iKala founder: AI has fully activated brain outsourcing.
After ChatGPT sparked a wave of generative AI, numerous AI tools have been widely applied in work, learning, and daily life, and humans' reliance on them has gradually deepened. However, are convenient AI tools assisting human thought, or quietly triggering a large-scale 'brain outsourcing'?
iKala founder Cheng Shijia today (6/18) posted, citing MIT's latest research, pointing out that long-term reliance on AI tools for writing may lead to a significant decline in human 'cognitive abilities' and accumulate a new form of 'cognitive debt'.
This study conducted by MIT researchers Nataliya Kosmyna and others primarily explores the impact of AI-assisted writing on the neural and behavioral aspects of the brain.
The research team invited 54 participants to wear EEG devices and complete multiple writing tasks in batches over four months.
Participants were divided into three groups: the 'ChatGPT group' that used ChatGPT assistance throughout, the 'Google search group' that could only use traditional search engines, and the 'pure brain group' that did not use any tools and relied solely on their own thinking and writing.
MIT's research results show that the 'pure brain group' exhibited the strongest and most widespread neural network connections, while the 'Google search group' showed moderate cognitive engagement, and the 'ChatGPT group' exhibited the weakest overall coupling.
Image source: MIT research.
In simple terms, when humans delegate cognitive responsibilities such as constructing arguments and organizing text to AI, the activity and connections in the corresponding areas of the brain will significantly decrease. Cheng Shijia metaphorically compares this to 'forgetting how to drive after hiring a chauffeur'.
83% of ChatGPT users experience cognitive amnesia, unable to remember the sentences they wrote.
More disturbingly, research found that as many as 83% of ChatGPT users cannot accurately quote sentences they wrote just minutes earlier (with AI assistance). MIT research refers to this as 'cognitive amnesia'; in contrast, only 11% of the 'pure brain group' faced the same difficulty, showing a significant gap.
Image source: MIT research.
MIT's research further confirms that participants in the ChatGPT group consistently performed worse in areas such as neural, language, and writing scores compared to the controls in the 'pure brain group'.
Cheng Shijia pointed out that this is because the processes of thinking and organizing have been completely outsourced, causing users to become 'movers' of information rather than 'internalizers' of knowledge.
Reliance on AI generates 'cognitive debt': trading future abilities for immediate convenience.
MIT researchers have named this over-reliance on AI tools, trading future cognitive abilities for immediate convenience, as 'cognitive debt', which is rooted in the brain's neuroplasticity.
Image source: MIT research.
Cheng Shijia pointed out that this is similar to how GPS has caused many people to become 'directionally challenged': over-reliance on GPS weakens our innate spatial awareness and navigation abilities. Nowadays, ChatGPT elevates this outsourcing level from 'direction sense' to 'thinking' itself, which is deeply concerning.
He emphasized that cognitive ability is the key reason humans pride themselves as the most intelligent beings, and this scientific finding is undoubtedly a significant warning of potential risks to human core abilities.
The consequences of not using the brain are severe; one cannot neglect the fundamentals.
Cheng Shijia stated that the brain has the characteristic of 'use it or lose it': when we strive to think and solve problems, the relevant neural connections are strengthened. If these tasks are outsourced for a long time, these connections will weaken, leading to a decline in cognitive function.
To strengthen brain connectivity, the best ways include 'deliberate practice' and 'desirable difficulty'. The brain learns best and retains long-term memories most deeply when facing challenges that push beyond the comfort zone and have a certain level of difficulty.
However, the seamless convenience provided by AI precisely deprives us of the opportunity to experience this 'beneficial difficulty', causing knowledge to remain superficial and difficult to take root in the brain.
Cheng Shijia also cited the Japanese artisan spirit as an example, pointing out that sushi chefs in Japan refine their craft to perfection, continuously improving their fundamentals.
Japanese traditional crafts also emphasize the 'guard-break-depart' learning three-stage theory. If one uses generative AI tools, it is like an apprentice who has not even experienced the 'guard' stage yet trying to jump directly into the realms of 'break' and 'depart'. The results may seem splendid, but their foundation is unstable.
Image source: thejakartapost Japanese artisan spirit: Sushi chefs refine their craft to perfection, continuously improving their fundamentals.
A glimmer of hope: those willing to write drafts first will enhance their brain's capabilities using AI.
Although MIT's research results are concerning, they also provide a glimmer of hope.
The fourth phase of the experiment in the study showed that those who first independently completed drafts and then used ChatGPT for refinement and editing exhibited an enhanced trend in brain connectivity.
These participants in the 'Brain-to-LLM' group (originally 'pure brain group' who switched to using LLM) exhibited higher memory recall abilities and reactivated extensive occipital-parietal and frontal regions, similar to the 'Google search group'.
This means that if AI tools intervene at specific stages and guide users to deeper cognitive integration, it may bring positive benefits.
Image source: MIT research.
The path to coexistence with AI: first 'guard', then 'break and depart'.
Cheng Shijia proposed a clear path for 'coexisting with AI', emphasizing that the answer is not to ban AI but to use it wisely. That is, first complete the cognitive burden with one's own brain (guard), and then use AI as a reinforcement tool to seek breakthroughs (break and depart).
He called on the next generation of children and newcomers in all fields to see AI as an infinitely potential 'collaborator', not as a 'substitute' that can take over thinking. Tasks should not be handed over to AI without thought.
Cheng Shijia: Only in this way can we enjoy the dividends of technology while avoiding incurring unpayable 'cognitive debt', ensuring that we remain truly independent thinkers and do not outsource our brains from the very beginning.
Further reading:
Will using AI cause brain atrophy? Research: Students' grades dropped by 17%, and some universities have banned the use of AI for writing papers.
'83% of ChatGPT users experience cognitive amnesia! iKala founder: the consequences of brain outsourcing are not just this' This article was first published in 'Crypto City'.