Many of us have been there: staring at a blank Word document, watching the clock, sweating a little bit, as a deadline looms. Having an idea for an article but not finding the right words to do it justice, or worse, facing the dreaded writer’s block.

This is why, for communicators, a tool like generative AI (GenAI) is thrilling. It can help us move past the oh-so-tough starting point, providing a base of text to work from. When we want to include an analogy or snappy conclusion, it’s a great source of inspiration. When it comes to brainstorming, it’s a useful partner, generating ideas for all sorts of topics, and can provide an outline for said suggestions.

Weighing risk vs. reward

Using GenAI comes with several potentially dangerous risks. Perhaps most concerning initially is the risk of data loss and privacy issues. Inputting content to an AI model allows the software to train itself on your data, potentially exposing your work to outside users.

Another major risk is the potential for misinformation because of the word association framework that drives the back end of this technology (ever searched your own biography and discovered made-up information about your professional or personal background?).

Importantly, it also comes with a heightened risk of plagiarism. While we become attuned to this risk at a young age, collaborating with software to draft a piece adds new layers of nuance. All too easily, your ideas can be combined with those of GenAI, leading to murky waters around content ownership and just who phrased what. While normally this wouldn’t be a major cause for concern—ideas can be cited—the original source of an idea from GenAI gets more complicated. Do you cite the AI provider (e.g., ChatGPT)? What if, unbeknownst to you, the software has directly lifted text from a book, paper, or other article? Users must be able to appropriately describe how and where any ideas have originated—human or otherwise.

Creating best practices

Several best practices can be introduced to help mitigate the risks described above. This list is not all-encompassing; in fact, we would expect it to evolve as the technology evolves over time. However, for those early adopters who have begun to practice brainstorming with GenAI, we suggest the following rules of thumb:

1. Have your topic laid out first. Understand what you’re writing about and have your main points before approaching GenAI for suggestions.
2. Do your own research. GenAI is a large language model that pulls in text that closely resembles its understanding of your question. While incredibly sophisticated, this can also mean it hallucinates “facts” from fiction. Rely on your own rigorous research methods over those of AI.
3. Similarly, one of GenAI’s weaknesses is math. There are other tools better suited for calculations—focus the large language model on language.
4. Be detailed with what you want. Looking to brainstorm a catchy conclusion? Ask it to suggest five to ten ideas for how your article might be concluded based on the topic you’ve provided. Or, if you have your topics and are looking for an outline, ask it to help you flesh out your ideas. The more detail you provide, the better the output will be.  
5. Refine and edit. Once GenAI has fleshed out a point for you, put it in your own words, similar to how you would with any other piece of research you’re utilizing. This will help you reduce the risk of plagiarism, while also ensuring the style stays true to your own voice.
6. Use anti-plagiarism software as an extra safeguard to help flag potential issues in your writing. A number of anti-plagiarism programs point out AI-generated content as well as potential plagiarism.
7. Ensure you don’t lose the human element. GenAI is very good at topics and statements, but the content can be a little dry and repetitive. While designed to replicate the way humans communicate, it can’t replicate your experiences. Bring your piece to life with relatable examples.

As finance and investment professionals, we have an obligation to uphold the incredibly high standards expected of us. Clients and the industry trust us. Our data has to be reliable, and our writing has to be clear, transparent, and free of errors. With a powerful new tool like GenAI, there comes a lot of opportunity to create efficiencies and a lot of enthusiasm around how we can put it to work. However, with great power comes great responsibility. Don’t let efficiency destroy your credibility.