What are our data really worth when AI consumes them?
Confidentiality, data, digital privacy
The unsettling question
The more you give AI, the more useful it becomes. That is how it works. The more useful it becomes, the more you give it.
At the end of the road, your histories, emotions, voices, faces, work conversations, intimate questions and 2 a.m. doubts all feed systems you do not control. The contract is implicit: your privacy in exchange for a little magic.
And nobody ever reads the privacy policy.
The starting point
AI models train on mountains of data. Their interfaces continually collect what we tell them to improve, personalise or train future versions. The boundary between “the tool I use” and “the tool that uses me” has become blurred.
The question is no longer whether we share data: spoiler alert, we do it constantly! It is which data, with whom, for what purpose, for how long and under whose control. We also need to recognise the imbalance: they know everything they hold about us; we know almost nothing.
There is another unsettling effect: today’s data will train tomorrow’s AI systems, which will then talk to strangers. A message you write in confidence now could indirectly shape a model’s personality five years from now.
Tomorrow
Tomorrow, AI will be in healthcare, education, work, the home, the car, on your wrist and even in a chip under your skin. Every conversation, click and heartbeat could become input for a model.
Privacy will need defending more than ever, even if you think you have nothing to hide.
Possible angles
Reading to get started
GDPR and CNIL guides on AI and personal data
To understand the legal framework in force and individuals’ rights in relation to AI systems.
Shoshana Zuboff — The Age of Surveillance Capitalism
A demanding but essential book. Read at least a summary to understand how personal data became the raw material for a new form of capitalism.
Ann Cavoukian — Privacy by Design
An approach that puts privacy before product design, not after it.
Articles on differential privacy and federated learning
To discover techniques for training AI without collecting everything.
MyData / NOYB
To discover organisations actively defending digital rights in Europe.
Try it yourself
- Read the entire privacy policy of the AI tool you use most. Note what surprises you.
- Ask an AI assistant: “What do you know about me? Do you keep a record of our conversations?” Compare its answer with its official policy.
- Try deleting your data from an AI service. Measure how long it takes and how many steps are involved.
- Compare what you told someone close to you this week with what you wrote to a chatbot. Is the boundary where you thought it was?
- List everything your everyday apps know about you. Be honest.
Which context would you anchor it in?
The issue of personal data takes on different significance in healthcare, learning and work.
Higher education
Student monitoring, learning analytics, progress and behavioural data
Business
Internal AI assistants, employee monitoring, customer data and contracts
Creative industries
Rights over works used for training, audience data, consumer profiles