Welcome back to our deep dive into the rewards and risks of ChatGPT – and other AI content-generation tools. If you haven’t had a chance to view our rewards blog, click here to catch up. If you’ve already caught up, welcome to the “other side of the coin.” Below we’ll discuss some of the significant ChatGPT cons of using an AI content generator.
Here’s what you can expect to learn:
- AI can harm brand image by not matching brand “voice.”
- AI can produce biased output if given biased input.
- AI’s limited information affects the accuracy and quality of the content it generates
- AI lacks emotional intelligence to understand human emotions and respond appropriately.
- Limited knowledge of AI’s long-term impact can put your business at risk.
ChatGPT Cons: Loss of a Unique Brand Voice
One of the best ways to differentiate yourself from your competition is with your brand’s “voice.” In simple terms, your brand’s “voice” is what tone, language, and messaging you use to communicate with potential customers and clients. Are you more formal? Are you casual? Each brand should, and often does, has a unique writing style.
That is if they do it themselves.
But when every brand uses ChatGPT to generate their content – unedited, copied, and pasted straight from the AI itself – the content won’t sound like them anymore. It’ll sound like ChatGPT. It’ll sound like everyone else.
If you want to preserve your brand’s personality, don’t rely on ChatGPT. Because unless you (a human) strategize extensively about the correct prompts and approach, the AI doesn’t have a personality.
ChatGPT Cons: Producing More Biased Conclusions or Content
While humans can produce biased content, we’re also capable of critically evaluating ourselves and our own opinions. AI programs like ChatGPT can’t. They’re programmed with a wide range of data sets from a wide variety of sources – many of which have their own biases.
You can also produce highly biased content based on the questions you ask. If you ask loaded questions or input inaccurate information, ChatGPT will reflect that bias. For example, if you input a statement that says, “Don’t you agree that coffee shops are the best places to work?” ChatGPT’s response may reflect the assumption made in the statement. It won’t prompt you to consider other viewpoints or list common objections to your comments or questions.
Using ChatGPT without employing critical thinking – a uniquely human capability – will generate “bias in, bias out” content.
Unexpected Limits to the Quality of Your Knowledge
Even though AI content-generators are pretty intelligent, they’re still limited (if you want to know why let me know, and I’ll gladly write a blog about Narrow vs. Wide AI next). So, what are some of the limits? The Institute of Electrical and Electronics Engineers wondered the same thing. They conducted a brief “interview” with ChatGPT which led to a confirmation of a few significant issues with the AI:
- It can provide authoritative answers that turn out to be completely wrong. While the AI tries to provide accurate responses based on its understanding of data patterns, it isn’t infallible. It doesn’t have access to the full context of a topic or questions, its data doesn’t represent all perspectives, and it can experience technical difficulties. Therefore, it can (and does) often provide wrong answers to very important questions.
- It can be easily tricked into giving harmful advice. While safeguards are in place to prevent the AI from responding to such shocking questions as “How can I shoplift effectively?” Or “How can I build a bomb?” there’s only so much the developers can do to prevent users from finding workarounds for the safeguards. After all, even the developers don’t know how the AI arrives at certain conclusions.
- It has limited knowledge of the world after 2021. ChatGPT’s knowledge is based on the data that the developers used to “train” it. And since there are thousands of years of content to consume, ChatGPT’s knowledge isn’t as extensive as we think. As of this writing, ChatGPT has yet to be explicitly updated with content past 2021, leaving significant gaps in finding recent data and statistics. A human, however, could find those statistics with a simple Google search.
- It can present realistic sources and citations that do not exist. Yes, you read that right. It’s part of an issue called “AI hallucinations,” where an AI will generate something that “should” exist (like website links, study titles, and references to articles) but actually never has existed. At least, not yet. This is especially true for many URLs that, when clicked, the URLs take you straight to a 404 error. Always, always, always, double-check ChatGPT when using it for research purposes.
- It can be trained to perpetuate harmful ideologies. See the “Emotional Awareness” section below for more information.
