Note: this article covers the MIT Study (read on to find out more) and my recent AI Adoption and Ethics survey that I partly shared early insights at the recent Digital Future Conference and Exhibition 2025.

I’ve recently seen quite a lot of posts on social media taking the MIT study titled “Your Brain on ChatGPT” out of context – mostly to scare people. It’s gone viral across TikTok and Instagram, with dramatic captions like “ChatGPT makes you dumber” or “AI is killing your brain”.

They are completely missing the point. For your information, in case you didn’t read through the whole 200 (I think) pages (I mean, who has the time?), the study isn’t saying we should stop using AI. It’s not even anti-AI. What it actually reveals is much more interesting and far more useful.

The research dives into how our brains respond when we use tools like ChatGPT to write. And instead of serving as a warning to stay away from AI, it highlights something we really need to pay attention to: how we use AI matters.

“The potential of LLMs to support meaningful learning depends significantly on the user’s approach and mindset” (Kosmyna et al., 2025, p.16).

“Educational interventions should consider combining AI tool assistance with tools-free learning phases to optimize both immediate skill transfer and long-term neural development” (Kosmyna et al., 2025, p.139).

This couldn’t come at a better time for Brunei. As we roll out national efforts like the Data and AI Strategy (yesterday I attended the EU-Brunei Policy Outreach & Exchange Seminar on AI and this was mentioned by the MTIC Acting Permanent Secretary in his opening remarks) and push towards the goals of Wawasan 2035, we need to ask ourselves these questions: Are we preparing our people, not just in schools, but across industries to use AI wisely? Or are we slipping into habits that make us overly reliant on it, without even realising it?

What the MIT Study Really Found

The MIT research team worked with 54 participants who were asked to complete SAT-style writing tasks under three different conditions: writing on their own without any tools, using a search engine like Google or using an AI chatbot like ChatGPT. While they worked, their brain activity was recorded using EEG headsets.

So what happened?

Participants who wrote without any help showed the most active and distributed brain activity, especially in the alpha and beta wave regions linked to memory, reasoning and creativity. Those using search tools showed a moderate level of engagement. But those using ChatGPT had the weakest neural activity – not just while writing but also when asked to recall what they had written.

To put it simply, most of them didn’t remember much. Around 83% of the AI users couldn’t accurately quote their own writing. Many said they didn’t feel very connected to the final product either. Their work tended to sound the same, less variety in structure, less distinct voice. That’s what the researchers called cognitive offloading.

They introduced a term for this: Cognitive Debt. It’s a bit like credit card debt. You get convenience now, but you end up paying for it later – with reduced memory, understanding and a weaker sense of ownership over your ideas.

But, and this is the part that often gets left out in social media posts, that debt isn’t permanent. The study found that when people started with their own thinking and then brought in AI support, their brain activity remained more engaged. In those cases, AI actually helped support creativity and deeper thought because it wasn’t being used as a shortcut, but as a tool.

“Strategic timing of AI tool introduction following initial self-driven effort may enhance engagement and neural integration” (Kosmyna et al., 2025, p. 139)

The Bruneian Context: Early Adoption, Uneven Understanding

At the national level, Brunei is moving quickly on digital transformation. We have an updated Digital Economy Masterplan which has been rebranded to the Digital Brunei Transformation Masterplan (as mentioned by Yang Berhormat Pengiran Dato Seri Setia Shamhary bin Pengiran Dato Paduka Haji Mustapha, Minister of Transport and Infocommunications at the Digital Future Conference and Exhibition 2025 Brunei) and work on a National Data and AI Strategy is underway. But while the planning side is progressing, public-facing AI literacy is still just getting started.

To understand where we really stand, I conducted a nationwide survey this year on Generative AI adoption and ethics. It drew 309 responses from people working across different sectors. What came through strongly reflects some of the same concerns raised in the MIT study.

Here’s what early data showed:

• Only 24% said they clearly understood the difference between “AI” and “Generative AI.”
• 274 people had used tools like ChatGPT mainly for writing tasks and work-related content.
• Average confidence in using AI tools was 3.29 out of 5.
• Trust in AI-generated content was also moderate, at 3.14 out of 5.
• 84 respondents felt using AI was unethical or similar to cheating.
• 67 said these tools often felt biased towards Western perspectives.

