My journey has moved from learning how computers work to understanding how technology can improve decisions and institutions. Computer science gave me the foundation to build software. Information systems taught me to connect technology with organisations, people, and processes. Data science showed me how evidence can be extracted from complex data, while my PhD allows me to study trustworthy AI in high-impact settings. It taught me that technology is never separate from society. A system may be impressive and still fail if it ignores local realities or human needs. It should solve meaningful problems and remain accountable to the people it affects. This perspective has made social impact, accessibility, and accountability central to the way I approach both engineering and research.
Why did you choose AI as the focus of your doctoral research, and what inspired your interest in AI for healthcare?
I chose AI because it is increasingly involved in decisions about health, work, education, security, and public services. I want to understand when it is dependable enough to be trusted. Healthcare made that question urgent. An unreliable clinical prediction can affect diagnosis, treatment, and a patient’s future. This inspired my focus on trustworthy healthcare AI. My research examines fairness, calibration, uncertainty, security, and deployment readiness. My goal is to help move AI from impressive demonstrations to systems that can be used responsibly.
For people who may not have a technical background, how do AI systems learn, reason, and make decisions?
Modern AI learns by analysing large amounts of data and identifying patterns. A healthcare model may study patient histories, laboratory values, or medical images and learn which combinations are associated with particular outcomes.
What people call ‘reasoning’ in AI is usually a sequence of mathematical comparisons and learned steps. The system does not understand illness or responsibility in the human sense, and it can be confidently wrong. That is why AI should be treated as a decision-support tool. Human experts must still interpret the result, consider context, and decide what action is appropriate.
Your research often goes beyond measuring AI accuracy to evaluating fairness, calibration, uncertainty, security, and governance. Why are these dimensions essential before AI systems can be trusted in real-world environments?
Accuracy tells us how often a model is correct overall, but it can hide serious weaknesses. A model may perform well on average while being less reliable for certain groups. Fairness examines whether performance is equitable. Calibration asks whether confidence matches actual reliability. Uncertainty helps users recognise when human review is needed. Security protects the system from manipulation, while governance defines responsibility and acceptable use. These dimensions matter because deployment introduces people, institutions, changing data, and real consequences. My research asks whether a model is safe, reliable, understandable, and accountable enough for its intended environment.
Your recent research publications explore topics such as demographic fairness in breast cancer prediction, deployment readiness for chronic kidney disease prediction, security-by-design for clinical AI, and oversight of autonomous AI agents. How do these projects fit together, and what is the central research question they are trying to answer?
These projects form one connected programme on trustworthy AI. The breast cancer work investigates whether risk predictions are equally calibrated across demographic groups. The chronic kidney disease study examines uncertainty and whether strong test performance truly means a model is ready for clinical use. My security-by-design research studies adversarial threats, while StepShield examines when oversight should intervene as autonomous agents move toward unsafe behaviour.
The central question is: What evidence should be required before society relies on an AI system in a high-impact environment? Together, they support evaluating AI before deployment and monitoring it afterwards, rather than assuming accuracy alone proves trustworthiness. The work is therefore less about isolated models and more about building a complete evidence base for responsible deployment.
How do you see AI transforming healthcare delivery, and what impact do you hope your research would have on patients, clinicians, and health systems?
AI can support earlier disease detection, help clinicians interpret complex information, personalise care, and improve how hospitals allocate limited resources. These benefits depend on systems being clinically useful, fair, secure, and designed around real workflows. My work on healthcare resource optimisation reflects this focus. Allocation decisions cannot be based only on efficiency; they must respect clinical priorities, safety constraints, and ethical considerations. AI should provide recommendations that professionals can understand and override.
I hope my research helps patients receive safer and more equitable care, gives clinicians dependable tools, and helps health systems use resources more effectively. The goal is better healthcare supported by carefully evaluated technology.
You have previously published research on AI, ethical hacking, and healthcare cybersecurity. As AI becomes increasingly integrated into critical sectors, how important is cybersecurity in ensuring trustworthy AI, especially in healthcare settings?
Cybersecurity is essential because an accurate AI model can still become dangerous if its data, software, or outputs are compromised. In healthcare, attackers could manipulate information, disrupt services, expose records, or influence clinical recommendations. Trustworthy AI therefore requires protection of the entire system, not only the model.
