By Abhik Sengupta
AI research firm OpenAI’s ChatGPT is capable of analysing complex text inputs and providing answers in simple English. The AI-backed chatbot is even capable of writing long essays and solving math problems, though the output may be inaccurate at times. However, if you are aware of the context, humans can proofread the text or review arithmetic problems, making tasks much easier. This is why Elon Musk, owner of Tesla and Twitter, predicts that AI chatbots could practically end homework.
Reacting to a Twitter post, Musk says that in the “new world”, one can bid “goodbye” to homework. Some New York-based colleges are planning to ban the use of ChatGPT to complete homework. As a part of the measure, systems linked to university servers will not open the ChatGPT website. According to a report, the education department has restricted access to ChatGPT, citing “negative impacts on student learning and concerns regarding the safety and accuracy of content.”
The report citing education department spokesperson Jenna Lyle, highlights, “Due to concerns about negative impacts on student learning, and concerns regarding the safety and accuracy of content, access to ChatGPT is restricted on New York City Public Schools’ networks and devices.” However, the platform will still be accessible to students on their personal computers and tablets.
Although Musk’s remarks are satirical, concerns related to AI-backed platforms can be concerning. Last month, the Wall Street Journal experimented with one of the journalists by enrolling her in a school, where she used ChatGPT to finish her literary project. It was found that the platform offered a generic summary of the topic, though the results were good enough for the journalist to achieve a B+ grade for the project without hardly reading the text or doing thorough research.
However, some companies and researchers are working on platforms to detect texts generated by AI-backed chatbots like ChatGPT. Originality.AI, a website, claims that it can detect ChatGPT-generated content and check for plagiarism. TechCrunch reports that OpenAI, the creator of ChatGPT and Dall-E, is also working on a secret watermark to aid tools in detecting AI-generated texts. Texas Professor Scott Aaronson, who is also a guest researcher at OpenAI, says the tool can help prevent “academic plagiarism.” Interestingly, if we ask ChatGPT whether its answers are plagiarised, the chatbot points out that “text generated by GPT could be similar to text that appears in other sources. In such cases, a plagiarism tool might flag the text as potentially copied.”
However, many companies are realising the potential of the AI-backed chatbot, and fans regard ChatGPT as the future alternative to Google that can also answer human queries based on news articles and online sources, which can be filled with jargon. Microsoft is planning to launch a version of Bing search in a collaboration with OpenAI. Google’s management has also asked its AI teams to create a platform inspired by ChatGPT. (IndiaToday)