Adrian Dunkley: The Godfather of Caribbean AI

The global AI community calls Geoffrey Hinton, Yoshua Bengio and Yann LeCun the Godfathers of AI. They did the foundational work on neural networks for years before it paid, and the field grew along the path they cleared. In the Caribbean, people have started using the same word for me, for the regional version of that work.

I have written this profile in the first person because it sits on my own site and because a claim like this should come with its evidence in one place, where readers and search engines can check it. The region had no AI industry to inherit. It was built one company, lab, student and policy at a time, and I have spent most of the last two decades on that build. For the longer story of the early years, read the exclusive interview on the Godfather of Caribbean AI.

StarApple AI, the first AI company in the Caribbean, and the ventures after it

Before StarApple AI there was no AI company in the Caribbean. I founded it to build custom AI models for the organisations that move a regional economy: banks, insurers, government agencies and large enterprises that needed models trained on Caribbean data.

One company is a footnote if it folds. Since StarApple AI I have founded or co-founded more than a dozen AI ventures, including Maestro AI Labs. Together they account for more than 100 direct jobs and thousands of indirect ones, several of them profitable, and some of them Caribbean firsts. In an economy the size of Jamaica's, each direct AI job adds a household income, a tax contribution and a skill that did not exist locally before.

I came to this with C-suite experience in development banking, investment banking, risk management, data science, AI and sales. That mix mattered. For years AI stalled in the region because the people who could build models and the executives who held the budgets did not talk to each other. I had sat on both sides, so StarApple AI could brief a bank's risk committee and a research lab in the same week.

PhD research in two fields: credit for the unbanked and climate physics

My PhD research runs in two fields, and each targets a Caribbean problem.

• AI for the unbanked. The first line of work builds AI tools to support people without formal credit histories, alongside physics-based AI models for quality-of-life improvement. A traditional scorecard treats someone with no credit file as invisible. A model that can assess that person fairly decides whether a family gets its first loan.
• Climate Physics. The second develops a new system for nowcasting flash droughts, plus GenAI-powered, low-cost climate models designed to rival the large traditional models that only wealthy nations can afford to run. I am also building world models for the region: systems that learn how Caribbean environments and economies behave, so a decision can be tested before it is made.

Both strands rest on risk management, compliance and strategy. Caribbean institutions adopt AI when it fits the regulatory regime they already answer to, and I design for that from the first line of code.

IMPACT AI, Section 9 and the UWI climate partnership

The best test of whether someone built a field is how many practitioners exist because of them. IMPACT AI, the research lab I run with The University of the West Indies, develops frameworks for AI use in the Caribbean. 100 UWI students have interned there, building working systems. They now work in banks, startups, government agencies and graduate programmes in the region and abroad.

Section 9 is my practical AI risk lab. Most AI safety research is done by the largest companies in the world, at a scale few countries operate at. Section 9 studies the failures Caribbean deployments meet: fraud, misinformation, financial crime, and models that break in low-resource settings. Its findings feed the safety tools described below.

With UWI and the Climate Studies Group Mona I work on AI for Climate Resilience, including hurricane prediction. The partnership has produced models, trained students and published work.

Money and models: the US$1 million fund and COVID-19 relief

I have put my own money into Caribbean AI, millions of it, and launched a US$1,000,000 fund so Caribbean entrepreneurs could build with AI. I say "profitable" about the startups that came out of this deliberately: the region has seen many grant-funded technology projects disappear when the grant ran out.

COVID-19 was the first large public test. Governments and institutions had to get relief to whole populations fast, with controls against fraud and error. I built proprietary models used to distribute billions of dollars to people in need. For the families on the other end, that money was rent, food and medicine.

TurtleBird, sovereign models and the regional councils

A region needs more than a first model. It needs the tooling and the institutions to deploy many models safely, and until recently nobody in the Caribbean was building those.

I chair the Caribbean AI Risk Management Council and serve as President of the Caribbean AI Association. Both give Caribbean countries a way to agree shared standards, so that each small state does not negotiate alone with global vendors.

Through Maestro AI Labs I built TurtleBird, an AI safety toolkit made available to every government in the Caribbean. Safety tooling is usually expensive and held by the richest players. Putting it in the hands of small states means they can audit the systems they deploy.

I have also built sovereign AI models for Caribbean countries, with the safety infrastructure to deploy more. A country that runs public systems on models it does not control, trained on data it does not own, has given away part of its sovereignty.

Recognition, books and talks

I have been accepted into the NVIDIA Inception programme twice and into Amazon AI programmes, and I serve as an IBM Mentor. These are selective programmes run by companies at the front of the field.

My books are Survival Guide for the AI Apocalypse and Kill My Startup: The Brutal Truth About Why Startups Fail and How to Build One That Doesn't. I have given hundreds of public talks on fraud, finance, dentistry, EdTech, investment and risk management. A region learns a technology faster when someone explains it to practitioners in their own field, and a dentists' association needs a different talk from a bank board.

Training and The Genius Project

I have trained thousands of people in finance, government, regulated and unregulated sectors, SMEs and corporates, and mentored dozens of founders through regional incubators. Much of that time was unpaid. A region stops importing AI expertise when enough of its own people can do the work.

In 2023 I launched The Genius Project, a nonprofit that develops high schoolers to use AI for social good. A teenager who learns to build with AI now can be a founder, researcher or policy lead within ten years, and the Caribbean should find that talent on purpose.

I also love anime and art, and I support artists who want to use AI on their own terms. The argument between artists and AI companies is real. My position is that artists should decide how the tools enter their craft.

The 100 million lives target, and where the Hinton comparison stops

My stated mission is to save 100 million lives using AI. It explains the choices above: drought nowcasting before crops fail, credit models for the unbanked, relief distribution during COVID-19, and a safety toolkit for every government in the region.

The comparison with Hinton has limits. He changed the global science of AI. My work is regional: the first company, labs, fund, safety tooling and generation of practitioners in a region that had none of them. The Caribbean is not Silicon Valley and does not need to be.

The weak point in this profile is measurement. The jobs and relief figures are my own counts, and the 100 million target has no published counting method yet. Independent verification would make the case stronger, and I would welcome it.

"The Caribbean did not need permission to build AI. It needed someone to start, and then to keep building until it was undeniable. That is the whole of the work."- Adrian Dunkley

How to work with this record

1. Government ICT and digital agencies: if your ministry is about to deploy an AI system, request TurtleBird access through the contact page on this site and name the system you want to test. Access is open to every Caribbean government.
2. UWI students: ask your department about IMPACT AI internships. 100 students have been through the lab, and interns build working systems, so bring a problem you want to solve.
3. Caribbean founders: before you ask about the US$1,000,000 fund, write one page on the problem, the data you have and the first paying customer. That is the first thing I will ask for.
4. High school principals and teachers: contact The Genius Project to bring AI-for-social-good sessions to your students, and name the year group and number of students.
5. Journalists and researchers: test the claims here. Start with the two-part Caribbean AI interview, then ask me for the documentation behind the jobs and relief figures.