The researchers, who carried out the study in Qatar, recommend that AI to be taught in primary schools – and that teachers show students how to use AI tools, to help with their schoolwork.

“AI is rapidly making its mark in almost every sector of the economy and will continue to increase its influence, making it vital that young people are equipped with the skills to thrive in an AI-driven future,” says researcher Dr. Zubair Ahmad, of Qatar University Young Scientists Center (QUYSC).

“However, previous research has shown that students often struggle to master the fundamentals of AI concepts. We wanted to know why.

“We found that students who believe in their ability to learn and use AI tend to do better in the subject. This link between confidence and results is much stronger in boys than in girls.

“Good teaching and access to resources are also important, although their impact was stronger for boys than for girls.

“By building girls’ belief in their ability to master AI, schools can help ensure that both genders do well in the subject and help prepare the next generation for life in a world that is being rapidly transformed by AI.”

For the study, published today in peer-reviewed journal Cogent Education, Dr Ahmad and colleagues developed a 35-question survey that explored the relationship between students’ confidence and belief in their ability to learn and use AI (AI efficacy), how much they had learnt (AI learning outcomes) and how much support they got from their teacher and school (institutional support).

The questionnaire was completed by 743, 15- to 18-year-olds who were studying computing and IT at high school in Qatar. Participants comprise both Qatari nationals and students from diverse Asian and African backgrounds.

Analysis of their answers showed that students who were more confident at AI did better in the subject, particularly if they were boys.

Students who received higher levels of institutional support – including teacher guidance, hands-on learning experiences, and access to educational resources – demonstrated better AI learning outcomes, with this effect being significant for male students but not for female students.

It is thought that students who are more confident at AI are more likely to persevere when the topic gets difficult, while those with a belief in their abilities may shy away from tricky areas or give up.

Good teaching, meanwhile, can inspire and motivate and so help turn confidence into results.

Dr Ahmad says there are several possible reasons why the links between confidence, institutional support, and learning outcomes were stronger in boys than in girls.

“Technology and AI are often perceived as a male-dominated domain which can influence students’ belief in their abilities and their engagement in the subject,” he explains.

“And, as a result female students may have lower confidence in their abilities or be less likely to experiment with AI tools.

“The teaching style may matter, too. We know, for example, that some students prefer very structured lessons, others thrive on the combination of gentle guidance and the freedom to explore.”

The study’s findings can be used to amend curricula, to boost students’ AI self-efficacy and their AI skills, say the authors.

They suggest that:

• Students should be taught the basics of AI from early stages of school education. More complex concepts can be introduced progressively across higher grade levels.
• Lessons should be interactive, with students asking questions, doing hands-on work, and solving real-world problems, rather than passively listening.
• Teachers should give students feedback soon after they complete a task, while it is still fresh in the students’ minds.
• Educators should also teach students about how to use AI ethically. This could include showing them how to use AI tools to help with their schoolwork without cheating or being dishonest.

“Schools should also do more to support girls specifically,” says Dr Ahmad. “This could be done by providing them with more female role models in AI, creating a classroom environment in which all students feel comfortable, so that girls perceive that they are equally supported, which will boost their belief in their abilities to engage with AI.

“One way of doing this is through what we call guided practice. An example would be a teacher demonstrating how to use an AI tool, then allowing students to practise, while giving them guidance when needed. The teacher then gradually reduces the level of support as the students become more proficient. Such approaches will ensure that students, particularly girls, feel supported throughout the learning process.

“This builds skills, while also boosting confidence.”

A limitation of the study, includes that it has “conceptualized institutional support with a broader coverage of the AI learning aspect”.

The authors recommend future research may investigate the influence of various aspects of AI learning individually.