The rise of ChatGPT in late 2022 has revolutionized the way millions of people access knowledge. The appeal of large language models lies in their ability to provide quick and polished summaries of information, making learning feel effortless. However, a recent study co-authored by marketing professors Jin Ho Yun and Shiri Melumad suggests that this ease of use may come at a cost.
The study, based on seven experiments with over 10,000 participants, compared the effectiveness of learning through ChatGPT versus traditional Google searches. Participants were asked to learn about a topic, such as vegetable gardening, using either ChatGPT or Google search. The results consistently showed that those who used ChatGPT developed shallower knowledge, invested less effort in writing advice based on their findings, and produced advice that was less informative and helpful.
One possible explanation for this difference is that ChatGPT users were exposed to less diverse information compared to those who used Google search. Even when the facts and platform were held constant, learning from ChatGPT summaries led to inferior knowledge retention compared to gathering and synthesizing information from web links.
The study suggests that the passive nature of learning through ChatGPT may be the reason for the shallower knowledge acquired. Active engagement with the material, as required in traditional web searches, leads to a deeper understanding of the topic.
The researchers emphasize that the solution is not to avoid using ChatGPT altogether but to become more strategic users of such tools. For quick factual answers, ChatGPT can be beneficial, but for developing deep and generalizable knowledge, traditional web searches are more effective.
Future research by Shiri Melumad will focus on making learning with AI models a more active process. By introducing healthy frictions or guardrails that encourage users to delve deeper into the original sources, it may be possible to enhance the effectiveness of learning with ChatGPT.
In conclusion, while ChatGPT offers undeniable benefits, users need to be mindful of its limitations when aiming to develop in-depth knowledge. By understanding when to rely on AI summaries and when to engage in more active learning processes, individuals can optimize their learning outcomes in an increasingly AI-driven world.
Original Article Source:
https://theconversation.com/using-ai-to-learn-why-you-may-be-better-off-sticking-with-google-198084
