- Ways generative AI tools may help your study
- Risks with using generative AI tools for study
- Generative AI tool examples
Generative AI, such as Dall.E or Google Translate, has seen amazing advances in recent years and is likely to continue to evolve rapidly. Generative AI is able to generate new data, such as text, images and audio, that is similar to existing data. This type of AI is trained using deep learning techniques and can be used in a variety of applications.
There is potential for you to use Generative AI responsibly to assist you in your studies but there are also risks that you must consider when using this type of technology.
Generative AI tools may be able to:
- generate quizzes or flashcards.
- recommend authoritative sources on a topic for you to follow up.
- summarise information on a topic to help you get started.
- improve your grammar, sentence construction and other language skills.
Generative AI can also:
- explain the solution to different types of problems to increase your understanding e.g mathematical problems, coding errors, formulas.
- analyse data e.g. create spreadsheets, tables and organise information.
- provide creative inspiration or suggestions that you can build on.
- restore low quality images or video.
You need to consider these risks when using generative AI tools for study or work.
- Do not provide any private information when using these tools.
- Verify any information provided by generative AI tools with credible sources and check for missing information.
- Acknowledge any generative tools that you use for your assignments or work and how you used them. For example, include the name, model or version, date used and how you used it in your assignment or work.
- Be sure to check with your course coordinators if you plan to use generative AI tools to help you complete assignments.
If you use AI tools and present the work as your own, you put your academic integrity at risk.
Our ChatGPT and other generative AI tools referencing guide has tips on how to cite or acknowledge your use of these tools.
Incomplete, inaccurate or offensive information
Information provided by generative AI tools may be:
- Out of date
- Biased or offensive
- Lacking common sense
- Lacking originality.
Generative AI, such as ChatGPT, have limitations on what information they can provide. ChatGPT doesn’t have access to:
- Personal information or private data.
- Events that happen after the knowledge cut off.
- Information not present in the dataset used for training.
- Information that is not available in written or spoken form.
- Information about very specific or niche topics that are not available publicly.
Source: Answer provided by OpenAI’s ChatGPT version 3 on 27 January 2023 (edited for brevity).
Text version of screenshot:
Question: Who is the current British monarch?
ChatGPT response: As of my knowledge cut off, the current British monarch is Queen Elizabeth II. She has been the queen since 1952 and is the longest-reigning current head of state in the world.
Actual correct answer: King Charles III is the current British monarch. Queen Elizabeth died on 8 September 2022. ChatGPT version 3 couldn’t answer this question correctly as it was trained on data sets from prior to 2022.
Under Australian copyright law, works created by non-humans are not eligible for copyright protection. In addition, there is a question about the materials used to train AI and what protection these materials had, which may lead to future legal challenges.
The Complex World of Style, Copyright and Generative AI blog discusses some of the issues around copying works to train AI and whether that should be considered copyright infringement.
Be sure to check the usage policies or conditions of any tools before you start using them.
Awesome Generative AI provides a list of modern generative AI projects and services.
Have you used generative AI tools for your studies?
Perplexity AI is an AI powered search engine. If you ask a question, the response includes in-text citations from web sources that you can follow up to learn more about the topic. Note: It does not include any scholarly sources in its responses. It is important to consult different types of academic sources for your assignments and not solely rely on information from webpages.
Bard is Google’s experimental generative AI tool based on their LaMDA conversational AI model. You need a Google account to use it. Google warns that some of Bard’s responses may be inaccurate so it is important to verify the information provided with other sources.
Bard allows voice input and you can export the responses to Google Docs or Gmail.
ChatGPT is a language model and can generate text responses based on text questions you ask it. ChatGPT is currently free to use during a ‘research preview’, but you must register your details.
The ChatGPT blog explains that sometimes it will give “plausible-sounding but incorrect or nonsensical answers”. Be careful to always check the information provided is accurate.
We asked OpenAI’s ChatGPT version 3 on January 27th, 2023 the following questions to help explain how to use the tool for study. ChatGPT’s responses have been edited for clarity and brevity:
What are recommendations for how to ask questions of Chatgpt?
Here are some recommendations for asking questions of ChatGPT:
- Be specific: The more specific your question is, the more accurate the response will be. Avoid asking open-ended questions that can have multiple answers.
- Use proper grammar and punctuation: ChatGPT is a language model and performs best when it is given well-formed sentences.
- Provide context: If your question is related to a specific topic or field, provide some context to help ChatGPT understand the question better.
- Be concise: Keep your question short and to the point. Long and complex sentences can be difficult for ChatGPT to understand.
- Avoid asking personal information: ChatGPT is a machine learning model and it does not have personal information. It also has a knowledge cut-off so it might not have the most recent data.
- Be patient: ChatGPT is a large model and it can take a moment to generate a response.
- Be aware of its limitations: ChatGPT is a model trained on a large dataset of text, so it may not have knowledge of certain information or events that have occurred after its knowledge cut-off date. It might also struggle with certain types of questions, like understanding sarcasm or idiomatic expressions.
Can Chatgpt provide information about authoritative sources for different study topics?
ChatGPT has access to a vast amount of information and can certainly provide information about authoritative sources for different study topics. However, it is important to note that not all the information provided by the model may be accurate or up-to-date. It is always a good idea to verify any information provided by ChatGPT with multiple sources and to consider the credibility of the source before using it for research or academic purposes.
To find authoritative sources, you can ask ChatGPT for things like:
- Can you give me some reputable sources for information on [topic]?
- What are some well-known journals or books on [topic]?
- Who are some experts or researchers in the field of [topic]?
Is ChatGPT able to critically analyse information?
ChatGPT can generate responses based on patterns and associations it has learned from the text it has been trained on, but it does not have the ability to understand the meaning or context of the information it is processing. Therefore, it cannot make judgments about the accuracy or reliability of the information it is providing.
It’s also good to keep in mind that the information is generated based on the data it was trained on, which might contain biases and inaccuracies.
It’s a good practice to cross-check the information provided by ChatGPT with credible sources before making any decision or taking any action.
OpenAI’s Dall.E 2 can generate images based on text input, edit existing images and create variations of existing images. Consider any privacy implications before you upload photos.
Dall.E 2 could help you:
- visualise a concept
- make plans
- experiment with designs.