Talk
Can AI Learn From Data It Cannot See?
Synthetic data, privacy, and responsible innovation in African fintech
Building and testing AI needs customer data, and privacy rules, rightly, limit who can touch it. This leaves a gap between what institutions want to build and what they may learn from.
One path is synthetic data, generated by AI in place of the real thing: to train and test a model without touching real customer data, to work with a partner without handing over customers, to develop in-house before anything goes live.
Synthetic data has limits: it carries the patterns, and the blind spots, of the data it came from, and a system tested on it still has to prove itself on real customers. The keynote looks at what institutions and regulators need to judge when it is good enough.
The closing section turns to the choices: what to share, with whom, under which safeguards, and who decides. What an institution may learn from sets the limit of what it can build.
Questions this talk answers
- 01What gap opens for institutions building and testing AI when privacy rules, rightly, limit who may touch customer data?
- 02Where does synthetic data help, and where does it fall short?
- 03Who decides what is shared, with whom, and under which safeguards?
Want this talk at your event?
Send your date, your audience and the format you have in mind. The talk will be adjusted to your room.
Invite Hamid with this talk