When people first saw the variety and quality of text that modern AI large language models (LLMs) can produce in seconds, many began to wonder if this technology would worsen the problem of plagiarism. While these concerns are quite reasonable, a better understanding of how LLMs work can help to clarify what is and isn’t problematic use of these powerful new writing tools.
In short, using AI to help with writing isn’t automatically an act of plagiarism, even if the outputs sound similar to other people’s writing. However, unscrupulous people can use AI to plagiarize other people’s work.
Plagiarism is mainly about accurately attributing credit for unique ideas and concepts we use. This was true before the introduction of generative AI, and it remains true now.
Large language models are created by training these computer systems using immense textual data sets that approach the size of the internet, itself. Text is fed through an AI model in a process often called ‘training’ where the objective is for the system to decipher and learn patterns that are present within the data. Across trillions of data points, the AI is measuring probabilities and relationships between small fragments of text and the order in which they appear.
Next, these models are pushed to predict new sequences of text based on the probabilities and relationships discovered during training. For example, a user might prompt the model with something like,
Write me a poem about ice cream
The AI would analyze all the words in this sentence, and then begin predicting what words should come next. Based on what it learned in training, the words it predicts should be in the format of a poem about ice cream!
If you’re thinking this is somewhat miraculous or unnerving, you’re not alone. Until just a few years ago, only people could creatively string together words in such evocative ways.
With enough data and enough computing speed, generative AI technology companies are refining the ability of computers to use powerful predictive algorithms to simulate our human language abilities.
So how do we think about using AI ethically to assist us when we write?
First, it’s a good idea to remember that writing something with AI assistant should start and end with you, the human writer. The scope of the piece should begin with you, the author, you can use AI to help with things like brainstorming and editing, and the final piece should be something you feel confident contains only your thinking unless otherwise cited. In this way, you’re accelerating the writing process, and staying accountable.
A nifty trick you can perform with AI is to ask it to assess whether or not a sample of your writing is close to any other public works. While AI systems are fallible and make mistakes, this can be a quick way to launch your own investigative journey to ensure you aren’t inadvertently treading on anyone else’s ideas. Based on what the AI replies, you can do some search engine research to see for yourself if any of the pieces offered by the AI is close enough to be concerned about accusations of plagiarism.
It is already the case that numerous writers are simply acknowledging when they have used AI as part of their writing process. It may very well be that we, as a society, will decide it is the norm or a best practice to clarify when we have used AI to assist with creating content.
For the time being, doing this is a matter of personal choice, but it may be interesting to test this out with audiences in low-stakes situations to see how they respond to such transparency.
Using AI in writing is not inherently an act of plagiarism, but it does require a clear understanding of how AI-generated content works. By ensuring proper attribution and maintaining the originality of the content, writers can ethically leverage AI tools to enhance their work. The key is to use AI as an assistant rather than a replacement, starting with human input and ending with a human sign-off to ensure accountability and authenticity.
As generative AI continues to evolve, it is crucial for writers to follow ethical guidelines and best practices to avoid common pitfalls. This includes being transparent about the use of AI and actively checking for any inadvertent plagiarism. By doing so, writers can harness the power of AI to improve their writing process while upholding the principles of originality and proper attribution.
