Imagine you have a pile of sand.
You remove a grain. Is it still a pile of sand?
You remove grain by grain until only one remains. Is it still a pile?
If not, when did the pile become a non-pile?
That heap paradox can serve you well today as the tension around AI-generative content grows thicker around the world.
Get Robert Rose’s take in this week’s CMI News video, or keep reading for the highlights:
Over the past couple of weeks, Italy became the first Western government to ban ChatGPT due to data privacy concerns. On the other hand, the British government rolled out a white paper detailing guidance for a “pro-innovation” approach to AI. In the United States, the Center for AI and Digital Policy petitioned the Federal Trade Commission to halt OpenAI’s release of ChatGPT models until safeguards are put in place. And an open letter signed by more than 10,000 tech leaders, mostly from academia and Elon Musk, urges a pause in any development of AI beyond GPT-4.
Most relevant AI news for marketers
But the most interesting news for short-term content marketing strategies came from guidance issued by the U.S. Copyright Office. It clarifies what constitutes ownable content – the work must be created by a human (as has always been the case.) Thus, anything authored or created by generative AI tools cannot be protected by copyright laws. So, if you create content using an AI generator, you (or your brand) do not own that creation.Guidance from the @CopyrightOffice reiterates what constitutes ownable #content: It must be created by humans via @Robert_Rose @CMIContent. #AI Click To Tweet
Now, if you consider the multitude of legal actions taken around AI content, including Getty Images vs. Stability AI and a lawsuit against Microsoft, GitHub, and OpenAI over their use of AI technology to create Copilot, you might think the courts will settle the AI-related cases quickly.
But CMI’s chief strategy advisor Robert Rose says no and uses the U.S. Copyright Office memo to make his point:
In the case of works containing AI-generated material, the Office will consider whether the AI contributions are the result of ‘mechanical reproduction’ or instead of an author’s ‘own original mental conception, to which [the author] gave visible form.’ The answer will depend on the circumstances, particularly how the AI tool operates and how it was used to create the final work. This is necessarily a case-by-case inquiry.
Now Robert is not a lawyer. He’s a marketing practitioner who’s talked to a few lawyers about the subject and found consensus doesn’t exist. So his advice to use on a case-by-case basis comes from his marketing perspective.
Contentious copyright call
“In very short order as an industry bridging content creators and AI technology, you will decide if the tool in question is true AI, operating from a true learning model or if it is fake AI simply scraping content and reassembling it,” Robert says.
Be careful in drawing your conclusions. Given the hype around AI, some nefarious solutions will pop up that aren’t true artificial intelligence and make it hard to tell the difference.
That’s why, Robert says, you need to know the AI tool’s learning model and how it may use your content. For example, Adobe Firefly only uses its stock image library in its learning model. No doubt that weakens its ability to do what other image-generating AI tools can do, but it may end up as a much safer application. Midjourney, though, uses the hashtag – #AllTheImages – to inform its AI learning model. Is that a problem? No one knows yet.Marketers should know an #AI tool’s learning model and how it may use their #content, says @Robert_Rose via @CMIContent. Click To Tweet
Frankly, it all may come down to how much a human changes the content.
The U.S. Copyright Office punts the issue. Its counsel concludes with a call for disclosure. As Robert says with more than a hint of sarcasm: “I’m sure everybody will comply with that … right?”
But Robert doesn’t blame the copyright office given how difficult, if not impossible, the task of determining when something becomes human-created vs. AI-created.
Like the grains from a pile of sand in the heap paradox, when does removing or altering the content change the pile of content from AI created to human created?
Choose the content to own
If your content marketing team blithely and proudly churns out blog posts, longer content articles, ad copy, or images created 100% by AI. In a weird way, the more amazing the AI-generated content, the riskier it becomes.
Robert emphatically explains why: “You. Don’t. Own. It.”
Instead, take a more effective approach to generative AI tools and use them for content you won’t care if it gets “stolen” or exists without copyright protection. It never was yours. Robert says to let those AI tools create those summaries, sales emails, short blog posts, FAQs, etc. And let your humans focus on creating content you want to keep (and own) only for your brand.Use #AI-generative tools for content you don’t care to own – summaries, sale emails, FAQs, etc., says @Robert_Rose via @CMIContent. Click To Tweet
And then, it won’t matter when a pile of content transforms from AI-generated to human-created because you’ll have two distinct piles for each origin.
HANDPICKED RELATED CONTENT:
Cover image by Joseph Kalinowski/Content Marketing Institute