Artificial intelligence: “ChatGPT”, the tool that can write instead of humans

Artificial intelligence: “ChatGPT”, the tool that can write instead of humans

A new artificial intelligence tool is taking the tech world by storm. It’s called “ChatGPT” and it can do just about anything. Type in a query and it can write an essay, compose a song or even solve a math problem in a few seconds.

But some point risks and deviations of such an application which could also allow cheating during an exam or writing fake news articles.

How do you know if artificial intelligence wrote the text you are reading? We posed the question to Muhammad Abdul-Mageed, an artificial intelligence researcher at the University of British Columbia, Canada:

“The weak point of ChatGPT is that it is trained on outdated data. With data dating from 2021 or 2022, it would not be able to detect recent events. For example, it would not be able to tell you if France won or not the FIFA World Cup in Qatar.”

There are other clues to look out for. If several people ask the exact same question to “Chat GPT”, the tool will generate almost the same answer for each of them. If you are a teacher and you correct several assignments using the same construction, the same examples or the same reasoning, then it may be a text generated by Artificial Intelligence.

A 22-year-old student developed a application called “GPTZero” who, according to him, can detect if a text is written by “ChatGPT. But this tool is not infallible, as Irene Souleiman, AI expert for the startup “Hugging Face” explains:

“There is no magic solution for AI detection. Detection models play a catch-up role but they are not able to give such an accurate result as to what is wrong and generated by AI.”

So what to do? Artificial intelligence becoming more and more efficient, maybe it’s time to rethink how students in universities are assessed and what kind of policies governments need to put in place to ensure that this powerful tool is not used to cause harm in the real world.