3 Reasons to Be Careful with Generative AI

You might own or work at a content business, or you’re an employee responsible for producing content for a business. You might even be a student or intern under a tight deadline. Let’s say you need to craft a health information piece in email format for patients who meet a target profile, but the information you’ve been provided about the disease state is overwhelming. You don’t have time to go very deep but you’ve been picked to create the content because you get a much higher rate of engagement (click-throughs, email open rates) compared to when SMEs (subject matter experts) write patient-facing content.

ChatGPT helps you out of your research overwhelm and writing slump. It quickly churns out a piece that you feel you can modify with enough original language. This isn’t bad – you are using it like a tool. But do you know enough about AI to be confident sending that piece over to your manager or turning in a paper? Here are 3 reasons you need to know and why you should run your content through AI Detector Pro before you press send.

AI Spreads Misinformation by hallucinating

Scientists say the correct word is “confabulates” but saying AI hallucinates is more fun. An AI hallucination is when the content you were given was completely made up. But the AI says it so confidently, you take it as fact. No amount of language reworking will make up for releasing misinformation. Companies are racing to fix this problem but AI technology is outstripping the solutions. That’s why you want to know what language in your piece would be flagged by an AI detector.  When you see a note the flagged language and your report score, you should escalate to your manager and ask for an SME to look it over and determine whether the content is factually correct.

If an SME agrees the content is factually correct, rework the language to be more engaging. However, unless you have the technical knowledge of the SME, you will never be able to figure out what is right vs. what’s wrong. You need to pinpoint flagged language to SMEs based on algorithmic detection to escalate effectively. SMEs also have deadlines, so they need to know what language is the most problematic. If you’re a student turning in a paper with factually incorrect information, AI Detector Pro would give you a chance to correct it. How would you explain to the professor that you turned in a paper with mistakes in it?

AI is biased

Many executives are excited about generative AI, without completely understanding how it works. AI generates content based on a Large Language Model (LLM). This means the LLM is going out and finding information across the internet to generate answers. It doesn’t think about that language or even question it. It just eats information and then spits out answers when asked. That means anyone could have produced the answers it gives you.

At a time when we are worried about systemic equity across many dimensions, do you want to release content that contains inherent bias? No number of prompts you read about will be able to help you get rid of the bias that was built into the original data. We noted this in our last blog for students. If you have a risk report that flags that language for you, you can think analytically (something an AI can’t do) about biased language before proliferating it and unnecessarily upsetting a lot of good people.

AI trades good content for fast content

The Cautionary tale of sabermetrics in baseball

Have you seen Moneyball with Brad Pitt and Jonah Hill? Brad Pitt plays Billy Beane, the General Manager of the Oakland Athletics. Billy decides that the traditional way to recruit into MLB could be reinvented for smaller clubs to be more competitive. While on a scouting trip, Beane meets Peter Brand, played by Jonah Hill, an economist working with the Cleveland Guardians.

Brand has innovative new ideas about how recruiting could be improved using statistical analysis. Billy recruits Peter to the Oakland As and together they develop a technique, now known as sabermetrics. When used by the Oakland Athletics to recruit using Brand’s innovative new methodology, they are able to compete well and win the American League West series, eventually losing to Minnesota in a different series. The epilogue tells us that John Henry adopts sabermetrics for the Red Sox, who win the World Series 2 years later.

Why are we talking about a baseball movie? Because it’s a cautionary tale. The hype around Generative AI for marketing would have you believe that if you don’t adopt it instantly, you will lose all your competitiveness. What happened in the case of sabermetrics is that once everyone adopted it, the method lost its edge and everyone ended up in the same place. When everyone uses AI to generate content, it’s going to make content boring. That’s when original content producers will regain their edge.

Use Detection to stay unique

If you want to make good, unique content, use a good detection technology to figure out what’s boring and reads like AI. If you’re a small independent marketing firm or an in-house marketing employee, we encourage you to use AI wisely. The best use of AI text generators is for content that doesn’t require a lot of creativity. But invest in a detector like AI Detection Pro to keep your edge. You may think you’ve reworked your GPT content to make it better but you want to be sure, line-by-line, what sounds generic. We can help you do this, and save time on figuring out what sounds unique and what reads like AI. Don’t fall prey to the next generation of scroll and deletes.   


We’re currently moving so fast with generative AI that businesses can’t correctly compute the risks and liabilities they’re incurring. Everyone is excited about speed-to-market and efficiencies and going to scale. Eventually the partying will stop, because it always does. Education has already invested in AI detection technology, started developing AI policies, and now flag and penalize students for exceeding limits on their campus policies. A big reason students use AI Detector Pro is to avoid being falsely flagged on original work. Businesses won’t be far behind. Look at our advice to students so you can begin shaping corporate policy within your workplace. Many of our clients use AI Detector Pro because we keep up with the latest versions of Large Language Models.


Leave a Reply

Your email address will not be published. Required fields are marked *