Generative artificial intelligence (AI) allows generating resources (text, images, audio, video, source code,…) according to the textual instructions provided by the user. Thus, answers to questions or document summaries can be requested, among other tasks.

Factors such as its effectiveness, versatility, availability and ease of use have increased its popularity. In addition, technology companies such as Microsoft or Google are integrating generative AI into products such as web browsers or office tools.

What role can artificial intelligence play for teachers?

There are different ways to introduce this technology in teaching. First of all, we can focus on it as an educational assistant: use it to improve and speed up learning. We can think about improvements for teachers, students or both. For example:

1. Generate educational resources through AI, whether they are teaching materials, evaluation activities or interactive scenarios.
2. Customize content. This includes textually describing graphic resources (photographs, diagrams…) for people with vision problems; adapt resources to the student’s level of prior knowledge (for example, adding or deleting examples or definitions), and adapt the examples of the materials (names of people, locations…) to the context of a given country.
3. Resolve students’ doubts in the form of an interactive tutor. AI-based tutors allow you to ask questions about a concept or exercise, receive an immediate response and ask for further clarification. This type of artificial intelligence tools for teachers supports the student when they are working autonomously, outside the context of a class.
4. Generate and refine ideas, giving support to the student. Among other things, AI can review texts and propose ways to improve their clarity and readability. It can also be used as a brainstorming tool , for example to propose ideas for an advertising campaign.
5. Search and synthesize large amounts of information.
6. Improve the assessment process and the feedback and evaluation provided to the student. For example, based on the teacher’s rating and evaluation criteria, the AI ​​can expand the teacher’s feedback with specific examples of errors or aspects of improvement in the student’s delivery.

What do students need to learn to be able to use it?

It is also necessary to train students so that they know how to use artificial intelligence effectively, ethically and critically. From this perspective, the teacher can offer them:

1. Resources on AI: what tools exist , tips for writing instructions, ethical aspects of their use…
2. Carrying out activities where AI is used to solve a problem, to learn how to use AI practically.
3. Critical evaluations of the responses generated by an AI, to understand its capabilities and limitations.

Finally, we can focus on applying AI as a lever to improve processes in educational centers: improve documentation management, simplify and automate procedures or facilitate access to information by students and teachers.

Artificial intelligence for teachers: choosing the tools

One of the main challenges of generative AI in education is the rapid evolution of technology. New tools are continually appearing to solve different tasks: chatbots, programming assistants or image, audio and video generation tools.

When choosing a specific tool or service, it is important to assess aspects such as functionalities, cost, integration with the center’s educational platform or data protection. In this sense, certain decisions may make more sense when taken at the level of an entire educational center than at the level of a specific subject.

What should we take into account when we apply it?

To make correct use of artificial intelligence by teachers, it is necessary to take into account the following issues:

1. The objective must be clear. Since generative artificial intelligence carries certain risks, it is not advisable to use it indiscriminately. That is, the objective to be achieved must be defined a priori.
2. Human supervision is essential. Generative artificial intelligence models can make many types of errors. For example, misinterpreting the request received, inventing data , giving biased answers or applying reasoning incorrectly. Furthermore, such errors can be presented convincingly, with well-written writing and arguments (apparently) in favor. For this reason, students must be made aware of the need to review the answers generated by artificial intelligence. On the other hand, teachers must review any text generated by this technology, and not delegate critical processes such as evaluation to it.
3. Instructions and data are the key . To use generative artificial intelligence effectively as teachers, you must give precise instructions and provide the information necessary to generate a response. Selecting the instructions and knowledge base to use is essential to ensure success.
4. Privacy and confidentiality must be ensured . It is necessary to ensure that the data supplied to an AI will not be used for other purposes. For example, many tools allow you to indicate that we do not share our data to train other models. We should use this option whenever possible and avoid sending confidential, private or copyrighted content to artificial intelligence services.
5. AI has a cost associated with it , even when it’s free. Many tools have a financial cost, whether it is pay per use or a monthly fee. In addition, certain tools can also be adapted to specific environments or problems, for an additional cost. Therefore, in certain contexts, the cost of certain tools or services may be prohibitive. On the other hand, in the case of free services, we must understand what companies gain by allowing their use. In many cases, companies reserve the right to use the data used to train other artificial intelligence models.
6. Students can use it too. Generative AI tools are available to anyone, without requiring technical knowledge and in many cases for free. This means that activities that are not performed in a controlled environment can be used. Although there are services to detect AI-generated texts, they are not 100% effective nor do they provide evidence to confirm suspected cases. This means that we must look for other types of strategies: tests in controlled environments, interviews with students, or activities where the use of AI does not affect learning.

Author Bio: Robert Clarisó Viladrosa is an Added Professor. Principal researcher of the “Systems, Software and Models Lab” group at the Internet Interdisciplinary Institute (IN3) at UOC – Universitat Oberta de Catalunya