In April 2023, the Washington Post published an article stating that ChatGPT had falsely accused a professor of sexually harassing a student, quoting another Washington Post article as evidence. The only problem was that this Washington Post article didn’t exist – ChatGPT had invented the accusation, as well as the evidence.
This revelation has led many people to question whether using artificial intelligence (AI) is safe for content creation. If writers decide to use AI, are they risking their professional reputation or worse, opening themselves up to potential legal consequences? This article outlines some considerations that content teams should factor in and provides suggestions on how you can best mitigate these issues.
AI Biases
Software developers train AI models using vast amounts of data, but this can introduce or reflect societal biases in your writing. ProPublica found that in courtroom risk assessments that use AI, black defendants were 77% more likely to be labeled as at higher risk of committing violent crimes than their white counterparts. As the World Economic Forum points out, humans train AI, and humans are biased and judgmental, even when they don’t mean to be.
Whether conscious or not, bias can have a detrimental effect on an organization’s reputation, so it’s vital that you do whatever you can to eliminate it from your content as much as possible. Our blog on avoiding biased language suggests checking your writing for insensitive language, recognizing individuality, and avoiding generalizations. Although this blog focuses more on academic writing, we feel that the tips are applicable to any type of writing.
You could also use sensitivity readers to check your writing for unconscious biases.
Legal Implications
There are some potential legal issues associated with AI that you’ll need to figure out before you use it to create content.
Copyright
Perhaps the biggest issue with AI-generated content is the question of who owns the copyright: the AI model or the person who creates the prompt. At the time of writing this article (June 2023), the US Copyright Office says that any work must be human-made to be copyrightable, meaning it’s not possible to copyright AI-generated work – however, it did launch an initiative to examine the copyright law and policy issues surrounding AI in March 2023.
But there’s also the issue of whether using AI content generators to create content is an infringement upon the rights of other writers and content creators. Because AI uses content from the internet to train its models, writers’ material is used in a way that they have not consented to. Using AI in this way is still relatively new, and lawyers are scrambling to create new laws to patch up loopholes in existing copyright law.
Liability and Accountability
Questions around who is responsible for any harm resulting from AI-generated content or whether it infringes upon rights have emerged with the increased use of AI. Is the AI model liable or the creator?
A white paper on AI from the European Commission proposed a legal framework for using AI, but it will need to continually evolve to keep up with developments in AI technology. And creators will have to be careful to follow the newest laws to avoid legal action.
Effect on SEO
You’ll probably come across conflicting information about whether using AI to help you create content will harm your SEO efforts. Google initially said that it would penalize AI-generated content, but it revised this stance in 2023, stating that if AI was being used to help create useful and engaging content, and not write it entirely, then it wouldn’t be flagged in Google’s search engine results page.
How Can You Use AI Safely and Ethically?
So now that we’ve established that AI is safe to use, how can we make sure we use it in the safest way possible? There are a few key factors to keep in mind:
1. Understand AI’s Limitations
AI models, such as language models, generate content based on patterns and examples from their training data, which is usually taken from across the internet. AI is incredibly proficient at mimicking human language, but it’s still a machine and therefore can’t understand sarcasm and tone. Tone is vital in communication, as this article in Psychology Today discusses, but it’s especially important in writing when we can’t get contextual clues from body language or facial expressions.
Content teams must recognize this limitation and check any content that AI produces to make sure that the tone is appropriate for the target audience. You don’t want to risk offending people because your tone isn’t what they expect! If this is something you struggle with, our blog will help you decipher the tone of your article.
2. Continue To Train and Develop
Developers are constantly training and fine-tuning AI models to better suit the market’s needs. If you’re developing your own AI model, you can work with your developer to customize the model, ensuring that its content aligns with your specific requirements and style guide.
But you should also continue to train your team on how to get the most from AI. This includes staying up to date with the latest advancements and best practices in AI. Staying up-to-date with ever-changing technology might sound like a nightmare, but Towards Data Science suggests five ways you can follow the trends in AI and technology.
3. Thoroughly Edit and Proofread Content Before Publication
Machines make mistakes, so it’s absolutely crucial that humans oversee AI content creation to ensure accuracy, relevance, and adherence to company guidelines. The first step includes checking for potential biases, errors, or inappropriate content that the AI model may have unwittingly generated. After that, the content should be proofread for spelling, grammatical, and punctuation errors.
At Proofed, we have lots of articles that outline the best proofreading practices and how to edit AI-generated content specifically.
Conclusion
With so much talk about the safety of using AI in content creation, it can be difficult to know who and what to trust. But the truth is that if you use AI to help you generate content, rather than relying on it to create content – and you thoroughly review the writing before you publish it – you shouldn’t encounter any issues.
Here at Proofed, our team of experienced editors and proofreaders has worked with more than 1,000 businesses to ensure their content is ready for publishing. If you’d like to see how our business editing services works for AI-generated content, schedule a call with one of our advisors today.
