Master LLMs with our FREE course in collaboration with Activeloop & Intel Disruptor Initiative. Join now!


AI Education Threat
Artificial Intelligence   Latest   Machine Learning

AI Education Threat

Last Updated on April 22, 2024 by Editorial Team

Author(s): Nima Shokouhfar

Originally published on Towards AI.

Crisis: AI Cheating? Source: By the author


Has the advent of large language models like ChatGPT deteriorated the education system? Just imagine a future surgeon, who cheated through their entire education using ChatGPT, holding a knife and about to perform surgery on you. Wouldn’t you be concerned? As shocking as this might be, the rise of LLMs isn’t entirely negative. For instance, the number of papers written on software programming has doubled since the rise of ChatGPT [1].

Crisis: AI Cheating? Source: By the author

This proves that these tools have empowered us to achieve more, advancing technology at an unprecedented pace. However, the question remains: does quantity equate to quality? What if these papers are filled with fluff?

The number of submitted papers. Source by Arslan Akram on arxiv

Can You Trust Academic Publishing Anymore?

Indeed, some of them are. Last month, my friend came across a paper that started in a manner typical of chatbots, which proves that no one read the article before publishing, and the publisher did not notice either. This underscores the severity of the crisis in the education system. Curious about this instance, we dug further. After an exhaustive search, we found a repeating pattern. Indeed, articles are already filled with this jargon. Here are other examples.

AI generated papers. Source: Google Scholar

This finding fundamentally questions the integrity of our education system to the extent that you might ask yourself: Shouldn’t we enforce severe punishments for researchers who use these tools?

AI Cheating and Punishment. Source by Ron Lach on Pexels

Can AI Bridge the Educational Divide?

Do not be so quick to judge. These tools are not as evil as you might think. AI is positively impacting the education system in various ways. For instance, students can have a personalized teacher cater to their needs using large language models (LLMs). Additionally, those in poorer countries who cannot afford to attend school can use chatbots as their teachers. This is democratizing the availability of knowledge for everyone.

AI teachers in developing countries. Source by director muuh on Pexels

However, this raises a more philosophical question: Is it fair if one child has access to AI and another doesn’t? Or if one student uses basic free versions like GPT-3.5, while another uses advanced paid versions like GPT-4?

Can Educators Keep Up with AI Cheating?

Last week, a cousin of mine was talking to me. He told me he is excelling in his studies thanks to ChatGPT. He admitted that he hadn’t written a single essay the entire year. Considering the strict parents he has, imagine if he gets caught. Oh man, that would be the end of the world for him. But are we currently capable of detecting cheaters like this?

Fighting fire with fire, tech giants like OpenAI have developed AI detectors, like GPT-Zero, to catch cheaters like my cousin. Undoubtedly, this improves the quality of the education system as we can filter cheaters out of the system. Yet, one question remains: How many teachers are aware of these tools? This is why it is essential to continue upgrading and educating educators to combat the misuse of technology in our education system. Educators who keep up with the latest trends could guide students to use technology in an ethical and sustainable way, ensuring a prosperous society in the future. However, a concern arises: Is this enough? For instance, is merely knowing about AI detectors sufficient to detect AI plagiarism?

The Future of the Classroom. Source: By Max Fischer on Pexels

Can AI Catch a Cheating Student?

Extensive research has been conducted on this topic, evaluating AI detectors like GPTZero and Giant Language Model Test Room (GLTR). According to one study [2], while some tools perform reasonably well in specific scenarios, overall, they struggle to consistently and accurately distinguish between AI-generated and human-generated texts.

The Paper’s Findings. Source: By Chaka ChakaA on arxiv

Furthermore, AI tools keep evolving. So, even if a detector works today, it may not work tomorrow. Imagine investing thousands of dollars in licensing at a university, only to end up passing cheaters with great grades, and unfairly penalizing hardworking students. Here lies our fundamental question: Is all hope lost?

Is all hope lost?

This question has been a subject of debate among tech enthusiasts and scientists for a while now. However, one thing everyone agrees on is that the landscape of education is changing rapidly, and we must act now. This involves changing our educational policies, upgrading our toolsets, and integrating AI into our curriculum. This is a war — a war for our future generations. Remember, prosperity is not given; it is earned.

Feedback Wanted: Let’s Collaborate to Improve!

Your insights and experiences are invaluable to me, and I’m eager to hear from you! Here are several ways you can reach out and contribute to our learning community:

  • Comments Section Below: Share your thoughts directly under this post. We’re looking for feedback on technical clarifications, content suggestions, and your personal experiences with the technologies discussed.
  • Social Media: Connect with me on LinkedIn, Twitter, Youtube, or Facebook

Your participation not only helps us improve but also enriches the community’s knowledge base. We’re excited to see your comments, learn from your experiences, and answer your questions. Together, let’s make AI more accessible for everyone!

Let’s make AI more accessible for everyone. Source by fauxels on Pexels

Join thousands of data leaders on the AI newsletter. Join over 80,000 subscribers and keep up to date with the latest developments in AI. From research to projects and ideas. If you are building an AI startup, an AI-related product, or a service, we invite you to consider becoming a sponsor.

Published via Towards AI

Feedback ↓