In the past few months, ChatGPT, an AI chatbot with the ability to write essays, make calculations, and fix code, has exploded in popularity. As the latest “large language model”, it is able to mimic human language, understand emotions, and converse naturally. Despite the controversy surrounding it, ChatGPT is being applied in many different fields. But how is it possible for a simple AI chatbot to complete so many different tasks in just a few seconds?
As mentioned earlier, Chat-GPT is a “large language model” (LLM), or a computer system (neural network) that is trained with millions of pages of text. This training allows it to understand relationships between words and phrases within the text, which is crucial because it makes it possible for the model to decipher your question (Kapronczay, 22). The training also allows the model to generate a response. As Kyle Wiggers of VentureBeat states, “The task of long-form question answering (LFQA) involves retrieving documents relevant to a given question and using them to generate a paragraph-length answer to that question.” (Wiggers, 21) Once the AI retrieves relevant documents, it typically paraphrases the information in them, producing an answer that is well formed and, sometimes, accurate.
While these responses can seem to convey emotion, ChatGPT doesn’t actually have emotions. As a BBC article states, “Chatbots currently have about as much capacity for real feelings as a calculator. Artificial intelligence systems are only simulating the real deal.” (Gorvett) As mentioned before, chatbots like ChatGPT are trained to analyze patterns in human language, which is what enables them to create natural, human sounding responses. ChatGPT is only able to simulate emotions - it doesn’t have any of its own.
While ChatGPT can seemingly do it all, there is still a lot of room for improvement. As Ridoan Karim of Monash University states “the ability of ChatGPT to generate conversational text raises ethical concerns about its potential to generate fake/wrong news or other misleading content. This could have severe ramifications, such as harming reputations, spreading false information, or even inciting violence.” Additionally, an article titled “ChatGPT and Large Language Model Bias” by CBS shows us that ChatGPT can produce responses that are sexist, racist, and factually incorrect (Croxton, 23).
Because of its shortcomings, many people are split over whether ChatGPT should be used. Meta (Facebook), Canva, and Shopify have started using chatbots similar to ChatGPT for customer service (Shrivastava, 23), while JPMorgan, Bank of America, and Amazon have limited employee use of ChatGPT (Wodecki, 23). Similarly, many school systems and universities have banned ChatGPT, because it can encourage cheating and plagiarism (Johnson, 23). While ChatGPT is clearly far from perfect, it appears to be a versatile tool that could have a huge potential in the future.
Works Cited
“ChatGPT and Large Language Model Bias.” Www.cbsnews.com, www.cbsnews.com/news/chatgpt-large-language-model-bias-60-minutes-2023-03-05/.
Gorvett, Zaria. “The AI Emotions Dreamed up by ChatGPT.” Www.bbc.com, www.bbc.com/future/article/20230224-the-ai-emotions-dreamed-up-by-chatgpt.
Johnson, Arianna. “ChatGPT in Schools: Here’s Where It’s Banned—and How It Could Potentially Help Students.” Forbes, www.forbes.com/sites/ariannajohnson/2023/01/18/chatgpt-in-schools-heres-where-its-banned-and-how-it-could-potentially-help-students/?sh=23994d386e2c. Accessed 11 Mar. 2023.
Karim, Ridoan. “ChatGPT: From the Horse’s Mouth.” Monash Lens, 13 Feb. 2023, lens.monash.edu/@politics-society/2023/02/13/1385448/chatgpt-old-ai-problems-in-a-new-guise-new-problems-in-disguise.
“Language Models Struggle to Answer Questions without Paraphrasing Training Data.” VentureBeat, 17 Mar. 2021, venturebeat.com/business/language-models-struggle-to-answer-questions-without-paraphrasing-training-data/. Accessed 11 Mar. 2023.
Shrivastava, Rashi. “ChatGPT Is Coming to a Customer Service Chatbot near You.” Forbes, www.forbes.com/sites/rashishrivastava/2023/01/09/chatgpt-is-coming-to-a-customer-service-chatbot-near-you/?sh=2e06ec0c7eca. Accessed 11 Mar. 2023.
Wodecki, Ben. “JP Morgan Joins Other Companies in Banning ChatGPT.” AI Business, 202AD, aibusiness.com/verticals/some-big-companies-banning-staff-use-of-chatgpt.