The article goes onto explain how applications are already being used, usch as tutoring systems and learning platforms. Tools, such as these can promote a more personalized and engaging educational experience.

Lastly, the articles pictures AI supporting education and learning for the foreseeable future. AI can provide real-time formative assessment tools and reduce gaps in education. As the articles does speak in detail, AI does need oversight. This would come from the teachers of the content.

One of the limitations of AI is it does not model social context or human reasoning. This shows that human teaching and interaction are still necessary for students. The articles raises concerns about social harms. For example, access to some AI could reinforce the privilege some students have over others. Some AI could pose an advantage for some groups while marginalizing others. The author makes sure to include that there is a bias with AI and students need to be aware of these issues.

Lastly, Selwyn states that AI use could become political. This distribution will need oversight, which can lead to someone or some entity having power and responsibility. The author urges research to look beyond what AI can do and look at AI in education in regard to social justice, ethics, and well-being. It should be used cautiously.

The authors categorize the AI used based on their instructional design. AI in education can be categorized into the following, personalization, tutoring, assessment, and feedback. This framework can help educators understand what AI can do and how it can align with their specific objectives and practices.

In essence, the article calls for a sober, critical approach: grounding expectations in current realities, unpacking the assumptions embedded in AI systems, and thoughtfully addressing the ethical, pedagogical, and structural challenges to embed AI in education in a meaningful way.