ITB School of Electrical Engineering and Informatics (SEEI) Informatics Research Group lecturer Dr. Eng. Ayu Purwarianti, S.T, M.T. said that there are several risks that have to be considered when using ChatGPT, for instance regarding regulations, plagiarism issues, and ethics in using ChatGPT, especially in the academic setting.

“ChatGPT is actually very useful in helping us learn, but we do have to be careful with our intentions of using it. If students for example were told to write an essay with the aim of increasing their analyzing skills and be more critical and creative, then they shouldn't use ChatGPT. They should write an essay using their own words then compare it with ChatGPT’s results,” Dr. Ayu explained.

Dr. Ayu gave a reminder to be more wise in using ChatGPT as a learning tool, because it brings numerous risks to the table. The first risk is the inaccuracy in ChatGPT’s answers, so it is hoped that users do a validation check or find another more reliable source in searching information.

Another risk is related to plagiarism where we do not know the data and answer sources provided by ChatGPT. So, for some cases regarding copyright, such as book writing and copywriting, we shouldn’t let ChatGPT do a take over since the final responsibility still lies with humans.

The next risk is how ChatGPT can cause potential misuse, since we can ask it to make a program code such as a jailbreak or something meant to pass through security.

However, with all the existing risks, it is very difficult to hold back on its development, because many people are now competing to develop something akin to ChatGPT at lower prices.

The European Union (EU) considers ChatGPT to be high risk, and Indonesia itself does not have any specific rules regarding the usage of ChatGPT yet. UNESCO has provided recommendations about the risk of using AI, but the readiness of each country is different to be able to follow said recommendations.

“Every institution has their own way of dealing with this, “ she concluded.

There are currently plagiarism checkers, however it is quite difficult to verify ChatGPT’s results since the target of said generator is to generate text that is as human-like as possible, so to see whether a text is machine or human made based on its writing style usually does not work.

The emergence of regulations to manage the usage of ChatGPT is an important step to navigate the new era of human-machine interaction. By adopting the right regulations, we can benefit from the amazing potentials of ChatGPT while also protecting the interests and safety of users.

Photo: Freepik
Reporter: Satria Octavianus Nababan (Informatics Engineering, 2021)
Translator: Hanna Daniela Ayu (Aerospace Engineering, 2021)