Inside a training workshop on AI-assisted energy systems planning for universal access in Nigeria
About 600 million people across Africa live without reliable electricity. For many of us, that’s a number easy to say, but hard to feel.
In Nigeria alone, the gap between urban and rural electricity access is stark. For the most part, national averages conceal huge variation: while some states have high grid penetration, rural communities in others have essentially none. The challenge, besides inadequate generation capacity, is poor distribution and planning, and the sheer complexity of delivering power to the right places efficiently.
So what happens when you invite 25 energy planners and policy professionals to the University of Port Harcourt, hand them AI tools, and tell them upfront: “You’re not programmers. You’re code pilots. Your job is to know what you want—the AI knows the syntax.”
Something very interesting happens. And it was my privilege to watch it unfold in real time.
A Course Born at the Intersection of Urgency and Possibility
In March 2026, Prof. Ogheneruona Diemuodeke and I co-delivered a three-day intensive training programme titled “Energy Systems Planning for Universal Access in Nigeria: Leveraging AI as Co-Intelligence.” The course was hosted by the Energy Technology Institute at the University of Port Harcourt, Nigeria. It was supported by the Transformative Energy Access Learning Partnership (TEA-LP) at the University of Cape Town, with invaluable coordination from Andrea Fitzpatrick and Mascha Moorlach.
Our premise was simple: energy planners in Nigeria operate in environments that are data-rich but insight-poor. Satellite imagery, population datasets, nighttime light data, and geospatial boundaries. Here’s what this boils down to: raw materials for transformative planning already exist. What’s missing is the analytical bridge between data and decisions. We designed the course to bridge that gap by using AI as co-intelligence, a concept introduced by Ethan Mollick that views AI not as a replacement for human judgment but as an amplifier of it. We tested the four principles for using AI advanced by Mollick:
- Always be ‘In the Loop’: Dive in and use AI regularly to understand how it can transform raw energy data into planning insights.
- Be the ‘Human in the Loop’: Stay in control by checking and guiding the AI’s work to ensure that energy strategies align with real-world human goals.
- Treat AI Like a Person (and Give It a Job): Interact with the AI as a collaborator and tell it exactly what type of energy-planning expert it should be.
- Assume This Is the Worst AI You Will Ever Use: Remember that AI tools for energy policy are rapidly improving, so current technical hurdles are only temporary.
Three Days That Changed How Planners See Their Own Expertise
Day 1: Foundations: Building the Mental Model. We started with a provocative distinction: traditional software follows rules you give it; AI learns patterns from examples you show it. A printed map tells you exactly which turn to make—that’s traditional programming. Teaching someone to read the stars so they can navigate anywhere—that’s machine learning. For energy planning, our problems are too complex to be solved by simple rules. Grid demand, weather, terrain, household income—these interact in ways no rulebook can fully capture.
We introduced the TRUST Decision Matrix, a governance framework for deciding when to rely on AI outputs and when to exercise human override. It was a recurring theme for the entire course: every analytical output the participants produced over three days had to pass through TRUST before it became a recommendation.
Day 2: Applied Analysis: From Spectators to Code Pilots. This was the day that shifted everything. A challenging one at that, for participants and co-trainers alike. We opened Google Colab notebooks and ran a national electrification gap analysis on real Nigerian data. Using geospatial algorithms, they combined Local Government Area boundary data, WorldPop population estimates, and VIIRS nighttime light data to produce choropleth maps that identify the highest-priority LGAs for electrification interventions.
Then we went micro: applying Multi-Criteria Decision Analysis (MCDA) to rank 30 candidate mini-grid sites in Rivers State. And then, a financial viability analysis to check whether technically optimal sites were actually worth investing in.
The secret ingredient was vibe-coding—a paradigm in which participants didn’t write code from scratch but instead directed an AI assistant to modify and extend prebuilt notebooks. Want to change the colour scheme of your choropleth? Ask the AI. Want to filter results to the North-West zone? Tell the AI what you need. Want to add a bar chart panel? Describe it in plain English.
Listening to energy economists who had never written a line of code offer instructions for ChatGPT to modify geospatial analyses in real time! That was the moment I knew we had something.