Despite all this, AI is already being used widely especially in education and the public sector. That’s the gap we need to pay attention to. People are using these tools without always understanding how they work, without feeling confident in their output and without systems in place to make sure they’re thinking through what the AI gives them.

Why AI Literacy Must Be Part of Brunei’s National Strategy

Brunei’s Wawasan 2035 outlines a future where the population is “well-educated, highly skilled and accomplished”. But getting there isn’t just about rolling out infrastructure or introducing new tech. It also means giving people the mindset and skills to use those tools wisely.

That’s why AI literacy shouldn’t be treated as an optional add-on. It needs to be part of the foundation. By that, I don’t just mean knowing how to open ChatGPT and type in a prompt but about:

• Knowing when AI should support you and when it shouldn’t. Not every task needs automation. Sometimes it’s better to think things through before turning to a tool.
• Being able to organise your own thoughts first. Writing a rough outline or reflecting on your point of view before asking AI to expand can help keep your voice in the final result.
• Questioning what AI gives you. Can you tell if it’s accurate? Are there gaps or assumptions you should check? That kind of awareness takes practice.
• Recognising ethical grey areas. Is the content original? Are you comfortable taking credit for it? Do you know when AI crosses into plagiarism or misinformation?
• Understanding bias and context. Many AI tools are trained on global, often Western-centric data, even to this day. You need to be able to spot when that skews the perspective.
• Building cognitive resilience. In simple terms, this means not letting your thinking get dull. AI can help speed things up but if you use it too passively, it can also do the thinking for you.

Without that kind of grounding, it’s easy to fall into habits that look productive on the surface but quietly erode critical thinking underneath. You get answers quickly, but you might not remember how you got there. Or worse, you start to feel disconnected from the ideas you’re sharing because they’re not really yours anymore.

“The convenience of instant answers that LLMs provide can encourage passive consumption of information, which may lead to superficial engagement, weakened critical thinking skills, less deep understanding of the materials, and less long-term memory formation.” (Kosmyna et al., 2025, p. 12)

“LLM group, while benefiting from tool efficiency, showed weaker memory traces, reduced self-monitoring, and fragmented authorship.” (Kosmyna et al., 2025, p. 138)

What Brunei Can Do Now

The good news is that Bruneians want to learn. My survey showed that:

• 230 respondents would take an AI training course if it were free.
• But 176 said their workplace or university currently provides no AI training at all.
• Many cited barriers like lack of time, cost and overly technical content.

This is the perfect moment for Brunei’s leaders (educators, policymakers, industry partners) to step in and design accessible, values-driven AI literacy programmes. These could include:

1. A National AI Literacy Framework

Brunei could adopt or adapt an AI literacy framework that works across levels and industries – from educators and civil servants to creatives, entrepreneurs and SMEs. A multi-tiered model (basic, intermediate, advanced) could help individuals understand not just how to use AI but when, why and with what level of oversight. This could be embedded into the Digital Brunei Transformation Masterplan and Brunei’s broader Data and AI Strategy.

One useful reference point is the OECD’s AI Literacy Framework, which outlines core competencies in understanding, using and evaluating AI systems. While originally developed with a focus on education, its structure is adaptable for wider use. It covers areas such as:

• Understanding how AI systems work
• Assessing reliability and bias
• Using AI responsibly in decision-making
• Navigating ethical implications

Another broader option is the UNESCO Digital Literacy framework, which includes AI and algorithmic awareness as part of 21st-century skills. For industry-specific guidance, elements from Singapore’s AI Verify framework and the European Commission’s AI Skills Strategy may also be worth reviewing, especially for public sector and business applications.

2. Cognitive Ownership Guidelines

In both education and the workplace, policies should promote habits that strengthen independent thinking before AI is brought in. This could be as simple as encouraging people to draft by hand first, jot down their main points or reflect on their ideas before asking AI to improve or rephrase them.

In my own work, I’ve seen this play out more times than I can count, people jumping straight into AI tools without understanding how it shapes what they’re creating or thinking through the implications of using it. The habits seem efficient, but they often come at the cost of clarity, memory and ownership.

