In 2025, you'll hear lots about artificial intelligence (AI) and education.
The British government presented its “Action plan for AI opportunities“Mid-January. As a part of the plan it’s excellent Funding £1 million (roughly $1.2 million) to 16 education technology firms to “develop AI tools for teachers for feedback and grading, driving high and rising educational standards.” Schools in some US states are Test AI tools of their classrooms. A Moroccan University is the primary company in Africa to introduce an AI-powered learning system across its facility.
And the theme for this 12 months's United Nations International Education DayThe event takes place annually on January twenty fourth and is named “AI and Education: Preserving Human Agency in a World of Automation.”
But what does AI mean on this context? In education, it is commonly used as a catch-all term and is commonly mixed with digital skills, online learning platforms, software development and even basic digital automation.
This mischaracterization can distort perceptions and obscure the true potential and importance of AI-driven technologies. These technologies were developed by scientists and experts in the sector and dropped at scale by large technology firms. For many individuals, the term AI brings to mind systems like OpenAI ChatGPTwho is ready to put in writing essays or answer complex questions. However, AI’s capabilities extend far beyond these applications – and every has unique implications for education.
I’m an authority within the areas of AI, machine learning, infodemiology – where I examine large amounts of knowledge using AI to combat misinformation – knowledge mapping (discovering and visualizing the content of various areas of data) and human language technology (constructing) models that use AI to To help people advance language, similar to live translation tools. I do all of it as Head of the Knowledge Mapping Lab, a research group throughout the Faculty of Economics and Management Sciences, and co-director of the Interdisciplinary center for the digital future on University of the Free State.
In this text, I explain the technologies and science behind the buzzwords to make clear what terms like machine learning and deep learning mean in education, how such technologies may be or are already getting used in education, and their advantages and pitfalls have.
Machine Learning: Personalization in Action
Machine learning is a subset of AI that involves algorithms that learn from data to make predictions or decisions. In education, this may be used to tailor content to individual learners – so-called adaptive learning platforms. For example, they’ll assess students' strengths and weaknesses and adapt lessons to their pace and elegance.
Imagine a math app that asks questions based on the curriculum, then uses a learner's answers to discover where they’re struggling and adjusts the curriculum to deal with those Basic knowledge before proceeding. Although scientific research is ongoing, this level of personalization could improve educational outcomes.
Deep Learning: Assessment and Accessibility
Deep learning is a branch of machine learning. It mimics the human brain through neural networks, enabling more complex tasks similar to image and speech recognition. In education, this technology has opened up recent possibilities for assessment and accessibility.
AI-controlled tools are utilized in the assessment may help E.g. mark-up, analyze handwritten tasks, assess speech patterns in language learning or translate content into multiple languages in real time. Such technologies may help teachers reduce their administrative burden while contributing to the training process.
Then there may be inclusivity. Speech to text and text-to-speech applications enable students with disabilities to have interaction with material in ways not previously possible.
Natural language processing: beyond ChatGPT
Natural language processing is a branch of AI that permits computers to assist understand, interpret, and generate human language. ChatGPT is essentially the most well-known example, however it is just one among many such applications.
The educational potential of the sector is large.
Natural language processing may be used to:
-
Analyze student writing for mood and elegance to supply real-time feedback on mindset, tone, and quality of writing. This goes beyond syntax and semantics
-
Recognize plagiarism
-
Give learners feedback before class, which deepens classroom discussions
-
Summarize papers
-
Translate complex texts into more comprehensible formats.
Reinforcement Learning: Simulating education and making it playful
Write/Getty Images
With reinforcement learning, computer systems learn through trial and error.
That is particularly promising in playful educational environments. These are platforms that apply the principles of gamification and education in a virtual world that students “play through.” She learn playfully. Over time, the system learns to adapt to make the content more difficult based on what the coed has already learned.
challenges
Of course, these technologies usually are not without flaws and ethical issues. For example, they raise questions on equity: What happens if students without access to such tools fall even further behind? How can algorithms be prevented from reinforcing biases already present in educational data? In the sooner math example, this will not be such an enormous problem – but imagine the unintended consequences of reinforcing bias in subjects like history.
Accuracy and fairness are also essential concerns. A poorly designed model could misinterpret accents or dialects and drawback certain groups of learners.
An over-reliance on such tools could also result in a decline in critical pondering skills amongst each students and teachers. How can we find the suitable balance between assistance and autonomy?
And from an ethical perspective, what if AI was allowed to trace and adapt to a student's emotional state? How can we be certain that the information collected in such systems is stored? be used responsibly and safely?
Experiment
The potential of AI have to be explored through experiments. However, this works best when managed in controlled environments. One option to achieve this is thru regulatory AI “sandboxes” – spaces where educators and designers can experiment with recent tools and explore applications.
This approach has been employed on the University of the Free State since 2023. As a part of the Interdisciplinary Center for the Digital Future, the sandboxes function open educational resourcesprovides videos, guides, and tools to assist educators and institutional leaders understand and responsibly implement AI technologies. The resource is open to each students and teachers of the university, but additionally ours Main focus is about improving the talents of educators.
AI in education is here to remain. If its components are properly understood and its implementation is driven by good research and experimentation, it has the potential to enhance learning while continuing to make education human-centred, inclusive and empowering.