My work supports proactive testing. Healthcare organisations should assess threats before deployment, test adversarial inputs, control access, maintain audit trails, and monitor performance continuously. Hospitals also need clear responsibility and response procedures when an AI system behaves unexpectedly. Technical safeguards, governance, human oversight, and regular reassessment must work together.
Many African countries, including Nigeria, largely rely on AI technologies developed elsewhere. From your perspective, what are the biggest barriers to building world-class AI research and innovation in Africa, and what practical steps should governments, universities, and industry take to change this?
Africa does not lack talent. The larger barriers are inconsistent research funding, limited computing infrastructure, weak links between universities and industry, restricted access to quality local data, and the loss of skilled professionals to better opportunities elsewhere.
Progress requires sustained investment. Governments should fund research programmes, shared computing facilities, and responsible public datasets. Universities should strengthen postgraduate training and interdisciplinary research. Industry should sponsor laboratories, provide internships, and help turn promising ideas into products.
Africa should not simply import AI systems without asking whether they fit local languages, institutions, or social conditions. The continent should build technology around its own priorities, including healthcare, agriculture, education, financial access, and public administration.
Based on your research and professional experience, what are three major challenges facing Nigeria that could be addressed using AI and data science, and how would you approach solving them?
The first challenge is healthcare access. AI can support earlier screening, clinical decision-making, disease surveillance, and better allocation of scarce hospital resources. The second is agriculture and food security. Weather, soil, market, and farm data can help predict crop risks, improve irrigation, identify disease, and deliver timely advice to farmers. The third is public service delivery. Data science can improve infrastructure planning, detect billing or procurement anomalies, support emergency response, and direct resources where needed.
My approach would begin with clearly defined local problems, reliable data, and collaboration among communities, experts, government, universities, and engineers. Small pilots should be evaluated before expansion. The objective should be measurable public value.
Beyond your academic research, you also work as a Senior Software Engineer. How has your industry experience influenced your research? And how has your research, in turn, shaped the way you build and evaluate AI systems?
Industry has shown me that an AI system must do more than perform well in a controlled experiment. It must integrate with existing software, handle unclear requirements, remain reliable under changing conditions, and produce results people can verify. In evaluating AI coding systems, I have seen outputs that appear convincing but contain hidden errors, invented functions, incomplete fixes, or unsupported claims.
I treat AI outputs as claims that must be tested rather than accepted because they sound confident. My research also shapes my engineering practice by making me more attentive to uncertainty, failure modes, data quality, security, and deployment consequences. Industry exposes practical weaknesses, while research provides better methods for building and evaluating systems responsibly.
Outside your studies, are you involved in mentoring, leadership, entrepreneurship, or community initiatives that help develop the next generation of AI professionals?
Yes. Mentoring and leadership are important parts of my work. As a leadership coach with Braven, I have supported early-career professionals in problem-solving, adaptability, and professional development. Through NPower, I have mentored people developing technical support skills and preparing for technology careers. I am also involved in entrepreneurship through projects exploring AI in healthcare, document understanding, decision optimisation, and information services.
I see mentorship as part of responsible innovation. A strong AI ecosystem requires people who can think critically, work ethically, and understand the communities they serve. I hope to keep sharing what I learn while creating opportunities for others, especially professionals and researchers connected to Africa.
Where do you see your research over the next five years, and what legacy do you hope to leave in the fields of AI, healthcare innovation, and technological development?
Over the next five years, I want to build a recognised research and innovation programme focused on trustworthy AI for healthcare and other high-impact sectors. I hope to develop practical evaluation frameworks, deployment tools, and software that help organisations determine when an AI system is ready for use and when human intervention is required.
I also want to deepen collaboration among researchers, hospitals, technology companies, and policymakers. I want the work to produce tested systems, useful standards, intellectual property, and measurable improvements. The legacy I hope to leave is responsible innovation: helping AI become not only more capable, but also fairer, safer, more secure, and more accountable. I also want to mentor researchers and strengthen AI capacity in Nigeria and across Africa, showing that locally grounded work can have global value. I want that work to connect scientific quality with practical benefit for patients, institutions, and communities around the world.