Many teachers already have access to a range of AI tools to enhance teaching and learning, and to prepare students for a world shaped by AI. A huge number of tried and tested AI tools for use in the classroom can be found in this list of AI Tools and Technologies across the curriculum, crowdsourced by the participants of the EU Code Week AI Basics for Schools MOOC.

AI applications such as language learning apps, language translators, math helpers, tools for automatic transcription and subtitling or digital assistants that offer customised learning experiences are already widely used to accelerate personalised learning. AI has also shown great potential in supporting students with special needs. AI-driven solutions might fundamentally transform assessment practices by providing students with in-depth assessment and timely and focused feedback. Effective use of learning analytics enables teachers to gain a deeper insight into how their students are learning, what problems they are facing, how motivated they are, how they are feeling and how they respond to a learning situation to select appropriate teaching methods and differentiate the learning process.

Nonetheless, poor design, improper use and negative consequences of AI systems can cause irreparable harm, especially to young people. I give you two examples, related to disinformation and algorithmic bias:

Rapid advances in AI have accelerated the production of synthetic media, colloquially known as deepfakes. Deepfakes refer to algorithmic generation, manipulation and modification of audio tracks, videos, images and text for the purpose of misleading people or changing its original meaning. This technology may seem advanced and, as such, out of reach for students, but it is far from being inaccessible. For example, TikTok users can use free apps, which allow for fast and easy face swapping in videos and photos, thus spreading fake media and causing harm to their peers. I strongly believe that raising awareness of fabricated media and learning how to critically analyse the content students create and consume is nowadays more essential than ever. I invite you to check out the website entitled Which face is real, an interesting project developed to raise awareness of deepfakes and how to spot them at a single glance.

In my opinion, one of the most relevant ethical concerns that AI has raised is algorithmic bias. It refers to errors that create unfair outcomes, such as discrimination on the grounds of gender, race, ethnicity or socio-economic background. It is driven by the quality and representativeness of data, intentional or unintentional biases of humans who design AI systems and the way these AI systems are developed and deployed. An example of gender bias is a language translator making assumptions that doctors and pilots are male, while nurses and flight attendants are female. Another example is deliberately adding racist or sexist language to a chatbot so that it communicates in a disrespectful, rude and offensive way.

It is still not clear what happens in the AI ‘Black Box’ and why ‘invisible’ algorithms make certain decisions that can have a tremendously negative impact on young people, their education and consequently on their future life opportunities. The AI decision-making process needs to be transparent and explainable. Unbiased and fair decisions need to be guaranteed for all students equally. A critical approach to understanding how AI works plays a significant role in raising awareness of algorithmic bias and increasing AI’s accountability, transparency and fairness.

Arjana Blazic is a teacher trainer and instructional designer. She is a co-author of the Croatian National Curricula for English Language Teaching and the Use of ICT as a Cross-Curricular Topic. She works as an external expert for EU Code Week where she develops educational resources and teacher training opportunities.