CO₂ Storage Breakthrough: Lessons for AI in Ghana

Prince Henry Sampson Eduam is 24, Ghanaian, and he just did something a lot of people said couldn’t be proven convincingly in a lab: he showed strong experimental evidence that a massive shale formation can trap injected CO₂ and hold it for geological timescales.

That’s climate science—yet it’s also a sharp reminder of how progress really happens: not by hype, but by measurement, iteration, and systems thinking. And if you’ve been following our “Sɛnea AI Reboa Adwumadie ne Dwumadie Wɔ Ghana” series, you’ll notice the parallel immediately. The same mindset that validates CO₂ storage in rock is the mindset that makes AI in Ghana’s fintech ecosystem useful (and safe) for akɔntabuo, mobile money, credit scoring, fraud detection, and compliance.

This post breaks down what Eduam discovered, why it matters, and the practical bridge to our campaign theme: AI ne Fintech: Sɛnea Akɔntabuo ne Mobile Money Rehyɛ Ghana den—how AI-led automation can strengthen trust, reduce losses, and speed up financial services in Ghana.

What Eduam’s CO₂ finding really proves (beyond headlines)

Eduam’s core contribution is simple to state: CO₂ doesn’t just “sit” in shale; it gets captured and becomes harder to move over time. That’s the part skeptics have challenged for years.

Shale formations are often seen as risky CO₂ sinks because they’re complex: ultra-tight pores, variable organic content, and fractured pathways that can behave unpredictably. The fear is obvious—inject CO₂, and it might migrate, escape, or fail to stay immobilized.

Eduam’s experiments provided a strong counter: CO₂ can be adsorbed, immobilised, and structurally confined in the Marcellus Shale (a huge unconventional reservoir system in North America). His work adds credibility because it doesn’t rely on one test. It uses multiple methods across scales—from molecular interaction to core flow.

Here’s the quotable takeaway:

Good climate claims need hard evidence across multiple layers of reality—chemistry, physics, and real-world conditions.

That same rule applies when people market “AI-powered” anything in fintech.

The “two-phase trap” insight: adsorption, then locking

Eduam measured how gases move through shale under reservoir conditions using pulse-decay permeability testing. He tested helium, methane, and CO₂—and CO₂ behaved differently.

He observed a two-phase decline in permeability after CO₂ injection:

1. Rapid initial drop: explained by adsorption—CO₂ binds to internal pore surfaces.
2. Further decline later: not just adsorption; the rock’s internal structure changes in a way that restricts flow even more.

He validated the “why” using:

• BET/BJH surface area and pore structure analysis: confirmed dense nanopores that preferentially attract CO₂.
• Raman spectroscopy: showed CO₂ interacts with kerogen (organic matter in shale), causing slight swelling/rearrangement—narrowing pore throats and reducing connectivity.

So the storage story isn’t “we injected CO₂ and hoped.” It’s “CO₂ triggers mechanisms that make storage improve over time.”

Why this matters to Ghana—even if the rock is in North America

The obvious impact is global decarbonisation, but there are two Ghana-specific angles that matter more than people admit.

First: Ghana’s credibility in high-stakes technical fields is rising. Eduam didn’t just publish; he competed at the highest level.

His paper—“Sorption-Induced Permeability Evolution and Chemo-Mechanical Interactions of CO₂ Sequestration in the Marcellus Shale”—took him through the Society of Petroleum Engineers student competition to the international finals, where only eight researchers worldwide were finalists.

Second: the breakthrough points to a practical model of development:

• Use existing infrastructure (in this case, natural gas assets)
• Repurpose it into new value (CO₂ storage)
• Prove safety with rigorous measurement

That “repurpose and validate” model is exactly how Ghana should approach digital financial infrastructure too.

Most institutions don’t need to rebuild everything from scratch. They need to modernize what exists—core banking systems, mobile money rails, agent networks—then apply AI carefully to improve outcomes.

The real bridge: what CO₂ storage science teaches AI fintech teams

CO₂ storage in shale and AI in fintech look unrelated until you focus on one word: trust.

• CO₂ storage only scales when regulators and communities trust the containment.
• Mobile money and digital finance only scale sustainably when customers and regulators trust the system.

Here are four direct lessons fintech teams in Ghana can borrow from Eduam’s approach.

1) Don’t sell “AI.” Prove performance under real conditions

Eduam didn’t rely on one result; he used a framework that tested multiple mechanisms. Fintech teams should do the same.

If you’re deploying AI for fraud detection, credit scoring, or transaction monitoring, “accuracy” in a demo isn’t enough. You need:

• Performance across customer segments (urban vs peri-urban, salary vs gig income)
• Performance during stress (holidays, salary weeks, election periods)
• Monitoring for drift (fraud patterns evolve like reservoir conditions)

A good internal rule:

If you can’t measure it weekly, you can’t manage it safely.

