Generative AI tools, like ChatGPD, are really good at creating or producing content, from quick answers to short stories or even complete essays. However, these tools cannot be trusted to always produce correct or accurate information. When using AI we must understand it's drawbacks in order to evaluate if we trust the information being produced.
There are several things to consider when evaluating AI Generated Content.
The content generated by AI tools should be carefully checked as it may contain errors, false claims, or plausible sounding, but completely incorrect answers.
Why ChatGPT and Bing Chat are so good at making things up (Edwards, April 2023)
Generative AI can also be used to create fake images or videos so well that they are increasingly difficult to detect, so be careful which images and videos you trust, as they may have been created to spread disinformation.
That photo of Trump being tackled by police isn’t what you think it is as the internet enters a new era of AI disinformation (Lajka et al., March 2023)
Generative AI relies on the information it finds on the internet to create new output. As information is often biased, the newly generated content may contain a similar kind of bias. Example of potential bias include gender-bias, racial bias, cultural bias, political bias, religious bias, and so on. Scrutinize AI generated content closely for inherent biases.
Rise of AI Puts Spotlight on Bias in Algorithms (Bousquette, March 2023)
AI content may be selective as it depends on the algorithm which it uses to create the responses, and although it accesses a huge amount of information found on the internet, it may not be able to access subscription based information that is secured behind firewalls. Lots of high quality academic research is located behind paywalls and therefore inaccessible to AI platforms.
Content may also lack depth, be vague rather than specific, and it may be full of clichés, repetitions, and even contradictions.
Generative AI tools create content based on data sets. Therefore, the information may only be as current as the most recent dataset it was trained on. At time of writing (May 2023), ChatGPD has limited knowledge of events after September 2021 (Chan, March 2023).
In some disciplines it is crucial to have the most recent and updated information available. For example, in the early days of COVID-19, research was being produced at a very fast pace. It was vital for researchers to have the most up to date data. Technology is another area that is constantly changing, and information that is valid one year might not be valid the next.
There are a few things to keep in mind when it comes to generative AI and sources.
Generative AI tools don't always include citations with their answers, and when they do they've been known to create incorrect citations and to make up citations to non-existent sources. This phenomenon is often referred to as AI Hallucination.
ChatGPT and Fake Citations (Rozear & Park, March 2023)
Why ChatGPT and Bing Chat are so good at making things up (Edwards, April 2023)
Not crediting sources of information used and creating fake citations are both cases of plagiarism, and therefore breaches of Academic Integrity.
It's important to also remember that most generative AI tools do not have access to text of articles behind paywalls. This means it only has access to abstracts and citation information for many academic articles, including ones found through the JIBC Library and Google Scholar.
The newest version of ChatGPT, which as of May 2023 is on limited release to some users, has a browsing feature that browses the internet. This feature is a step towards receiving real citations and resources. Remember, any sources found online must be evaluated. See the Evaluating Websites page for more information.
Generative AI tools rely on what they can find in their vast knowledge repository to create new work, and a new work may infringe on copyright if it uses copyrighted work for the new creation.
There is also current debate about who owns the copyright to work created by AI. Is it the person who wrote the code for the AI tool, the person who came up with the prompt, or is it the AI-tool itself? Although currently in Canada, AI-generated works are not copyright protected, this may change in the future.
Unless otherwise noted, this guide is licensed under a CC BY-SA 4.0 (Creative Commons Attribution-ShareAlike 4.0 International License).
The information on this page was adapted with permission from KPU's Artificial Intelligence LibGuide, created by Ulrike Kestler.