Day 3: Capstone: Autonomy Under Pressure. On the final day, the baton passed from facilitators to participants. Teams of mixed expertise, energy economists, engineers, and policy advisors, tackled real-world capstone scenarios. Each team had to run the complete analytical cycle: identification of the electrification gap, site selection, financial viability, and a TRUST-justified recommendation, all presented to the group under time pressure.
The deliverable wasn’t just “here are the top 5 sites.” It was: “Here is why we weighted the criteria this way, here is what we’d need to validate, and here is where human oversight is essential before any capital decision.” That distinction between technically optimal and responsibly recommended is the difference the TRUST framework (from the perspective of AI as Co-Intelligence) makes.
What Made This Training Different
Three things set this course apart from typical AI-assisted capacity-building workshops.
First, we looked at AI not as the subject, but as an active co-pilot. Rather than treating it as a passive instrument, we leaned into Co-Intelligence, giving the AI a seat at the table and a defined role in our energy problem-solving process.
Second, the hybrid model demonstrated a smarter way to deploy technical support. Without the need for Daisy or myself to navigate the logistics of travelling to Nigeria, we were able to provide high-level facilitation from the US and UK in real time. This ‘lean’ approach to technical delivery—relying on private facilitator channels and local room management by Prof. Diemuodeke—proves that world-class training can be delivered across Africa without the prohibitive costs of traditional fly-in models.
Third, we prioritised AI governance from the start. Too many AI trainings focus on capability and ignore responsibility. By embedding the TRUST framework from Day 1, every analytical exercise became a governance exercise too. Participants didn’t just learn what AI can do; they learned when not to trust it and why.
The Bigger Picture: Human-AI Relations for Sustainable Energy in Africa
This course sits at a fascinating intersection that I believe will define the next decade of energy transition in Africa: the evolving relationship between human expertise and AI capability in contexts where the stakes—universal energy access, climate resilience, economic development—are extraordinarily high.
The participants in Port Harcourt demonstrated something I’ve long suspected but hadn’t seen proven so vividly: domain experts who understand their context deeply can become powerful AI users in days, not months. The bottleneck was never technical skill. It was access to the right mental models, frameworks, and permission to experiment.
M-KOPA in East Africa uses AI-driven payment history to score creditworthiness for pay-as-you-go solar systems—reaching customers the banking system ignores. Geospatial algorithms are identifying optimal mini-grid sites across West Africa by combining satellite imagery with population data. These applications work not because AI is magic, but because it processes data faster than any human team could. The intelligence is still ours. We define the question.
This is the true essence of AI as co-intelligence: Far from being about replacing human judgment, the workshop was about giving decision-makers superpowers. In a country like Nigeria, where energy planning challenges are immense and world-class expertise is often under-resourced, this partnership is transformative.
Ultimately, our course was designed to put participants on a path to becoming what Mollick calls “AI Connoisseurs”—practitioners who, through hands-on practice, master the art of blending human intuition with machine logic to achieve order-of-magnitude improvements in the productivity and impact of energy planning. The 25 professionals who left Port Harcourt aren’t waiting for that future. They’re already building it.
What’s Next
The participants left with personal replication plans—concrete strategies for applying these tools in their own organisations. Several teams identified immediate applications: mapping electrification gaps in their own states, running MCDA for pending mini-grid investment decisions, and building the case for data-driven planning within their agencies.
We’re working on a journal article on the course that examines the evolving human-AI relations at the nexus of technology, environment, and society in African contexts. This is a story that deserves rigorous documentation, not just because of what happened in Port Harcourt, but because of what it signals about the future of capacity building across the Global South.
If you’re working in energy access, climate adaptation, or AI governance in Africa, I want to hear from you. If you’re a funder wondering whether AI-assisted capacity building actually works at ground level—this is your proof of concept. And if you’re a domain expert who’s been told you need to “learn to code” before AI can be useful to you: you don’t. You need to learn to pilot. Twenty-five Nigerian energy planners just showed the world how.
This training was supported by the Transformative Energy Access Learning Partnership (TEA-LP) at the University of Cape Town and hosted at the Energy Technology Institute, University of Port Harcourt, Nigeria.
