On Thursday, ChatGPT came out a year ago; before that, very few people knew anything about language models. We have compiled several articles examining what generative AI means for teaching.
New text field for written tests in German, English, and French
From May 2024, a new text field will be introduced for the written German, English, and French tests in grades 9 and 10, with mandatory use from the 2024/2025 school year. This new text box replaces traditional word processors like Word and Google Docs and includes no spelling and grammar checking. The goal is to ensure a more equal and fair trial process while minimizing cheating with word suggestion features, ChatGPT, and translation programs.
Students can continue to use dictionaries and grammatical summaries as aids, and schools can voluntarily sign up for the new format as early as May 2024. Registration for this voluntary phase requires the school to participate in all three foreign languages and comply with the latest test conditions. The Danish Agency for Education and Quality prepares new test guidelines, and the assessment criteria are adapted to the new conditions.
Schools not participating in the voluntary phase must use the text box from the winter term 2024. Continuously updated guidance material will be published on the ministry's website towards the school year 2024/2025.
Integrative technology understandings for a sustainable world
Mikala Hansbøl, PhD, has written an article about integrating technology understanding into the education system. She argues that educational institutions should focus on a holistic understanding of technology beyond traditional tool and instrument understandings. She criticizes both technology evangelism and pessimism, as these views overlook the complexity of technology and its impact on society.
Hansbøl highlights how digital technologies such as ChatGPT and AI affect our understanding of the world and stresses the importance of critical thinking and accountability in using these technologies. She points to the need for integrative understandings of technology that recognize the many facets of technology, including its impact on climate and culture.
The article calls for primary and secondary education to reflect on how technology can influence and shape our world. Hansbøl emphasizes that a deep, diverse understanding of technology is crucial to creating a sustainable and democratic future.
In the last week, many articles have been about a project at OpenAI called Q* - pronounced Q-Star, but we need to learn more about it. Reportedly, the project was why Sam Altman was fired in the first place, but now he has returned as CEO of OpenAI. What is interesting, however, is what Q* is; many guess they are approaching AGI (Artificial General Intelligence). In this state, AI can understand, learn, and apply knowledge across various tasks at a level similar to human intelligence. According to Reuters, the Q* program has been able to perform math at the elementary school level.
Mathematics is seen as a big challenge for the language models, and Q* had to be able to solve mathematical tests that were outside its training data - and without using plugins.
However, it is essential to note that total AGI is still a long way to go.
Artificial intelligence affects all disciplines in our education system, and we see many creative opportunities to use the technologies. Some time ago, Dave Villalva released a take on a Star Wars trailer, which is 100% generated by AI and shows how far technology has come. It may not be long before students use it in media productions, presentations, or other multimedia assignments.
Below, we select articles that have a slightly more scientific perspective. These are articles that we read ourselves to keep ourselves updated, and at the same time, we also know that they are not so interesting for most people.
LEDITS++
Researchers have developed a new method of image editing called LEDITS++. The exciting thing about this method is that the model can mask relevant areas in the images, thus replacing them with the text the user has written. At the same time, the model preserves the overall image composition and objects in the picture.
Eksempler hvor modellen har tilføjet elementer til billederne, men stadig bevaret den originale billedekomposition.
In graphic production subjects, we will see a shift in how you work with digital image editing programs, which opens up for experimentation with complex and creative image edits. If you can write what you would like to change in an image, artificial intelligence can also perform it in pictures.
To work and research in artificial intelligence and machine learning, SDU eScience Center has purchased 16 NVIDIA H100 GPUs. The 16 H100 GPUs will be made available via the national HPC service, DeiC Interactive HPC, run by a consortium of universities, including Aarhus University, Aalborg University, and SDU.