Generative AI And Cognition: Thinking Outside The Bots -By Stephen Agada

• Introduction

The year is 2020, at exactly 6AM, 17th May, this writer sits pondering how much he had to do for the day. So many tasks needing his attention. One question lingers on his mind: “where does he starts from?” The fear of being engrossed in one task and forgetting the others creeps in. However, this writer saves himself from such a dilemma as he opens his Calendly app on his phone; schedules all his tasks for the day and even sets a reminder for each activity. 20 years prior, that would have probably been called sorcery. Interestingly, in today’s world, innovations and solutions like this have been created and amplified through the use of Artificial Intelligence (AI). In fact, one would think that was the peak of it all. But as it is with all kinds of technological advancement, it is constantly evolving. At exactly 2pm WAT, November 30 2022, OpenAI released ‘ChatGPT’[1] thus marking the start of mainstream generative AI. Mankind had never experienced something of this nature. Humans could now create unique text or image results in response to user prompts.

While these technological advancements have largely improved our lives, this piece of work aims to balance the conversation around this subject matter by casting the mind of its readers to a major side effect to this technological advancement (generative AI) slowing eating into mankind. There has been a significant decline in human cognitive abilities since its inception. Humans no longer find it necessary to do critical thinking. They quickly resort to dumping all of their ideas as a prompt to these generative AI tools to do the thinking and analysis for them. Many now suffer from digital dementia. Thus, this writer earnestly desires that at the very least, readers become more enlightened about this issue and certain recommendations that will be made are implemented as a solution to this conundrum.

2.0 Understanding Key Terms

For the purpose of clarity and a clear sense of direction, clarifications must be made to certain key words like “generative AI”, “cognition”, “digital dementia”, etc.

There abounds a replete of explanation that have been provided regarding generative AI. Perhaps, the World Economic Forum sums it better when it highlighted that generative AI refers to a category of artificial intelligence (AI) algorithms that generate new outputs based on the data they have been trained on; unlike traditional AI systems that are designed to recognize patterns and make predictions, generative AI creates new content in the form of images, text, audio, and more.[2] There are a wide range of the applications of generative AI ranging from creating new images from existing ones, to writing articles and even scripts, to generating sound effects and even acting, etc, there is almost no limitation to what generative AI can do. Cognition on the other hand has been described by Cambridge as a range of mental processes relating to the acquisition, storage, manipulation, and retrieval of information.[3] Dementia is a group of symptoms associated with a decline in cognitive function, which may arise from various conditions that progressively damage brain tissue by harming cholinergic and glutamatergic neurons. This damage ultimately impairs cognitive abilities, such as memory, reasoning, and problem-solving, beyond what would be expected from the normal aging process.[4] Specifically, the term ‘digital dementia’ was proposed by a German neuropsychiatrist Dr. Manfred Spitzer in 2012 when he iterated the possible harm to the brain that might result from an over-reliance on technologies such as cell phones and google that can exacerbate attention deficit hyperactivity disorder (ADHD), memory loss, and cognitive decline.[5]

3.0 What the Numbers are Telling Us

The evidence is no longer anecdotal; it is measurable and worrisome. In a study funded by Horizon Europe and published by PubMed Central where 160 participants aged 18-35 years were observed and examined to compare the effects of using generative AI (intervention group) versus not using it (control group) on cognitive effort and writing performance in an analytical writing task designed as a hypothetical writing class assignment for college students. It was discovered that students performed better when using general-purpose generative AI tools but performed worse when these tools were taken away.[6] This rightly suggests that students may have relied on the tool to bypass cognitive processes essential for developing cognitive skills, which ultimately compromised their performance.[7] Furthermore, in a controlled MIT experiment, researchers used EEG to track brain activity in participants writing essays with ChatGPT, with a search engine, or with no tool at all. The results were stark, participants using GPT-4o showed lower cognitive engagement than those using search tools or their own reasoning, and 83.3% of the AI-assisted group could not recall a single correct quotation from the essay they had just written, compared with only 11.1% of the other groups.[8]  This is not an isolated finding. A 2025 survey by Microsoft Research and Carnegie Mellon University, involving 319 knowledge workers, found that people who trusted GenAI tools more experienced a decline in critical thinking, while those confident in their own reasoning retained sharper critical faculties.[9] All these put together describes a generation quietly trading depth of thought for speed of output.

4.0 Thinking Outside the Bots: Strategies to Break Free

Diagnosing a problem is easy. Living differently in response to it is the harder part, and this is where most conversations about AI overreliance quietly stall. It is not enough to know that the convenience that generative AI tools offer us is costing us our cognitive abilities; the real work lies in deciding deliberately to reintroduce the friction that generative AI has made it so easy to avoid.

The first and simplest shift is to change the order of operations. Before opening ChatGPT to draft an argument or summarise a case, spend fifteen minutes wrestling with the problem alone, even if the result is messy or wrong. This single change in sequence preserves the cognitive labour that actually builds understanding, while still allowing AI to play its proper role afterward which is refining, checking, and polishing what the mind has already produced. The MIT Media Lab study referenced earlier found something worth repeating here, that “participants who began with their own thinking and used AI only to revise showed far stronger brain engagement and a clearer sense of ownership than those who let the tool lead from the start.”[10] The lesson therefore is not to abandon AI, but to insist on going first.

Secondly, treat generative AI as a verification tool rather than a generation tool wherever possible. The reason for this is that the former outsources thinking while the latter sharpens it. A law student, for instance, loses very little by using AI to check whether a line of reasoning holds up, but loses a great deal by asking it to construct that reasoning in the first place.