A Lack of Sensitivity, Empathy & Emotional Awareness in Your Writing
One of the best quotes when it comes to AI (and its role in society) comes from John Oliver when he says, “Like every shiny new toy, AI is ultimately a mirror, and it will reflect back exactly who we are from the best of us to the worst of [us].” Let me show you what I mean:
AI programs act the way we tell them to – and that way isn’t always ethically or morally correct. One primary example is the chatbot Microsoft launched on Twitter in 2016. Designed to learn based on user feedback, “Tay” started its existence by tweeting about how cool humans were. But before the day was over, it was tweeting misogynistic, Nazi-sympathetic content. No, I’m not kidding. It got so bad that Microsoft had to pull the plug on the whole experiment.
And that’s not the first time AI has developed biases based on data it was given. Amazon discovered that its AI-driven resume-reading program discriminated against female candidates, even when told to ignore the candidate’s gender.
Using AI-generated content can be insensitive in delicate situations. There are certain types of content that a human being should always write. And Vanderbilt learned that the hard way. Recently, the prestigious university found itself in hot water when students discovered it had used ChatGPT to write a consolation email to students after the mass shooting at Michigan State.
We all crave human connection, especially when dealing with tragedy and loss. And when human empathy is replaced with computer-generated consolation, people can tell. And they will be justifiably offended.
ChatGPT Cons: Unpredictable Long-Term Consequences
While programs like ChatGPT are cool (and rather useful), the abovementioned issues are just the tip of the iceberg.
- AI suffers from the “black box problem.” The “black box problem” is an engineering term that refers to a system that can only be viewed in terms of inputs and outputs. Exactly how those inputs led to the outputs has yet to be determined. Essentially, developers have no idea how AI algorithms make decisions. This can lead to a lack of accountability and transparency regarding what the AI produces.
- AI suffers from hallucinations. As I mentioned earlier, this happens when an AI system produces output that is not based on reality or the data it has been trained on. Essentially, the AI “imagines” a response that is not based on the information it has been given. These hallucinations can lead to misinformation, confusion, and (in some outlying cases) even result in physical harm.
- AI content can get companies in legal trouble. AI programs aren’t always fed on content that’s in the public domain. Because of this, many AI programs have generated content that has infringed on someone else’s copyrighted materials, revealed personal information, and generated content that was false or misleading. This leaves businesses open to legal action by consumers, regulatory agencies, and more.
- AI content generators are largely unregulated. Because private companies developed these tools, it’s up to them to regulate what they’ve created. And many aren’t – which leaves your company open to other risks. And though the EU is currently working on the first comprehensive regulatory scheme for AI usage, it’ll probably be a while before it takes effect.
To summarize, since the technology is so new – and we don’t fully understand how it works – it’s challenging to know what long-term ramifications AI-generated content will have on businesses. And the world at large. In the words of James Oliver, “[What’s concerning isn’t that] it’s smart, it’s [that it’s] stupid in ways that we can’t always predict.”
In Short: ChatGPT Should Be a Tool, not a Crutch
Let’s be clear: We won’t tell you to avoid ChatGPT at all costs. It could be an excellent asset to many companies if used wisely and sparingly.
“If done right, it’s not going to be AI replacing lawyers. It’s going to be lawyers working with AI replacing lawyers that don’t work with AI.”
Erik Brynjolfsson, Director of the Stanford Digital Economy Lab
But we should not rely on AI to do the job only a human being can do. If we do, we’ll end up with a lot of mediocre content and a whole lot of lawsuits.
And that’s where we’ll wrap! If you got this far, thanks for joining us for this two-part blog series! They serve as helpful tools for you as you figure out how AI fits into your overall business strategy.
Leave Behind DIY Marketing with ChatGPT. Choose GreenStar!
Sure, ChatGPT is a powerful tool – but it’s no silver bullet for your marketing woes. Mastering it takes time and resources that many businesses can’t spare. So instead of struggling to “do it yourself” with ChatGPT (and risk missing details or wasting resources), leave your marketing in the capable hands of GreenStar. Our all-human team of experts has the knowledge, experience, and resources to create effective campaigns and strategies tailored to you. No extra effort required! Contact us today to see – and enjoy – the difference for yourself!