This was also reflected in the survey I conducted on Generative AI use in Brunei. While 274 respondents had used tools like ChatGPT, the average confidence level (as mentioned earlier) was just 3.29 out of 5. As as you may know from that as well, many also shared concerns around trust, ethics and the feeling that using AI might be “cheating”. People are using these tools but not always with the awareness or structure needed to use them well.

These are examples I’ve mentioned many times in my workshops: In schools, this could be as simple as asking students to submit a quick handwritten outline before using AI or having them explain how they used it when presenting their work. In the workplace, it might just mean including a short note or reflection before pasting AI-generated content into a report or email. They do not need to be complicated – just small steps to make sure people stay connected to their own thinking.

These small practices build metacognition and self-awareness. They help maintain a sense of authorship and prevent the quiet erosion of critical thinking that happens when AI is used too early or too passively.

More importantly, they align with Brunei’s MIB values especially the emphasis on personal integrity, accountability and purposeful contribution. Remember, it shouldn’t just be about doing things faster, but also about doing them meaningfully, with ownership of your own thinking.

3. Sector-Specific Modules

Training doesn’t need to be heavy or overly technical. What’s more useful right now are short, practical modules built around real situations people face in their work. The goal is to help them think through how to use AI without losing sight of what matters like ethics, responsibility and context. These are always my goals when designing modules for my trainings and workshops.

• For educators: Focus on how to bring AI into the classroom without compromising student learning or assessment integrity. This could include examples like: when it’s appropriate for students to use AI in drafts, how to design assignments that can’t just be copy-pasted, or how teachers themselves can use AI to support lesson planning while keeping the focus on pedagogy, not automation.
• For government officers: Cover how to use AI for productivity (things like summarising long documents or drafting responses) while being careful not to share sensitive or internal information. Real-life cases could include how to check AI outputs for accuracy, how to maintain confidentiality and how to avoid policy or data exposure.
• For SMEs and entrepreneurs: Help business owners use AI tools to create marketing content, respond to customers or develop product descriptions (without sounding generic or losing the tone of their brand). Practical tips could focus on how to combine AI with their own voice, how to check for originality and how to manage customer trust when using AI-generated materials.

Each module could be completed in under an hour, include sector-specific scenarios and highlight one or two key risks and habits to build. Over time, these micro-trainings could form the foundation of a broader AI literacy programme across the country.

4. Make AI Literacy a Public Conversation

Public understanding of AI can’t be limited to classrooms and offices. It needs to reach everyone: at home, on the road and on their phones. Just like we’ve done with digital safety and financial literacy, we should be making responsible AI use part of everyday conversations. We need to use channels people already trust and tune into:

• Short explainer segments on the radio
• Campaigns through social media that break things down in simple terms
• Community events or talks at schools, mosques or local centres
• Even posters or infographics in public spaces that raise basic questions like: “Did you know AI might not always be right?” or “Have you asked how this was generated?”

Authority for Info-communications Technology Industry of Brunei Darussalam (AITI) and Digital Brunei platforms are already doing this so well on their social media platforms. Again, things do not need to be complicated. What matters is that people start becoming more aware, not just of what AI can do, but of what they need to watch out for. Our goal shouldn’t be to scare people off AI. We need to help them use it with more care, more clarity and a little more confidence.

Final Thoughts: From Cognitive Debt to Cognitive Growth

The MIT study isn’t an attack on AI. It’s a mirror. It reminds us that AI use isn’t neutral. It shapes how we think, what we remember, and how much ownership we have over our own ideas.

But it also points to a way forward. The more intentionally we use AI (by drafting first, questioning the output and reflecting on how we use it) the more we can protect and even strengthen our thinking.

This is the moment to pay attention for us in Brunei. AI is already finding its way into our schools, ministries and workplaces. What we do now will shape whether it becomes a tool that supports human thinking, or one that slowly replaces it.

We shouldn’t see this (especially the MIT study) as a warning to turn away from AI. It’s a call to guide people toward using it with care, purpose and awareness before habits form that are hard to undo.

No doubt, the future will be AI-assisted. The real question for us is: will it still be critically, ethically and cognitively human?