26 Sep Don’t be Artificially Intelligent
By Chuck Gamble
Learning and Develop Manager
ChatGPT is a powerful tool when you use it correctly, but the problem for many people right now is that they are not using it correctly.
A New York attorney named Steven Schwartz recently learned a tough lesson when he placed his implicit trust into the Large Language Model.
Schwartz, representing a client suing an airline over an incident involving a serving cart hitting the client’s knee during a flight, used ChatGPT to research cases similar to his that he then cited in a filing brief. The problem? The cases that ChatGPT cited didn’t exist, which was quickly pointed out by the attorneys representing the airline. Not only did Schwartz and his partner lose their case, but Schwartz was sanctioned and fined by the judge.
This brings me to a point: always remember that the “A” in AI stands for “artificial’”.
ChatGPT spat out artificial results by providing this lawyer with fake cases, and if you rely too heavily on ChatGPT and use it for your writings, you run the risk of sounding artificial yourself.
I have always encouraged writers to use their own voices no matter how they appear to others. An AI or LLM doesn’t know your voice; ChatGPT doesn’t know how you think or speak. It only works by predicting thoughts and facts.
After reading a recent Linkedin post from the CEO of my former company, something caught my eye.
I had worked closely with him for more than 20 years, meaning I’d read every email he sent to my clients and myself, I’d read every blog he penned on our company website and I’d had countless conversations with him. I knew his voice and his writing style inside-and-out.
It was obvious that he didn’t write the post.
I also assumed the company’s marketing department drafted it for him, but I had worked extensively with that marketing department for years, and I knew the social media director well — it wasn’t her writing style, either.
The post just sounded — here’s that word again — artificial. So, I opened ChatGPT and entered a prompt for the subject of the post. Lo and behold, ChatGPT responded with an answer that was 95 percent verbatim to what the CEO’s LinkedIn post said. It had been copy-and-pasted directly from ChatGPT, no matter how fake it sounded.
TimelyText has employed and placed hundreds of instructional designers, technical writers and corporate communication professionals in North Carolina since 2003. Whether you’re looking for an expert to join your team for just a few months or someone to stick around permanently, we can help you find the perfect fit. Submit a contact form or reach us directly at firstname.lastname@example.org to discuss how we can help. Follow us on Linkedin for more updates!
There’s the danger. When you rely on ChatGPT — or any other AI program — to create your writings, you lose your voice, your creativity, your authenticity, and even your critical thinking skills. Critical thinking is a key trait of intelligence, and that’s what the “I” in AI stands for.
Going back to the attorney, Schwartz, he believed his filing was accurate. He believed he filed an intelligent brief that would support his case, and he was badly mistaken.
So, how can you tell by reading something if a person is using their real voice or AI? Good question.
Let’s try it together. I’m going to write two paragraphs below on how to dress for a job interview. One paragraph will be from ChatGPT, and one will be from my perspective. See if you can guess which is which.
It is wise to dress professionally for any job interview, no matter what the role or company is. For men, that means a suit, or at the minimum, a shirt and tie with dress slacks. Women should wear a matching suit or dress outfit. Both genders should be sure their hair is well groomed, their shoes are clean and appropriate, and jewelry should be kept to a minimum.
Dressing appropriately for a job interview is crucial. Consider the company culture, but generally, opt for professional attire: a suit (or business appropriate dress), well-groomed appearance, and minimal accessories. Dressing conservatively is usually a safe choice.
Do you know which one I wrote? The correct answer is Paragraph One; ChatGPT created the second one.
What stood out about mine?
- It’s longer. If you’ve gone through this blog, you’ll notice that I tend to write longer paragraphs.
- When I wrote “well groomed”, I didn’t put the hyphen in it. ChatGPT did.
- I used the words, “no matter” in my paragraph — words I’ve used twice earlier in this post.
Those are the observations you should be aware of when trying to determine if someone uses ChatGPT or other AI products to substitute for their work.
Now let’s talk about how you can use AI correctly in your writings. I will admit that I use ChatGPT to assist me, but I only use it as a way to get myself unstuck. (Writer’s block is a real thing!)
If I’m stuck for what to say on a subject, I will open ChatGPT and enter a prompt on the subject I’m working on. Of course, it will answer me. I’ll look at the response and think about how I would phrase ChatGPT’s answer in my own words.
That gets me unstuck, and it’s a productive way to use AI.
How do you use AI to benefit you the most without losing your voice?
Chuck Gamble is a a learning and development manager at Safe-Guard Products International, and he has more than 20 years of experience in the instructional design and training world. You can follow him on Linkedin here, where he shares L&D insights and other slightly less serious observations.
Would you like to share your expert perspective on instructional design, technical writing or corporate communications? Contact us here!