I have a love/hate relationship with ChatGPT.
On the one hand, it certainly saves a ton of time. I’ve used ChatGPT to create social media posts, usually editing what the bot comes up with by tweaking the wording. When I’ve asked ChatGPT to do some research for me, the results are quite impressive.
One interesting discovery with this powerful chatbot, though, is how generic the results can be. Many of the “articles” are no more than a series of generic comments all strung together to look like they are worth reading. No personality, No insight.
Deploying ChatGPT in your business could lead to some frustration, so as long as you know what they are, you can safely proceed. But thinking this chatbot is going to solve a lot of issues, save you money, or improve efficiencies at every turn — in other words, believing the hype about AI — could be a lesson in how to deploy technology.
The first point to make here is — make sure you move slowly and deliberately. One example is with those social media posts I mentioned. They do not exactly jump off the page, and I’ve even gotten to the point where I can recognize a post that was likely created by a bot. CUstomers of any business will eventually learn to detect when a bot is involved.
With the written content, it;s far worse. The generic content you might want to use reminds me of a stock image. Some are interesting and worth using, but many stock images just lack any real value. It’s one reason this classic YouTube video is still funny.
Having that same mentality with ChatGPT can tell. Is the content really worth using? Does it merely give you a starting point? Will customers know it was created by a chatbot? It’s always important to evaluate a deployment strategy and know what you are getting your business into and why. Some of the content you can create has value, some will only make your firm look generic and like you are cutting corners.
Another thing to consider about using chatbots is that they only produce content when you are specific and provide prompts that make sense. In testing ChatGPT recently, I asked the bot to write an article about Jeff Bezos. The results were bland. When I asked the bot to write an article about all the companies Jeff Bezos has started and runs, it was much better.
Then, I went further and asked the bot to write the article like something The New York Times might create, and specified the word count. It was still too generic and dull, but it was better. When you use smart prompts, you receive much better results.
The issue for many companies is that these apps and tools are so easy to use. In mere minutes, you can conduct research on a new product, or write articles and social media posts. When you spend extra time, customize the prompts, and then use the content only as a starting point for what I would call “real work” you’ll discover a lot of value. After that, run on all cylinders.