Young customers ditch Google for AI, with unknown penalties
School researchers reviewed practically 200 research on how synthetic intelligence impacts youngsters and younger individuals’s means to be taught and assume. They discovered that we all know little or no about how AI use impacts youthful college youngsters.
In temporary
Researchers needed to learn the way AI instruments have an effect on college students’ important considering and problem-solving skills. They analysed 173 cross-border research that had checked out using generative synthetic intelligence (GenAI) at school and schooling.
Findings:
- AI by itself neither strengthens nor weakens these skills.
- The researchers revealed that we nonetheless have restricted understanding of school-age youngsters.
- The research should not thought-about goal, they usually have a powerful regional imbalance.
“It is hanging how little strong information now we have about how AI use impacts primary expertise in youthful pupils. This is problematic, as a result of the affect happens whereas schoolchildren are nonetheless growing,” says Hilde Storrøsæter, PhD analysis fellow at NTNU’s Department of Teacher Education.
Storrøsæter has analysed 173 cross-border analysis articles on using generative synthetic intelligence (machine studying methods the place the objective just isn’t solely to analyse knowledge, but in addition to create new knowledge, for instance within the type of textual content, photos, audio, video or programming code. The most widespread use of generative AI is thru chatbots, comparable to ChatGPT) at school work and schooling. Her research was performed with a gaggle of colleagues from Czech universities.
The objective of the literature evaluate was to search out out whether or not AI instruments strengthen or weaken youngsters and youths’ means to assume critically and resolve issues independently. Many individuals imagine that these expertise have by no means been extra essential than now, within the twenty first century.
The twin nature of AI
“The most essential discovering is that AI in and of itself doesn’t make us any higher or worse at considering. It’s straightforward to imagine that the instruments will both strengthen or weaken studying. We discovered a transparent twin impact. The similar expertise can do each,” mentioned Storrøsæter, who can be a geographer and instructor who focuses on subject-matter pedagogy.
“The research confirms the twin nature of AI. The similar expertise can have each constructive and damaging results. We had been shocked that this got here throughout so clearly,” says Hilde Storrøsæter, a PhD analysis fellow at NTNU’s Department of Teacher Education. Photo: NTNU
AI instruments assist essential expertise when they’re used to problem, ask questions, examine completely different viewpoints and construct one’s personal line of reasoning.
“But when AI turns into a shortcut to a ready-made reply, the possibility of overdependence, extra superficial considering and weakened unbiased drawback fixing will increase,” Storrøsæter says.
Abandoning Google – selecting AI
The Norwegian Media Authority’s survey (in Norwegian) on the media habits of 9 to 18-year-olds exhibits that using Google searches fell from 72% in 2024 to 47% in 2026 for this age group.
The report exhibits that 39% of Norwegian 11 to 12-year-olds use AI.
The web site ung.no – Norway’s on-line data and communication channel for younger individuals aged 13 to twenty – can be dropping visitors as a result of younger individuals are switching to AI instruments to search out solutions to their questions. The variety of visits to ung.no fell from 24 million to 18 million from 2024 to 2025. Google’s new AI summaries have additionally influenced the decline in visitors.
A information hole uncovered
The researchers analysed research from 2022 – the 12 months that AI language fashions swept onto the globe stage – and as much as the spring of 2025. Few predicted the ability and velocity with which expertise would change society. AI impacts how data is created, used, shared and disseminated. The penalties are notably vital in faculties and academic settings.
The schooling sector has already been formed by AI. The query is whether or not we are able to form using AI.
Research on using AI in schooling ought to present course for pedagogy, coverage and analysis within the area to assist authorities and different establishments make choices on AI use.
Storrøsæter and her colleagues uncovered three main gaps in what we learn about AI use.
Knowledge hole 1: Children and youths
In 80% of the research the researchers analysed, the individuals had been college or faculty college students. This implies that we all know surprisingly little in regards to the educational and cognitive results of AI use on youngsters and youths, at a time whereas they’re nonetheless growing how they understand, course of and consider data.
“We can’t merely assume that what applies to a 22-year-old college pupil additionally applies to a 10- or 14-year-old,” says Storrøsæter.
“If AI takes over the considering course of earlier than pupils have had time to develop their very own expertise for important considering and drawback fixing, the implications might be fully completely different for them than for adults,” she says.
You may also like: “We must develop thinkers. Not crammers and fact-experts”
Knowledge hole 2: Lack of objectivity
Most of the articles within the literature evaluate are based mostly on self-reporting. This is taken into account a much less dependable methodology than goal research of modifications over time.
Knowledge hole 3: Geographic bias
The college researchers additionally uncovered a geographically skewed distribution in the place the research had been finished. South Asia, Southeast Asia and East Asia – primarily China – dominate with 28 research. In Europe, Spain and the United Kingdom led with 9 research every.
A complete of 15 research had been performed within the USA. Africa, South America and Central Asia are significantly underrepresented. This creates a analysis hole that must be addressed.
Overall, the research present optimistic results extra usually than detrimental ones.
Mostly optimistic findings:
- Of 139 research on the impact of AI on important considering, 67 reported solely optimistic results of AI use.
- 33 research reported solely detrimental results, whereas 39 reported each optimistic and detrimental results.
- Of 103 research that examined the impact of AI on problem-solving means, 57 had been solely optimistic.
- 23 reported detrimental results, the identical quantity that reported each optimistic and detrimental results.
The twin activity of instructor schooling
The researchers behind the brand new literature evaluate spotlight three essential points specifically: How AI instruments are used, lecturers’ understanding of AI, and the calls for lecturers place on college students and pupils.
“Teacher schooling has a twin activity. We have to coach lecturers who can use and relate critically to AI themselves. At the identical time, our schooling college students must also be capable to train youngsters and younger individuals easy methods to use AI. It just isn’t sufficient to simply train them to make use of some AI instruments,” says Storrøsæter.
For us to succeed with AI in faculties, instructor schooling has to succeed.
“Students want to grasp what expertise does to the educational course of itself, comparable to what occurs when AI solves elements of the duty. Which elements of the considering course of are essential for the pupils to nonetheless do themselves? How can we design duties the place AI turns into a software, and never a alternative, for considering?”
Can create elevated disparities
Storrøsæter just isn’t fearful about college students’ or pupils’ use of AI in and of itself.
“I assist being restrictive by way of not utilizing AI with very younger pupils. But when AI is used, it has to occur on the instructor’s phrases,” she says.
If AI is used uncritically, you don’t construct any expertise with it, and also you miss out on lots of studying conditions. This will solely enhance disparities between those that use AI critically and as a software, and those that use it uncritically merely to finish a activity sooner. If that occurs, faculties and universities will each face vital challenges.
We must form using AI, not the opposite means round
Technology will most likely proceed to develop sooner than we are able to put curricula and laws in place. Our primary activity is due to this fact to domesticate AI literacy, by defending and growing competence in important considering, drawback fixing, supply criticism and the power to make unbiased assessments.
“The schooling sector is already strongly influenced and formed by AI. The query is whether or not we are able to form using AI in order that it helps college students’ considering as an alternative of taking it over.”
Storrøsæter stresses that if we’re to succeed with AI in faculties, instructor schooling has to succeed.
Reference: Tomáš Bendl, Lenka Krajňáková, Hilde Storrøsæter, Emil Drápela, Martin Hanus: Dual Nature of GenAI: Catalyst or Barrier for Critical Thinking and Problem-Solving Skills? A systematic Review, Journal of Information Technology Education: Research • Volume 25 • 2026 • https://doi.org/10.28945/5834
