The Problem With Creative Entitlement
This Week In Writing, we explore how AI tools amplify the sometimes problematic relationship between creator and consumerThe world of generative AI has gotten pretty strange. With Adobe adding "generative fill" to Photoshop, people are using it to "expand" classic album covers and paintings.
I guess this is cool if you want to know what a computer thinks was going on behind the Mona Lisa. From a use case, I'm curious about how helpful this is to digital artists. Then again, I don't use Photoshop or pretend to be that level of a digital creator. So, I find the whole thing interesting. But, it is part of the larger issue of entitlement connected to generative AI.
For now, we're going to set aside the conversation about generative AI plagiarizing existing content and focus on this idea of generative fill -- using existing work as the base to remix, expand, or create something new. This isn't just about digital art and graphics. The same is possible with text, voice, and video. AI can be trained on any media and then used to produce new versions of that content.
Take the story of Stefanie Sun, a Chinese pop star. She hadn't released an album since 2017, which, apparently, wasn't good enough for her fans. They trained an AI on her work and created covers of classic songs deepfaked by Stefanie Sun.
While the coder who trained the AI justifies their actions by saying, "I wanted to listen to her sing other songs," Sun has a much more dire look at the situation:
My fans … have accepted that I am a passe singer faded into irrelevance while my AI persona is the current hot property.
Yikes.
The story of Stefanie Sun highlights the big question raised when training an AI to remix or generative fill existing work: is anyone entitled to another people's work?
Filling in the background of a classic album cover or painting is relatively harmless, but is it that much different from training an AI to replace a living pop star? I don't think so.
If you're a digital artist and use generative fill to expand your work, go for it. If you want to train an AI on your writing and use it to create more articles in less time, I guess that's fine? I don't actually know how I feel about that. Regardless, it's much better than doing the same thing with someone else's work.
I grew up with the dawn of the internet. I'm very comfortable with the reblog and remix nature of the internet. I remember all too well the old meme explaining, "everything is a copy of a copy of a copy of a copy..." And I believe the internet is a fundamental human right that should be open and available to everyone. That said, fans of a creator are not entitled to their creativity, time, experience, or expertise. What happened to Stefanie Sun is wrong and will only increase in the coming years.
We will face more of these issues as AI continues to evolve. Right now, there is seemingly no protection for creators. Regulations and copyright laws will take years to adapt and likely will never catch up to the exponential technological increase. Instead, creators must decide how and when to use these tools.
I'm still torn on the ethics of AI tools. I still think it's important for creators to learn the tools and determine how they can enhance existing workflows. However, as more stories like Stefanie Sun come to light, I'm moving more and more toward regulating and limiting generative AI. I guess the bigger question is about copyright and using one's likeness. If you're a lawyer, please comment and shed some light here.
You may have noticed I stopped using Midjourney to create featured images a few weeks back. While the decision was purely economic, I think seeing where some of this technology goes is prudent. Yes, it contradicts my early adopter nature, but I also don't want to cause another Stefanie Sun situation inadvertently.
Generative AI and large language models are accelerating faster than we can comprehend. We're truly in uncharted territory. Ultimately, each of us has to decide how and when to use the tools. Whatever you decide, I implore you not to use them to generate someone else's work. You're not entitled to them.
WordPress Takes on Substack (plus a Ghost update)
WordPress recently announced a new set of tools that enable paid subscriptions. The move is clearly designed to challenge Substack (and Ghost) directly and could be a big deal since roughly 44% of the internet is powered by WordPress.
Notably, this announcement is only for the WordPress.com platform and not self-hosted WordPress.org installs. Atomattic, the company behind WordPress, also announced a generative AI tool built into JetPack for $10/month -- you can loop back to the top for more feelings on that tool.
While this WordPress moves and their work on ActivityPub are intriguing, it's not enough to get me to switch back from Ghost. I'm really happy with my transition.
There have been a few hiccups along the way. I learned the email settings in Ghost are different for site emails and newsletters. Why? I don't know. But I sorted that nonsense out, and the first issue of my personal newsletter, This Just In, went out to subscribers from my website yesterday! It's now an exclusive for subscribers on my site.
There's still no plan to move This Week In Writing to my website. However, what I'm learning by sending out This Just In will pave the way for an eventual transition. For now, you will continue receiving This Week In Writing on Substack. However, if you do sign up on my website, you will likely also get it from there. It gives you the chance to pick your platform.
Personally, I prefer the way Ghost sends emails to Substack. I also like the lack of a 10% Substack cut.
I'll share more about my continued transition to Ghost in a future newsletter. Know that it's going very well. Though there is a steep learning curve to self-hosting Ghost, so it's not exactly for the faint of heart.
Do you have questions about the transition to Ghost? Let me know.