Limitations of AI ChatGPT
It is important to be discerning of the output from ChatGPT and to remember this AI tool is still in development with various versions being released as the technology improves over time. There have been a number of versions of ChatGPT released over the past few years with ChatGPT 5.1 being the latest one.
- Hallucinations – ChatGPT has been known to produce completely incorrect citations and references provided for sources of information.
- Bias - There can be bias in information that is input from users and bias in the output from the consulted sources which can affect the validity of the information.
- Limitations - ChatGPT is designed to be helpful and wants to aid with your presented queries however it does not acknowledge it’s own limitations. ChatGPT has been known for information falsification, which can be extremely compromising or difficult to decipher in a very specialised subject area. It may only be obvious to certain experts that the information is untrue.
- Currency of data from ChatGPT - It is important to be aware that ChatGPT 5.1 version has only gathered data up to June 2024 and does not include the most recent research conducted in a study field. However newer versions of generative AI tools include a live web search function to respond to up-to-date queries. Even so, it is important not to over-rely on information from ChatGPT and other models due to inaccuracies that may present. It is advisable to fact check information on reputable websites or to gather information from the SETU online library resources. No longer searching for information in traditional ways, can lead to acquiring outdated or incorrect data.
