The term AI can be used in a wide range of ways but there have been some misconceptions on the recent AI chatbots. An article about how ChatGPT spreads misinformation states, “Today’s AI chatbots work by drawing on vast pools of online content, often scraped from sources such as Wikipedia and Reddit, to stitch together plausible-sounding responses to almost any question” (Verma & Oremus, 2023). This causes some answers that be incorrect since it only gives an answer based on patterns it has already seen. Since the bot is trained on the internet it allows individuals to converse with the internet but like the internet not everything you find is correct. Hughes explains how ChatGPT works, and it uses, he states, “As a language model, it works on probability, able to guess what the next word should be in a sentence” (Hughes, 2023). These AI chatbots are trained on tons of information that is online. AI then uses the patterns it sees to generate answers to prompts. In a way, it is like a parrot since it uses what it knows and repeats it back but does not know the reasoning. In the same article, Hughes writes, “The model was trained using text databases from the internet. This included a whopping 570GB of data obtained from books, web texts, Wikipedia, articles and other pieces of writing on the internet” (Hughes, 2023). While having so much knowledge sounds like a good thing it causes conflicts and makes it difficult to find exactly what is being asked. When Google searches for a prompt it was given there could be billions of results many with good information. Chatbots can make it easy to see all the good information in one place, but they can also use misinformation and make it sound correct. In Oxborrow’s article about the controversies of ChatGPT he asks ChatGPT to explain how it works, “’I'm sorry, but I am not familiar with ChatGPT’, it responded. ‘I am a language model trained by OpenAI, and I don't have the ability to browse the internet or learn about other AI models. I am only able to provide information based on what I have been trained on and what I can generate from that information’” (Oxborrow, 2023). This is due to the fact it was not trained on current information. Since ChatGPT was not yet released when it was trained it could not have known how it works unless given other information from its creator. However, when I recently tried asking ChatGPT how it worked, it gave me a different response. While the use of the term AI is not incorrect, there are misconceptions about what it means.
Comments