Children, technology & education
A child may ask a voice assistant about homework, health or a family problem. A fluent answer sounds personal and confident, but it does not arise from a teacher–student relationship. It is the output of a system whose knowledge, rules, memory and handling of recordings vary by product and setting.
1. What can be substantiated
In its guidance on generative AI in education, UNESCO emphasizes privacy protection, age appropriateness, human oversight and pedagogical validation. The guidance is not proof that a particular application is safe, but a framework for assessing it.[1]
The GDPR requires lawfulness, transparency, purpose limitation and data minimization. For information society services offered directly to a child, Article 8 governs the conditions for consent; the specific lawful basis, however, depends on the service and circumstances.[2]
Whether speech is transcribed locally, sent to the provider, stored in the history or used for other purposes cannot be determined from the label “voice assistant”. The technical and contractual terms governing the particular account are decisive.[1][2]
2. How to interpret the claims in context
A child may not distinguish between confident language and verified truth. An adult should therefore do more than set parental controls; they should also demonstrate how to ask a follow-up question, verify a source and recognize a topic that needs to be brought to a person.
Privacy does not begin with a ban on speaking. It begins with the rule that an assistant should not receive classmates' names, health details, an address, passwords or another person's secrets. Both parents and schools must know the same rule.
The educational benefit must be measured by what the child independently understands and can explain after the conversation. Quickly completed homework without understanding may look like help, but it may merely outsource the thinking to the service.
- A generative assistant can produce plausible-sounding errors.
- Children's data are subject to enhanced protection and transparency.
- The technical handling of speech is a feature of the particular service.
- The long-term effect of different usage patterns on a particular child's learning.
- Exactly how a particular product will process a recording without reviewing its current terms.
- Whether parental settings cover every connected service and account.
3. Five questions to ask
- What is the recommended age, and what legal rules apply to the service?
- Where is the speech or transcript processed, and how long is it retained?
- Can the child verify the answer and explain it in their own words?
- Which topics must always be taken over by an adult?
- Can the history be deleted and further use of the data disabled?
4. Conclusion
A voice assistant can be a tool, not a substitute for an adult's relationship, judgment and responsibility. The safe question is not only “what can it do?”, but also “what will the child take away, what will the service store, and who will correct an error?”
— Jiný Kontext