Thirdly, build in AI-free blocks of deliberate practice, the same way an athlete would never skip strength training simply because a machine could lift the weight for them.[11] This might mean writing a first draft entirely by hand before typing anything, memorising a short passage or case citation instead of looking it up each time, or setting aside specific study sessions where no digital assistant is within reach. These blocks will feel inefficient, even uncomfortable, but that discomfort is the entire point. Cognitive stamina, like physical stamina, only grows under load.

Fourthly, practice active recall instead of passive consumption. After reading or researching a topic, close the laptop and try to reconstruct what was just learned from memory alone, then check for gaps afterward.[12] This is a far more demanding exercise than simply rereading notes or re-running a query, but it is also one of the most well-established methods for converting information into lasting knowledge, precisely because it forces the brain to retrieve rather than simply recognise.
Finally, and perhaps most importantly, cultivate an honest relationship with AI itself. The truth remains that it is not every shortcut that is harmful, and not every use of AI constitutes decline. The danger lies specifically in reflexive unexamined reliance, i.e the instinct to reach for a tool before even attempting the task oneself. Breaking that reflex does not require rejecting the technology; it requires restoring a pause between the question and the answer, which entails a moment in which the mind is given the first chance to try. That pause, small as it seems, may be the single most important habit standing between a generation of engaged thinkers and a generation quietly outsourcing the very faculties that make them so.

5.0 Implication for Policy (Law) Makers and Stakeholders

Having understood how important this discourse is, there’s the need to also draw the attention of policy makers and stakeholders. Most existing AI governance frameworks were designed to address a narrower set of risks which includes bias in algorithmic decision-making, data privacy violations, safety failures, the misuse of AI for fraud or manipulation, etc. The EU AI Act, for instance, organises its entire regulatory architecture around a risk-based classification system, sorting AI applications into categories of unacceptable, high, limited, and minimal risk. Yet cognitive decline, the slow cumulative weakening of memory, reasoning, and independent thought through everyday reliance on generative tools, fits awkwardly into this structure. It is not a single catastrophic failure that can be flagged and penalised. It is a diffuse, gradual harm that accumulates quietly across millions of ordinary interactions, and regulatory frameworks built to catch dramatic breaches are simply not designed to notice a decline this slow.

Although the GDPR offers a useful parallel where “it restricts automated decisions made without meaningful human involvement, recognising the danger of removing humans from consequential processes.”[13] Yet this is not enough. Policymakers should broaden AI regulation to address cognitive and developmental harm, especially among students, and fund longitudinal research tracking these effects. For Nigeria and other African jurisdictions still shaping AI governance, this is a rare chance to embed cognitive welfare into law before dependency takes root at scale.

6.0 Conclusion

The stance of this writer is clear, generative AI itself is not the enemy; over reliance on it is. Reclaiming our cognitive independence stands as a panacea to digital dementia. As law, education, and policy evolve alongside this technology, the real challenge ahead is ensuring humans remain the ones doing the thinking.

[1] “ChatGPT, the Generative AI Chatbot, Is Released” (ThisDayInHistory, November 4, 2025) https://www.history.com/this-day-in-history/november-30/chatgpt-released-openai. accessed July 17, 2026

[2] Routley N, “Generative AI Explained” (World Economic Forum, February 6, 2023) <https://www.weforum.org/stories/emerging-technologies/generative-ai-explain-algorithms-work/> accessed July 17, 2026

[3] “What Is Cognition?” (Cambridge Cognition, August 15, 2015) < https://cambridgecognition.com/what-is-cognition/> accessed July 17, 2026

[4] Ali Z, Janarthanan J and Mohan P, “Understanding Digital Dementia and Cognitive Impact in the Current Era of the Internet: A Review” (PubMed Central, September 23, 2024) <https://pmc.ncbi.nlm.nih.gov/articles/PMC11499077/> accessed July 17, 2026

[5] Spitzer M, “Digital Dementia: What We and Our Children Are Doing to Our Minds” (2014) 3 Cognitive Remediation Journal 1 <http://cdzjesenik.cz/journal/>

[6] “Effects of Generative Artificial Intelligence on Cognitive Effort and Task Performance: Study Protocol for a Randomized Controlled Experiment among College Students” (2025) 1–1 PubMed Central 1 <https://pmc.ncbi.nlm.nih.gov/articles/PMC12255134/#notes2>

[7] Ibid

[8] Lu J, Zhao GG and Zheng A, “Generative AI Use in Entrepreneurship: An Integrative Review and an Empowerment-Entrapment Framework” (MIT, April 2026) <https://www.researchgate.net/publication/403420864_Generative_AI_Use_in_Entrepreneurship_An_Integrative_Review_and_an_Empowerment-Entrapment_Framework> accessed July 17, 2026

[9] “The Impact of Generative AI on Critical Thinking: Self-Reported Reductions in Cognitive Effort and Confidence Effects from a Survey of Knowledge Workers” (Microsoft, May 1, 2025) <https://www.microsoft.com/en-us/research/wp-content/uploads/2025/01/lee_2025_ai_critical_thinking_survey.pdf> accessed July 17, 2026

[10] Lu J, Zhao GG and Zheng A, “Generative AI Use in Entrepreneurship: An Integrative Review and an Empowerment-Entrapment Framework” (MIT, April 2026) <https://www.researchgate.net/publication/403420864_Generative_AI_Use_in_Entrepreneurship_An_Integrative_Review_and_an_Empowerment-Entrapment_Framework> accessed July 17, 2026

[11] Germain T, “Think Outside the Bots’: How to Stop AI from Turning Your Brain to Mush” (BBC, May 6, 2026) https://www.bbc.com/future/article/20260505-how-to-use-ai-without-turning-your-brain-to-mush accessed July 20, 2026

[12] Ibid

[13] Section 4, Article 22, General Data Protection Regulation (GDPR),(EU) 2016/679.