The American Psychological Association “Tests” Educational Technology

A child advances to the next level in a math app. A teenager finishes an assignment with the help of a chatbot. The platform reports that almost all the exercises are complete. Everything seems to be working, but these results leave one question unanswered: What have the students actually learned?

That is the central theme addressed in the Expert Report on Children’s and Adolescents’ Learning With Educational Technology (APA Expert Report on Children’s and Adolescents’ Learning With Educational Technology), published in September 2026 by the American Psychological Association. The document reviews the available evidence on the use of these tools among children and adolescents aged 5 to 18. In its conclusions, it avoids a blanket verdict: their effects depend on what students do with them, how adults intervene, and the activities they end up replacing.

The report uses the concept of educational technology in a broad sense. That is to say: it includes products created for teaching (reading apps or smart tutors), general-purpose tools used in the classroom (search engines, word processors, and chatbots), and the infrastructure that enables their use, ranging from devices to learning management platforms. Each category requires a different evaluation based on its distinct nature: a math app will need to demonstrate that it improves that skill; however, for a chatbot, the results will depend on how it is used for educational purposes.

For the APA, learning occurs when an experience results in a relatively lasting change in a student’s knowledge or skills. Time spent using the tool, the number of activities completed, or the ease with which a student progresses within an app provide information about their engagement, but the decisive test comes later: when they can recall what they’ve studied, explain it in their own words, and apply it to a different situation without the tool’s prompts or assistance.

Based on this distinction between performing well during an activity and learning to apply what has been learned in the future, the report outlines its ten recommendations for teachers, families, government agencies, and developers.

Define what you want to teach

A tool should be evaluated based on the specific competency it practices. If it works on basic arithmetic, it will need to show that the student can then solve similar problems outside the application. Having learned to complete its exercises or follow its hints shows mastery of the program, but not necessarily of the content. The APA calls for special caution with claims of improving general abilities such as intelligence or critical thinking: practicing a specific activity rarely produces gains, on its own, in such broad skills. Those effects can only be considered valid if they also appear in different tasks and contexts.

Measure learning, not just activity

Minutes logged in, exercises completed, levels passed or user satisfaction show how much a platform has been used, but say little about what has been learned. The report recommends checking whether the student can recall, explain and apply the content after closing the application. If they can only complete the task correctly when they have its hints and aids available, they may have learned to use the tool, not the subject matter it was meant to teach.

Notice what screen time is replacing

Counting hours is not enough, because a screen hosts very different activities. What matters is what the student does, when, with whom and what they stop doing instead. Technology can displace sleep, reading, physical activity, conversation or hands-on experiences. Among younger children, play and exploration deserve special protection. In adolescence, it is rest and in-person relationships. Even a good application loses value if it takes up the wrong moment.

Choose tools based on how we learn

Technological novelty contributes little if it ignores what we know about memory and practice. The APA advises looking for tools that require retrieving information without looking it up, revisiting content at different times, alternating types of problems and explaining mistakes. Intelligent tutors can help when they offer hints and adjust difficulty without solving the task for the student. The criterion is simple: the tool should keep the student thinking.

Reduce stimuli and distractions

Points, badges, streaks, sounds and animations can prolong the use of an application while diverting attention away from the content. The same problem arises when the student switches between tabs, checks messages or responds to notifications. Multitasking is linked to poorer comprehension and weaker recall, although students tend to underestimate that cost. The report recommends simple designs and devices configured to limit interruptions.

Preserve intellectual effort in the face of artificial intelligence

Artificial intelligence can improve an answer or a piece of work without producing equivalent learning. When it generates ideas, explanations or solutions, it also makes it harder to know what the student actually understands. The recommendation is to use it after a first attempt of one’s own: compare both answers, identify mistakes and revise the work. The APA also calls for keeping tasks that allow knowledge to be demonstrated without assistance, and for extreme caution with chatbots that encourage trust, emotional dependence or the disclosure of personal information.

Maintain adult guidance

Children and adolescents learn more with technology when an adult helps them identify the important ideas and relate them to other experiences. This involvement changes with age. With younger children, it can mean sharing the activity; with older ones, asking them to explain what they have learned, check the information and justify their decisions. The tool should reinforce the educational relationship and the teacher’s judgment, not replace them.

Find out who benefits and who is left behind

A positive result on average can hide the fact that some students face barriers. Language, disability, reading level, cultural background, the device available or family support all affect how a tool is used. Features such as subtitles, dictation or text-to-speech can improve access, but each product must be tested with diverse students. It is also worth noting who needs constant help, who avoids the tool and who gives up before reaching the content.

Protect students’ data

Data makes it possible to detect difficulties and adapt teaching, but it can also be used to build profiles, show advertising or develop products. Each piece of information collected should have a clear educational purpose. Families, teachers and students need to know what is stored, who can access it and when it will be deleted. Free applications deserve special attention when their business model depends on data. The recommendation is to collect only what is strictly necessary.

Plan for everything that comes after the purchase

Acquiring licenses or distributing devices is only the beginning. Teachers need ongoing training, time to plan and technical support. Schools also need to decide how they will measure results and when they will review the tool. Many studies show that a product works better than standard teaching, but not that it is superior to other effective practices. Evaluation must accompany the entire implementation and allow course corrections when learning does not improve.

The test continues

These ten recommendations shift the focus from the novelty of the tool to the student’s experience. An application may be useful for practicing one competency and poorly suited to developing another; a platform may make learning easier for some students and create barriers for others; artificial intelligence may explain a concept or prevent the student from ever formulating their own explanation.

That is why educational technology never receives a final grade. Its test is repeated with every use, every context and every student. The decisive test comes when the supports disappear and the student has to remember, explain or solve something on their own. What they are able to do at that point shows how much the technology really helped them learn.