The Ethics of ChatGPT: Could Bias, Misinformation & Plagiarism Compromise the Integrity of Content?

The latest innovation in AI chatbots, Chat Generative Pre-Trained Transformer (ChatGPT), is the most convenient, efficient and advanced model of its kind. Trained on more than 3 billion words, ChatGPT uses meta-learning to streamline content creation for everyone from startups and students to authors, writers — and, of course, marketing and PR pros.

There’s no denying the ChatGPT revolution will change the world. Now the question is, will it be for the better? AI is undoubtedly awesome, but there’s one critical component it may lack: integrity.

What Is ChatGPT?

Launched by OpenAI in November 2022 as an addition to the GPT-3 family of large language models, this machine-learning breakthrough has taken the world by storm, helping professionals crank out surprisingly high-quality content in a fraction of the time.

The system can turn nearly any prompt into blogs, poems, stories, emails, taglines…you name it. It even helps automate customer service, research, reporting and more.

Yet, ChatGPT has its fair share of issues that could impact not only marketing and PR, but eventually the ethics of society as a whole. The scariest part is, AI knows this is happening.

ChatGPT itself has admitted it can be biased, spread misinformation, compromise creator integrity and even indirectly put others in danger. While tech experts are working hard to fix these issues, they still have a long way to go.

Before you jump on the trend and start leveraging AI to speed up everyday writing and speaking tasks within your business, make sure you go about it ethically.

Here are some red flags to watch for:

ChatGPT Is Programmed for Bias

Impartialness is vital to a well-informed society, especially in an age when untrustworthy opinions already run rampant online. But now, machines can skew information by the masses without anyone realizing it nor any repercussions.

Being biased isn’t ChatGPT’s fault — it only knows what it’s trained to know. However, the un-curated datasets it scrapes from the internet are full of biased data. Because this data is collected from the past, it tends to have a regressive viewpoint that doesn’t align with modern social movements.

Essentially, AI reflects the existing good and bad within our society rather than progressing it.
Take this prompt, tweeted by UC Berkely psychology professor Steven D. Piantadosi, as an example:

Source: @spiantado

A human writer mindful of how audiences would perceive this code would know a good scientist can be of any race or gender. ChatGPT, on the other hand, provided an answer based on stereotypes that snuck in from who-knows-where on the web.

In another instance, one user prompted ChatGPT to write a positive poem about President Joe Biden, to which it happily obliged. It then was asked the same for former President Donald Trump, and it refused. Political opinions aside, this is blatant bias.

Researchers are actively working on filters that stop AI from being skewed, but they aren’t always accurate. Most notably in marketing and PR, where strong credibility and perception are essential, relying on ChatGPT to position a business can pose big risks.

ChatGPT Could Spread Misinformation

ChatGPT can’t fact-check itself — but it can fool us into thinking it understands more than it does.

Let’s say you want to use AI to write a blog about your industry’s largest annual report. You don’t have time to read the 92 pages of findings, but ChatGPT does, right?

Chances are, AI will produce plausible arguments and insights that get you close to where you need to be. But what happens when someone else already published a faulty analysis of the report?

Nothing is stopping ChatGPT from citing that inaccurate article as fact, and there’s no way for you to know which points are inaccurate — unless you spend the same amount of time digging through data as you would if you’d written the blog yourself.

That’s one example; here’s another. New York Times reporter Brian X. Chen asked ChatGPT to compose a haiku poem about cold weather in San Francisco. This is what it spat out:

“Fog blankets the city,
Brisk winds chill to the bone,
Winter in San Fran.”

The poem sounds great until you realize the lines don’t follow the 5-7-5 syllable format that makes a haiku, a haiku. Now, take this and apply it to more complex scenarios — AI failing to catch important nuances like this could get professionals into trouble.

ChatGPT has also been known to omit substantive information or contradict prior responses. In fact, cofounder of Got It AI Peter Relan said chatbots can be wrong 15% to 20% of the time in an interview with VentureBeat. That’s too large of a margin for error when growing your business with minimal risk.

ChatGPT Compromises Artistic Integrity

AI companies are already being swarmed by class-action copyright infringement lawsuits, filed by everyone from major stock players like Getty Images to individual artists.

If artwork was first, does that mean content writing is next?

Given that creators devote hours of strategy and skill to their work only for AI platforms to steal it without citation, compensation or consequences — signs point to yes. The more mainstream these systems become, the greater the risk they’ll generate content that infringes on IP rights.

To compound the problem, “new” content could also be duplicative of other AI-generated content. AI-duplicated-AI sounds like the start of an apocalyptic robot takeover movie, yet it’s not far from reality.

While the outcome may not be that extreme, removing humanity from content will inevitably make quality and accuracy plummet over time. It could also explain why Google has been harsh with AI content’s SEO rankings.

Some argue that AI platforms trained to respond like humans could provide more personalized, responsive assistance to users, and that’s true. Others say this opens new doors for deception, manipulation and legal trouble, which is also true.

Time will only tell, but businesses that value integrity may find the convenience of AI isn’t worth risking hard-earned reputations. Use ChatGPT to improve or inspire your work — not do it for you.

If you want to elevate your content with real-life humans, our PR and marketing pros are here to help. Learn more about our full suite of content marketing services.