Chat Generative Pre-trained Transformer, or ChatGPT has recently become a factor in everyday scholarly work and writing.
It's becoming a common saying that AI or Large Language Model trained chat generators function by "predicting what the human would say next." Many users compare it to an elaborate version of autofill.
Ask ChatGPT what ChatGPT is, and it responds: "ChatGPT is a language model developed by OpenAI, specifically GPT-3.5, designed for conversational interactions. It utilizes deep learning techniques to generate human-like responses based on the input it receives. ChatGPT has been trained on a vast amount of text data from the internet and can generate coherent and contextually relevant responses across a wide range of topics. It is capable of understanding and generating text in multiple languages, providing information, answering questions, and engaging in interactive conversations."
ChatGPT can have conversations, and answer questions, as well as create outlines or even proof-read a paper.
What ChatGPT cannot do, however, and this is a biggie: It cannot return valid research articles to support research or substantiate its own claims!
To use ChatGPT, you must make an account. According to OpenAi's Privacy Policy, your personal information may be shared without "further notice." ChatGPT is currently free, but OpenAI could later put it behind a paywall. OpenAI has already launched a subscription plan (Chat GPT Plus).
ChatGPT isn't the only generative AI that students and LTC staff may encounter. Many services and graphic design tools, such as Canva, now feature generative AI. DaLL-E 2 is a generative art tool, which can create images based on a text description. Using DaLL-E2 can help students create illustrations or art for computer design without worrying about copyright issues.
DaLL-E 2 is not a free service. At the time of this information, it was $15.00 for credits to begin.
Here are some other resources that are paid vs. free (Please note that educators can create an Educator account in Canva that is free and provides premium content):
Chat GPT's information used to "train" it is dated prior to September 2021. After that date, ChatGPT does not have a reference point, so one of the drawbacks of using it is that current events may not be a part of the information used for responses from the Chat.
ChatGPT doesn't have access to real-time information or updates. Therefore, responses from the Chat may not be up-to-date with recent events or developments that have occurred after that knowledge cutoff date.
Another huge limitation in using ChatGPT as a student is that ChatGPT works by issuing responses based on patterns and examples from its training data. It will give responses that includes references that it is generating, and not references that exist in a scholarly journal. In other words, it will make up sources and citations.
As a student, you will want to use Chat GPT for a starting point, or to help brainstorm. Turning in or relying on ChatGPT's conversations that you haven't vetted can result in you turning in plagiarized work, spreading disinformation, or making decisions based on faulty assumptions.
One other major problem for ChatGPT is that, just like the information from a wide swath of internet datasets used for its learning, it can have bias. Developers are working to avoid this, but racist or sexist viewpoints, and other forms of bias, can be present in responses or information used to produce the responses.
Just like how ChatGPT can make up resources, it can sometimes make up other events (called "hallucinations" when referring to ChatGPT) so your friendly librarians remind you to use ChatGPT as a starting point or a sometimes-reliable-sometimes-faulty tool. Take all conversations with ChatGPT with a grain of salt, and never turn in a conversation or response as your own!
In a cringe-inducing court hearing, a lawyer who relied on A.I. to craft a motion full of made-up case law said he “did not comprehend” that the chat bot could lead him astray.
...At times during the hearing, Mr. Schwartz squeezed his eyes shut and rubbed his forehead with his left hand. He stammered and his voice dropped. He repeatedly tried to explain why he did not conduct further research into the cases that ChatGPT had provided to him.
“God, I wish I did that, and I didn’t do it,” Mr. Schwartz said, adding that he felt embarrassed, humiliated and deeply remorseful.
“I did not comprehend that ChatGPT could fabricate cases,” he told Judge Castel.
Read the complete article using your GALILEO access here.
From the article in the New York Times, "The Lawyer Who Relied on ChatGPT Explains Himself. It Was Awkward."
______________________________
Bard's financial mistake:
After being unveiled...the Google AI tool Bard is already getting called out for presenting inaccurate information in a demonstration meant to show off the tool's abilities.
Google shared an example of Bard answering the question "What new discoveries from the James Webb Space Telescope can I tell my 9 year old about?" The AI tool responds with three bullet points. The last one incorrectly states that the JWST, which launched in December 2021, took the "very first pictures" of an exoplanet outside our solar system. The first image of an exoplanet was taken by the European Southern Observatory's Very Large Telescope in 2004, according to NASA.
The mistake was called out by astrophysicists on Twitter and reported earlier Wednesday by New Scientist.
After the mistake gained notice, Google parent company Alphabet lost roughly $100 billion in market value. Its stock saw sharp declines...
From the article "Google ChatGPT Rival Bard Flubs Fact About NASA's Webb Space Telescope," on CNET.
Using generative illustration is only as good as user input! The image displayed was created by requesting a "1980's style librarian using a computer."
College students may encounter several problems when using ChatGPT. Here are some common challenges:
Lack of Context Understanding: ChatGPT may not always fully grasp the context or nuances of a conversation. It might misinterpret ambiguous queries or struggle with complex or specific questions that require deep domain knowledge.
Inaccurate or Biased Responses: ChatGPT learns from the data it has been trained on, which can include biased or incorrect information. As a result, it may occasionally generate responses that are factually incorrect, misleading, or reflect inherent biases present in the training data.
Sensitivity to Input Phrasing: The way a question or prompt is formulated can significantly impact the response from ChatGPT. Different phrasing or rephrasing of the same question may lead to inconsistent or contradictory answers, highlighting the model's sensitivity to input variations.
Generating Plausible but Unreliable Information: ChatGPT aims to generate coherent responses, but it may occasionally provide plausible-sounding yet incorrect information. Students should verify information from reliable sources rather than solely relying on ChatGPT.
Lack of Real-Time Information: ChatGPT's knowledge is based on text data available until September 2021. It does not have access to real-time or up-to-date information. Therefore, it may not be aware of recent events, news, or developments that have occurred after its knowledge cutoff.
Overreliance on ChatGPT: One potential problem is overreliance on ChatGPT as a sole source of information. It is important to utilize ChatGPT as a tool to assist learning and research, but not to substitute critical thinking, independent analysis, or consulting authoritative and credible resources.
