (or cognitive outsourcing) is the act of delegating to a technology a task that normally requires our brain, such as memorization or problem solving.
GPS, which saves us from having to memorize a route, is a classic example. Conversational AIs have greatly facilitated this outsourcing for complex tasks.
The real danger lies in the temptation to delegate effort to these new tools, transforming our brain as a simple consultation terminal rather than as a creative engine.
Studies already show that this dependence has a negative and rapid impact on our cognitive skillsour memory and our ability to learn. By stopping making the effort to research, connect ideas, and validate information, we lose the drive that keeps our intelligence sharp.
This is the very essence of the Observatory’s warning: the comfort of instantaneous response has a hidden intellectual cost which is paid for by an erosion of our own know-how.
How to use AI without sacrificing our critical thinking?
The solution lies in an active and critical approach to the tool. Reid Hoffman, co-founder of LinkedInsuggests a powerful method: using AI as a contradictory agent.
Instead of asking for a ready-made answer, we can present our idea and ask: “ What is wrong with my reasoning? “. This approach forces AI to become a sparring partner, which pushes us to renforcer nos arguments and to identify the flaws in our own thinking.
Likewise, academics encourage responsible use where AI is used to challenge concepts or explore solutions collaboratively, but never to “ outsource thinking ».
The idea is to maintain control of the intellectual process. AI can generate drafts, synthesize sources or propose alternatives, but the validation, analysis and final decision must remain human.
