CHRE has concerns about the potential misuse of generative AI (e.g., large language model chatbots such as ChatGPT and Claude).
We are especially concerned about the use of generative AI in abusive and neglectful settings, where it may exacerbate the effects of existing poor treatment. That said, we recognize that for some homeschooled children in abusive/neglectful settings, AI may be their only interaction with instruction or with even simulated empathy. Caregivers should encourage children to be aware that AI lacks both the normal friction of human relationships and the ability to provide appropriate socialization, and encourage them to establish relationships with trustworthy peers and mentors when possible.
Despite the existence of abusive and neglectful homeschooling caregivers, we believe there are significant numbers of caregivers who seek to provide their homeschooled children with appropriate education and socialization. Below we provide recommendations for those caregivers.
Summary of recommendations
| ❌️Don’t | ✅️Do |
| ❌️Don’t substitute AI for human judgment, parenting, and teaching. | ✅️Use your judgment, in consultation with experts, to plan and teach. Learn the tells of AI-generated materials and avoid their use. |
| ❌️Don’t let a child’s most rewarding option be to submit AI-generated work in lieu of their own. | ✅️Know what work your child is submitting and whether their work is consistent with their understanding of the subject matter. Ensure that it is more rewarding for your child to do their own work than to submit AI-generated work. |
| ❌️Don’t punish your child for submitting AI-generated work. | ✅️Try to understand why your child submitted AI-generated work and address the cause appropriately. |
| ❌️Don’t avoid the topic of AI entirely. | ✅️Provide homeschooled children with education about generative and other forms of AI as part of a broader technology and media literacy education. |
| ❌️Don’t allow AI relationships to substitute for peer and mentor relationships. | ✅️Ensure children have access to a variety of peer and mentor relationships outside the home and have appropriate support for dealing with issues. |
| ❌️Don’t allow children to use AI for lengthy periods of time, especially in lieu of other normal life activities, such as education, socialization, and sleep. | ✅️Ensure children’s access to AI is limited, reasonable, and purposeful. |
Recommendations
Don’t substitute AI for human judgment, parenting, and teaching. Instead, use your own judgment in consultation with experts. Learn the tells of AI-generated materials and avoid their use.
Substituting AI for parenting and teaching removes a major source of social development for children and can substitute potentially unevaluated AI output for professionally evaluated teaching materials. Reliance on AI can also erode performance and persistence, meaning that caregivers who rely on AI can lose important skills of their own.
Instead, use your own judgment, in consultation with experts, to choose and implement appropriate curricula, learning materials, and assessments. If you don’t have the time, energy, or expertise to do this, consider that homeschooling may not be the right option for you at this time.
We also recommend learning common tells of AI-generated materials so that you can identify and avoid them even when they are not labeled as AI.
Don’t let a child’s most rewarding option be to submit AI-generated work in lieu of their own. Instead, know what work your child is submitting and whether their work is consistent with their understanding of the subject matter. Ensure that it is more rewarding for your child to do their own work than to submit AI-generated work.
Substituting AI-generated work for self-generated work can harm learning.
To catch AI-generated work, be aware of what work your child is submitting and aware of your child’s level of knowledge. Their work should be consistent with their general understanding of the subject and their general skill level.
Ensure doing their own work is more rewarding than submitting AI-generated work.
Don’t punish your child for submitting AI-generated work. Instead, try to understand why they submitted AI-generated work and address the cause appropriately.
Punishing children can set up an antagonistic situation without addressing the root cause.
Instead, if your child submits AI-generated work, try to figure out why. For example: did they not understand the material? Were they afraid of disappointing you? Do they not yet have the self-regulation skills necessary to defer gratification? Then, try to address the cause appropriately.
Don’t avoid the topic of AI entirely. Instead, provide homeschooled children with education about generative and other forms of AI as part of a broader technology and media literacy education.
Avoiding the topic of AI entirely prevents children from gaining a realistic picture of AI’s uses and limitations. It also prevents them from developing the skills to help detect AI-generated content.
Instead, provide children with a technology and media literacy education that includes generative and other forms of AI, including understanding AI’s drawbacks and shortcomings.
Don’t allow AI relationships to substitute for peer and mentor relationships. Instead, ensure children have access to a variety of peer and mentor relationships outside the home and have appropriate support for dealing with issues.
Allowing AI relationships to substitute for peer or mentor relationships risks depriving children of learning to handle the friction and variability that are part of normal human relationships.
Instead, ensure children have access to a wide variety of peer and mentor relationships outside the home, and have appropriate support for dealing with issues that may arise. Additionally, ensure children are aware of relevant issues with AI such as AI sycophancy.
Don’t allow children to use AI for lengthy periods of time, especially in lieu of other normal life activities, such as education, socialization, and sleep. Ensure children’s access to AI is limited, reasonable, and purposeful.
Allowing AI use to crowd out normal life activities risks negative mental health outcomes.
Instead, ensure that if children are using AI, they are using it briefly and reasonably.
###