📋 Tender Overview
Artificial intelligence (AI) is helping people do more things faster and better. But better results dont necessarily mean more learning. As AI participates more and more deeply in the thinking process, a new question arises: What are the real abilities that belong to the learner and what do people still need to learn, remember, think and practice on their own? Download Image created by AIThe Government Electronic Newspaper had discussions with international experts in education, learning science and AI, to clarify the issue of how to leverage AI to help improve capacity without letting technology replace the processes necessary for that capacity to be formed. Lesson 1: AI helps do better, but does it really learn better? In just a few years, from a more or less unfamiliar tool, AI has become an assistant in learning and work. millions of people. A piece of code can be created in a few seconds, a large amount of documents can be read and synthesized in a short time, an essay or solution can be completed with the help of AI. But it is this ability to help people do better and faster that raises an important question: Do better results mean people actually learn more? Studies cited by three international experts who spoke with the Government Electronic Newspaper show that these two things do not always equal. parallel. As AI becomes increasingly capable of doing part of the thinking work, the challenge is not only to teach learners how to use technology, but also to ensure that real competence is still built in the learner. When performance and competence are no longer the same Download Professor Mutlu Cukurova According to Professor Mutlu Cukurova, professor of learning and artificial intelligence at University College London (UCL), UK, when evaluating learning in the age of AI, it is necessary to distinguish between performance results and the learners actual capacity. Performance is what a person can produce at a given time with the available support of AI. Competence is what remains when that support is no longer there. According to him, AI is causing these two factors to no longer necessarily go hand in hand: The results that learners show may be better but do not reflect their true abilities. This is a change that the education system is currently not fully prepared to handle. A randomized study of about 1,000 math students found that practicing with a versatile AI tutor increased results by up to 127%. But when not using AI in the exam, students not only did not show progress but also performed a bit worse. Another study also showed that with AI support, learners tend to be less persistent, give up earlier and skip more exercises. When learners entrust their thinking to AI Download Dr. Stephan Vincent-LancrinDr. Stephan Vincent-Lancrin, Deputy Director of the Center for Educational Research and Innovation, level education economist OECDs high level, also points out a well-documented risk: Learners can assign assignments or learning tasks to artificial AI to do them for them - a phenomenon called cognitive offloading, which means transferring part of their thinking activities to tools. Another risk is that learners ignore essential parts of the thinking process when interacting and seeking help from AI - a phenomenon called metacognitive laziness, which is the tendency to rely on AI instead of thinking and checking for themselves. and adjust the way they solve problems. Sharing the same opinion as Professor Mutlu Cukurova, Dr. Vincent-Lancrin said some studies show that people perform tasks better when using generative AI, but that does not mean they actually learn. The difference is revealed when learners no longer have AI support. This is an especially notable issue as AI has become popular in education. According to Dr. Vincent-Lancrin, available information shows that the use of generative AI has become very popular at the high school and university level. AI is mainly used to find information, summarize and translate text; At a lower level, learners use AI to replace tasks or as an interactive tool to support learning. Better work does not mean more learning. Professor Barbara Oakley (Oakland University, USA), co-founder of the famous course Learning how to learn, believes that the impact of AI on learning depends greatly on when the tool is introduced into the learning process. Download Professor Barbara Oakley She led a study comparing student essay writers with different forms of support, including ChatGPT, tutors, and a group that received no support. The group using ChatGPT produced better essays, but subsequent tests showed that those better results did not mean they learned more. The researchers also observed that they were less self-correcting, less reflective, and less deeply engaged in the learning process. Another study with high school students in math showed that the group using GPT-4 performed better than their peers during practice, but their results dropped sharply when entering the final exam without AI support. Theo Professor Oakley, AI has replaced the cognitive activities that help form understanding, while learners only follow the process without directly performing those cognitive activities. The difference between experts and beginners is even clearer. People with a strong knowledge base can use AI as a power amplifier: They can evaluate the results AI produces, detect errors, adjust requirements and use what is useful. Meanwhile, beginners do not have a knowledge framework to verify the AIs results, so it is easy to mistake the fluency of the AIs answers with their own understanding. How do we know the learners true ability? Professor Cukurova believes that there is a fairly simple test to determine which ability really belongs to the learner: After a while, when encountering an unprecedented problem, can they solve it themselves? This can be assessed in many ways. First, at least in some assessments, learners need to perform the task independently, withou
