Embedded Files
Explaining ChatGPT
Explaining ChatGPT
ChatGPT, Chat Generative Pre-trained Transformer, is a large language model-based chatbot. Large language model-based chatbots like ChatGPT function by ascribing text taken from the internet with additional data such as identity, dialogue, and entities. Ascribed data allows ChatGPT to return answers to user questions in text-based communication alike to texting with another person. When used properly, ChatGPT is capable of assisting in the research and writing process.
Accessing ChatGPT
Accessing ChatGPT
ChatGPT 3.5 is a free open AI tool. The button above will take you to the login screen. Click "Sign up" if this is your first time using ChatGPT or click "Log in" if you already have an account. ChatGPT 4.0 is not free, however, premium features are not necessary for students.
Prompting for ChatGPT
Prompting for ChatGPT
Effective Prompts
Effective Prompts
Ineffective Prompts
Ineffective Prompts
Follow Up Prompts
Follow Up Prompts
Engineering Research Prompts
Engineering Research Prompts
Revision
Revision
Shortcomings of ChatGPT
Shortcomings of ChatGPT
Although ChatGPT can reduce the labor for research, there are many possible faults within the responses ChatGPT creates. ChatGPT produces valuable, but not infallible information. ChatGPT responses must be used with skepticism in order to provide the most accurate information for research purposes. Some of the shortcomings of ChatGPT are:
- Informal or Casual Language
ChatGPT produces responses in a casual format. ChatGPT attempts to sound human-like, leading to informal sentence structure and vocabulary. ChatGPT’s responses are derived from the training data provided, which includes casual writing techniques. ChatGPT will attempt to respond formally if requested but oftentimes the response is still not technical enough for professional writing. This means that in order to mimic a conversational flow, ChatGPT may output including phrasal verbs, "it," contractions, colloquial terms, and even slang. ChatGPT cannot be used to produce technical writing itself, but can be a valuable resource to assist and revise research.
- Outdated and Inaccurate Information
ChatGPT cannot produce real-time information, bringing drastic limitations. ChatGPT’s data has knowledge up to January 2022. Any information after January of 2022 is not updated within the ChatGPT system. Any cutting-edge research that utilizes ChatGPT will be inhibited by ChatGPT's limitations. Any evidence or data provided by ChatGPT must be utilized with skepticism.
- Biased Outputs
ChatGPT, despite being AI, still has biases. These biases come from many sources. ChatGPT’s response is not justified with what the correct information truly is, but based on what is believed to be the intended response. ChatGPT assumes user intent and thus produces a faulty response. ChatGPT also has bias within data training. The text gathered from the internet contains biases and those biases are reflected in ChatGPT’s response
- Volatile Responses based upon Wording
ChatGPT is extremely susceptible to change based upon input values. Subtle changes in how a question is asked will produce dramatically different responses. This can range from formality in writing to content production. Although the idea of a prompt may be the same, ChatGPT can change the output data based upon user input.
Ethical Usage of ChatGPT
Ethical Usage of ChatGPT
ChatGPT returns information related to the requests that were inputted into the chat, and when requested, ChatGPT may give a limited amount of information. The capability of ChatGPT is derived from what is inputted to the search bar. The program tries to satisfy all the requirements set in the prompt asked. This means ChatGPT is capable of writing full paragraphs. Quality may diminish as the program continues to write condensed forms of information into a paragraph with commentary. However, the information ChatGPT provides should not be used for academic work . ChatGPT in an academic setting functions as a collaboration tool to cycle through ideas and fine-tune thoughts. The program should not do the thinking for the writer. As mentioned before the online program should be used for brainstorming not product production. ChatGPT does not filter the information ChatGPT gathers from scholarly sources. The program also does not indicate where the information is being extracted from. ChatGPT is not a source and cannot be used as a citation for an academic paper. ChatGPT guides writing. Inaccuracies with ChatGPT are dependent on the complexity of the question. Binary questions are more accurate than descriptive questions. Ai can source through the information for more complex questions. One must separate the question into pieces and make connections between the responses that ChatGPT outputs. This process is similar to a conversation. Writing involves a synthesis of ideas at some point in the process of writing an outline that places thoughts in a linear, concise form. The outline's purpose is to be a condensed form of the finished product. Writing should be authentic as the goal when writing an argumentative essay is to convince and persuade the reader when proving a point in the essay. To persuade writer must understand their audience and what values the audience has are meant to be shown. Writing is communicated more effectively when taking into account these factors.
GPTZero and Other AI Detectors
GPTZero and Other AI Detectors
A topic sentence and introduction paragraph about the history of zombies written by ChatGPT.
The same paragraph in GPTZero, a free AI detector, identified as 99% likely written by AI.
Text-based content can easily be tested for AI writing. AI detector tools are as easily available as AI text generators. The tool GPTZero is both effective and free. Similarly, Turnitin, a popular plagiarism checker for educators, has developed an AI detector for their customers. Students intending to directly use AI-generated text for school work should be made aware that AI detectors are easily available. Alike to how large language model-based chatbots like ChatGPT work, AI detectors scan large amounts of text data, both human-written and AI-generated to determine what is human-written. Ethical use of AI tools is necessary to avoid academic probation or other punishment.
Work Cited
Work Cited
Baker, Pam. ChatGPT TM. John Wiley & Sons, Inc., 2023.
ChatGPT, GPT-3.5, OpenAI, 14 Nov. 2023, chat.openai.com/chat
Hamilton, Scott. "W3D2: Writing Workshop."ENGL 147: Writing Arguments for STEM, 12 October 2023, California Polytechnic State University, San Luis Obispo. Keynote.
Sarrion, Eric. Exploring the Power of ChatGPT Applications, Techniques, and Implications. 1st ed. 2023., Apress, 2023, https://doi.org/10.1007/978-1-4842-9529-8.
Page updated
Google Sites
Report abuse