Yes! While powerful and useful, there are some things you need to know about generative AI:
- It's not very smart. In fact it's not smart at all. It's basically a very fast version of what you see when you're texting and the texting app suggests words to follow the words you have already entered or when you start typing a search query and suggested searches appear under the search box in response to what you're typing. It knows, for a given context, the words that usually follow each other to make a coherent responsive to your prompt, but it doesn't know anything
- Therefore the answers you get will sound plausible, but could be wrong.
- Generative AI will make things up if it doesn't have the information as part of its dataset. When it does so, this is called "hallucinating"
- A common hallucination is making up information sources such as books, reports and journal articles
- The dataset used by the generative AI tool might be out of date (Chat GPT's dataset is behind the times) and therefore useless for current events or up-to-date knowledge on a topic
- Generative AI datasets are created by humans and can include fiction, jokes, satire, opinion, lies and biased information, so what you see might be biased or just plain wrong
- These tools might keep and use any personal data that you enter
When using generative AI tools:
- Be aware of these limitations
- Verify, using sources you know are reliable (or using the SIFT method) information given by generate AI tools
- Check to see that the information sources referenced by ChatGPT or other Chatbots actually do exist
- Read the terms and conditions of use and any data use & privacy statements that are available