2) Model the system, not just the prediction

Eduam measured permeability changes, then used spectroscopy to explain why it changed. That’s causal thinking.

In AI-driven financial services, the prediction is only one piece. The system includes:

• User behavior (PIN sharing, SIM swaps, social engineering)
• Agent networks (cash-in/cash-out liquidity issues)
• Product design (limits, fees, reversal processes)
• Compliance requirements (KYC, AML)

A fraud model that improves “precision” but increases false declines for genuine users can quietly destroy trust—especially in mobile money where customers expect speed.

3) Use existing rails, then automate the bottlenecks

Marcellus has decades of infrastructure. The opportunity is transition, not replacement.

Ghana’s fintech and banking ecosystem also has mature rails:

• Mobile money wallets and agent networks
• Interoperability and bank-to-wallet flows
• Merchant payments and bill payment ecosystems

The highest ROI AI use-cases often sit in bottlenecks like:

• Automated KYC document checks and risk flags
• Customer support triage (ticket routing, intent detection)
• Transaction anomaly detection at scale
• Collections prioritization and affordability assessment

This is Sɛnea AI reboa adwumadie in real terms: faster operations, fewer errors, lower cost per customer.

4) Safety isn’t a feature; it’s the product

Eduam’s work is fundamentally about containment and long-term integrity.

For AI in fintech, safety translates into:

• Fairness: avoid systematic bias against certain regions, professions, or demographics
• Explainability: internal teams must understand why decisions happen (especially in credit)
• Auditability: regulators and risk teams need logs, thresholds, and model governance
• Resilience: models must degrade gracefully when data is missing or messy

In Ghana, the best AI deployments will be the ones that reduce fraud and credit losses without turning digital finance into a frustrating “computer says no” experience.

Practical examples: where AI can strengthen akɔntabuo and mobile money

The campaign focus—akɔntabuo ne mobile money—is where AI can have immediate, measurable impact. Here are concrete examples I’d push for (and what “good” looks like).

AI for mobile money fraud: faster detection, fewer customer losses

Answer first: AI reduces mobile money fraud by spotting unusual patterns across millions of transactions in near real time.

What it can catch well:

• SIM-swap risk signals (device change + location shift + high-value transfers)
• Agent float manipulation and abnormal reversal patterns
• Social engineering bursts (many similar transfers to newly created wallets)

What “good” requires:

• Human-in-the-loop review for high-impact actions (freezes, blacklists)
• Tiered responses (step-up verification instead of outright blocking)
• Continuous feedback from confirmed fraud cases

AI for akɔntabuo automation: cleaner books, faster decisions

Answer first: AI improves akɔntabuo by reducing manual reconciliation and catching errors early.

High-value workflows in Ghanaian SMEs and institutions:

• Auto-categorising transactions and receipts
• Matching payments to invoices (especially across bank + wallet)
• Predicting cashflow gaps before they hit payroll

If you’ve ever watched a finance team scramble in late December to close books, you know this matters. December is busy, mistakes are expensive, and nobody wants surprises in January.

AI for credit: smarter risk checks without excluding people

Answer first: AI can expand access to credit by using more signals than traditional salary slips—if governance is strong.

Potential signals (used responsibly):

• Transaction consistency and wallet inflows
• Merchant payment history
• Bill payment regularity

Non-negotiables:

• Clear adverse action reasons (why a customer was declined)
• Regular bias testing
• Conservative limits until models prove stability

What Ghanaian innovators should copy from Eduam’s playbook

Eduam’s success isn’t only the science; it’s the process. If you’re building AI products for financial services in Ghana, steal these habits:

1. Design experiments that can fail honestly. If your pilot can’t disprove your idea, it’s marketing.
2. Test across layers. Prediction metrics + operational metrics + customer trust metrics.
3. Document everything. Model versions, thresholds, data sources, and exceptions.
4. Compete globally. Not for applause—because global standards force discipline.

The reality? Ghana doesn’t have a talent problem. We have an execution and trust problem. Fix those two, and adoption follows.

Where this goes next: AI can accelerate climate and finance together

Eduam’s research points to a future where climate solutions scale because we can model the subsurface better. AI is already used globally for geological interpretation, anomaly detection in sensor streams, and simulation speed-ups.

Now connect that back to finance: climate projects need funding, insurance, carbon accounting, and transparent reporting. The same AI approaches used to validate CO₂ storage integrity can support MRV (measurement, reporting, verification) and reduce greenwashing.

So yes—this is a story about a Ghanaian researcher in a petroleum engineering competition. But it’s also a story about how Ghana can build systems people trust, whether those systems store carbon underground or store value in a mobile wallet.

If you’re leading a bank, fintech, SACCO, or enterprise finance team and you want AI to improve akɔntabuo and mobile money operations, start with one question: what would “secure storage” mean in your system—of money, data, and customer trust?