ChatGPT can be a useful tool, but it is not perfect. If you use this model in the classroom, it is important to recognize its limitations and help teach students how to identify these limitations. This can also be a good opportunity to emphasize critical reading and thinking skills, which OpenAI encourages as an effective application of the tool.
It may sound correct but actually be wrong
Sometimes ChatGPT sounds convincing, but it may provide incorrect or misleading information (often referred to in the literature as “hallucinations”).
It may even fabricate citations or quotations, so do not use it as your sole source for research.
Sometimes it may say that a question has only one answer when in fact it does not, or misrepresent different aspects of an argument by incorrectly giving equal weight to each side.
It does not know everything
ChatGPT’s knowledge is not up to date, so in most cases it does not know about current events or trends.
ChatGPT is currently trained primarily in English.
OpenAI cannot determine what it knows and does not know, and cannot fully understand when it is expressing confidence in an incorrect assertion.
No access to tools such as calculators or the internet (mostly)
ChatGPT cannot browse the internet or access the latest information on the internet unless plugins are enabled.
It cannot verify facts or perform complex calculations without internet access or the use of plugins. |