Explore the phenomenon: Chat GPT

Chat GPT was launched for the public on 30 November 2022, and it conquered the world in no time. The service got 100 million users in two months, whereas for TikTok, a popular platform among young people, this took nine months. Chat GPT is an AI-based service that gives answers to various questions. AI as such is not a new thing. However, Spotify’s recommendations, Google Translate’s translations or Siri’s answers to questions have not stirred up debate on how AI affects everyday life, society and working life. Chat GPT has.

ChatGPT is an example of generative AI, meaning AI solutions that can automatically generate new content such as text in this case. Chat GPT is based on the GPT-3 language model developed by the organisation OpenAI, and this model has been trained by using the vast amount of texts available on the internet. Unlike the chatbots we encounter in online customer service, Chat GPT gives answers to questions on various topics, giving the impression that we are communicating with a person. At the time of writing (on February 2023), one of the restrictions is that the newest information used to teach the language model is from the year 2021, so Chat GPT does not know more recent events.

How can Chat GPT be used? Chat GPT can, for example, pick up the main points from a text, summarise a phenomenon with bullet points, or suggests dishes for a Sunday brunch. Chat GPT can work as a sounding board when we write by providing inspiration for and feedback from our texts. Chat GPT can also help in organising lessons and creating questions for exams (see examples from the list of links below) and act as a tireless study partner who can explain things we do not understand over and over again.

Chat GPT is a language machine which can generate finished text based on our requests and questions. The service can create questions for exams, pick up the main points from a text or answer students’ questions for us – similarly, it can also complete homework and write essays. This naturally stirs up debate on cheating. Going forward, how do we approach written assignments if we cannot be sure about who is the actual author?

At the same time, it is important to keep in mind that Chat GPT is not a reference book or a database. It is a machine that answers our questions by creating new texts based on probabilities. So it is guessing which word would most likely come next. This means that we cannot fully rely on the answers given by Chat GPT, as it can come up with its own “facts”.  Big mistakes are easy to notice but small and subtle errors might appear in interpretations and tones of voice, years and source references. Using Chat GPT requires thorough source criticism. Also asking good questions is important. Previously, we have had to learn to form great search terms when we are searching for information online. Now, we have to learn to ask questions that are precise enough.

The introduction of this kind of language machine also raises other questions and concerns. For example, the technology makes it increasingly easy to automatically create fictitious content, risking an increase in disinformation and propaganda.  However, ChatGPT is not the only example of generative AI; various solutions are now available for automatically generating different types of content, including images, sound and source code. In turn, this has given rise to discussions about how we can learn to use these tools wisely – as humans, we are prone to laziness, and the temptation of having material served on a silver platter may become great. As the solutions are trained using existing data, there is also a risk that the results we obtain are not representative or may reflect prejudices and other biases present in the training data.

Chat GPT might be the first AI solution that really forces us to think what development can actually mean in the future. How can Chat GPT – and new services that will come up in the future – act as parts of a teacher’s toolkit? How can students use them? We need to take a stand on questions on didactics and subject didactics: what, how and why? Meanwhile, AI makes media literacy increasingly important, and requires us to be aware of any negative consequences so that we can tackle them in the best possible way.

Other links

• Chat GPT: chat.openai.com
• AI in Education/AI och skolan (Johan Falk’s eBook in English and Swedish): https://github.com/Itangalo/AI-Education (you can download the book from the link “Ladda hem senaste versionen” at the bottom of the page).
• Example answers: ​​https://docs.google.com/document/d/1g3LytGyVOrzyTcsN__c_ppT9VlM_gAnuah0bjkLBkrg/edit?usp=sharing

Text: Linda Mannila

Image: Siru Tirronen

CC BY 4.0

The media landscape of children and young people is constantly changing and one phenomenon will be followed by another. It is essential to offer students tools to understand and process these phenomena. This learning module is part of the information and assignment material series Paths to new media phenomena. The material series includes the information texts and assignments for the teacher and the students. A meaningful way to get to know new phenomena is to use the How to cover new media literacy phenomena pedagogically? operational model, for example.