While AI-generated content can sometimes be challenging to identify, there are several strategies and techniques that can be used to recognise such content.
AI-generated content usually shows a consistent writing style that lacks the nuances and variations typically seen in human writing. Look out for:
Here's a site that almost certainly has been generated by AI. Do you notice anything about the style of writing?
Generative AI can’t refer to something it wrote earlier in the text because it doesn’t have a memory of what it wrote. If you wrote something and then repeated it, you might say something along the lines of "As I mentioned earlier...". Generative AI can't do this.
A human might slip up once and then correct the mistake; generative AI is likely to make the same mistakes over and over again.
Humans use parentheses (brackets). Generative AI tools don't (not yet anyway)
AI-generated content may contain inconsistencies in facts, dates, or details. Cross-reference the information with sources you know to be reliable to verify its accuracy.
Check for author information and credentials. AI-generated content often lacks a verifiable author or displays generic author profiles that will be written in the same robotic language (and may come with an AI-generated avatar).
Here's almost certainly an AI-generated author profile
AI-generated articles might include images with questionable sources or copyright issues. Look for discrepancies between image content and the text.
AI-generated content may be very concise or use the same keywords over and over. This can make the content feel unnatural.
AI-generated content may have consistent - repetitive! - formatting and structure, lacking the variability seen in human-created content.
AI-generated content may lack emotional depth or empathy in its writing.
Evaluate the context in which the content is presented. If it appears on a website or platform known for AI-generated content, exercise caution. Think of the CRAAP test!
Fact-check the information presented in the content using reputable fact-checking websites or resources. Think of the SIFT process!
Again, use the SIFT process - what do other sites and sources say on the topic? Here's an example: on Amazon in August 2023, there appeared a series of self-published Kindle books on wild mushroom foraging that used images of poisonous mushrooms to accompany text that stated that they were safe to eat. Warnings about using these books soon appeared online, so check elsewhere if something seems fishy to you.
User reviews and comments on sites may contain clues about whether the content is generated by AI or written by humans. Do bear in mind that some user reviews and comments may themselves be AI generated. Look for reviews/comments with depth, humour or a distinctive voice/tone
If you use Google to search, as most people do, then to find more information about a site, click on the three vertical dots next to a result to see more information about it.
There may be a Wikipedia page for the source site and if not, then it will be stated when the site was first indexed by Google. Sites first indexed in 2023 may be more likely to contain AI-generated material than older sites.
Generative AI is a very fast-moving field. Keeping up to date with the latest developments and headlines may help you become more skilful at recognising AI-generated content, or at least more aware of generative AI's capabilities. There is a subreddit for generative AI that might be useful for staying aware of new developments in the field.
