Generative AI is Convenient and Cool, but Make Sure Your Content is Copacetic

If you are on board with using generative AI as a content creator, good on ya. I am, that’s true. But I am an ethical person and I do my due diligence. Are you those things? Because if not, I’d steer clear or else definitely don’t use it in your professional sphere. There are serious legal concerns all the way around.

Sam Altman himself offered to the U.S. Senate that our government should put some serious guardrails on artificial intelligence before it jumps the berm entirely. Open AI’s most popular product, Chat GPT, has boomed in popularity, quickly becoming a household word before most people know what it really does. Users confuse it as a hybrid creator and google search tool rolled into one. It is one of a collection of new tools that fall under the umbrella of generative artificial intelligence, capable o creating text content, images, and computer code strings based on human-entered text prompts.

Many small companies toyed with the tool by letting their teams explore content generation using the generator, and soon large systems like Mailchimp were on board, experimenting with marketing and promotional copy strings.

The hangup here is mostly that generative AI tools are so new that they are tricky. Not tricky to use or “play with.” They are so easy in that regard that anyone can use them. They are astonishingly intuitive, really, but that there are some dangers built-in that the average user can fall into without some attention.

First, generative AI is not a fact-checker. Human beings are not super-efficient at this, but we can be diligent. And we must be. AI text generators can churn out content that sounds amazingly convincing, highly refined, and yet in the end it is false. The algorithm is created to produce a set of words that are most likely to occur after a prior set of words. It’s important to understand that. It is not designed to generate the set of words guaranteed to be the most truthful or accurate after the prior words.

Therefore, if I were still in higher education, I would insist that my students could use AI generation tools the same way that I allowed them to use Wikipedia – as the starting point, but absolutely not the ending point. Wikipedia has always been a community-curated information source, prone to accidental misinformation by dint of its openness to free editing. If we think of AI generation in a bit of the same way, we know that a human fact-checker is required to follow up, every time.

Another pitfall is in the sourcing of generative AI. Eager users who want to create new work may be inclined to protect their work, and in most cases that is totally fine. Encouraged, even. If you write a new play, poem, or novel you should definitely copyright that and keep all the rights to it. But if you asked a generative AI tool to do most of the work for you based on a prompt you gave it, and you sat back while it spat out 20,000 words, you should (must?) disclose the artificial nature of your work to some extent. As of March, 2023, the U.S Copyright office says that works of art, music, or writing where the human generates the prompt but the work itself is the product of an A.I tool is not generally going to be covered by copyright law. This is good news for all of the human creators out there who have been doing some (deserved) hand-wringing over what might become of the original art in the world. It won’t take a whole lot to separate the wheat from the chaff, as it were.

And while we are on the topic of legality, it is wise, dear creators, to not go about willy-nilly creating images of celebrities and other public figures that are wholly realistic and palming them off as the real deal (cough, cough paparazzi style) or even trying to mimic existing artists’ styles. There have been cases wending their way through the courts already ferreting out the rights of writers, photographers, and painters (along with poets and songwriters, soon to be more) that outline the rights of what those creators can retain in style, not just substance. So a new creator won’t be able to claim that something is “theirs” when they asked a generative AI tool to write “in the style of Taylor Swift” or to write a short story “like Stephen King.” It won’t make them any money anyway.

My last word in the area of protecting and legality (and there are many, many more last words – I just can’t write a whole book here) is this one: protect your IP. Your proprietary information needs to stay just that. If you put your information into a large model system (like Chat GPT, for instance), it stays there. That is how the whole thing gets to grow and “learn.” If you put proprietary information in so as to generate a cool string of output, that information is retained by the generative tool and becomes part of the pool of larger information – forever.

When I was in grad school, it was the start of using the plagiarism checker “Turn It In.” This software aggregated a bank of student papers against which to check other student papers So the gist was that I was supposed to turn my paper in electronically and it would scan my work against thousands of other papers to see whether or not I had plagiarized. It would see if I had grabbed from the internet, or other students’ work.

I refused to submit my paper to the program unless I was fairly compensated for my content. I wrote a short argument paper to my professor stating that the company, Turn It In, was profiting because the University bought their software, and relied upon my work and the work of other students, but they offered me absolutely no assurance that they were not keeping my work in their repository. The only way that the company could remain successful was to continue to “bank” student papers to check other student papers against. I told the professor that he was of course welcome to self-check my paper by entering any number of phrases into Google or any other search engine but that I did not authorize Turn It In to retain my work for their purposes.

He loved my moxie, and allowed me to retain the rights to my academic work.

I am not a professional writer and editor – I retained my proprietary information and did not authorize its use elsewhere.

Use generative AI. Figure out how it benefits you and your company. Build your own in-house tools. The algorithm is not difficult. It’s fascinating, it’s useful, and it is powerful.

But be careful and be ethical. Be responsible for what you do, and be mindful of your customers and their information as well.