It was big news when Shutterstruck reached an agreement with OpenAI two weeks ago to incorporate their text-to-image generator (DALL-E) with the Shutterstock platform, thus allowing users to generate royalty-free images directly on their website.
A major point of contention is fairly compensating the countless artists whose work is used as data to train these generative AI models.
Unfortunately, this conundrum is about to get way bigger because the Shutterstock partnership is only a small step compared to the giant leap that OpenAI announced.
Shutterstock is no longer the only creative platform that will benefit from this AI integration.
OpenAI has launched a DALL-E API in a public beta. The API allows companies to add DALL-E's text-to-image functionality to their products. It will be rate-limited, to begin with. Customers will not be vetted for how they will use the technology. There are three resolution tiers, and customers will be charged per image generated.
Microsoft's new Designer app is one significant instance of this API. Fast Company called it “The Canva Killer” for its ability to make graphic design as easy as typing a search into Google.
However, other companies utilize this API for everything from interior decoration to clothing design. There are even people posting job board listings on Upwork looking for help making their own generative AI bots or integrating existing models into their websites.
Having easier access to these tools is great, but if you've ever used a generative AI, you know the power is in the prompt. The AI’s artistic output is limited by your ability to craft the right sentence.
Most people struggle with writing, and even more people struggle with creativity (Ray Bradbury has one antidote). Becoming proficient at AI-generated art requires both. Thus, we can’t expect people to know how to talk to the AI properly. Giving people greater access to these tools (through platform integrations) won’t magically replace all of the visual artists of the world.
OpenAI knows that prompt writing is the limiting factor in making DALL-E an everyday tool for millions of people. That's why they made an 82-page book on how to write AI prompts.
Personally, I'm spending a lot of time researching how I can improve my AI prompt skills.
Obviously, everyone isn’t going to dedicate the time to learning this skill. Outsourcing creativity is already a common behavior. Thus, the job role of Prompt Designers will grow and marketplaces for prompts (such as Prompt Base) will become increasingly useful.
The Use Cases
What it takes Beeple an hour to create in Photoshop you can do in seconds with DALL-E. He knows this and has leaned into this paradigm shift. From time to time, he will host AI-generated competitions where he asks his followers to make something similar or better than his daily art piece. And the top three winners get a signed Beeple work.
There’s a hierarchy of ability in every creative industry. This field is no different. We already have an emerging class of “artists” who excel at writing prompts for DALL-E, Midjourney, and Stable Diffusion.
Below are a few of the top prompt writers for each tool.
Overall, the AI-assisted creativity field has been one of the most discussed technology advances this year because of its implications. It’s a developing industry that I plan on continuing to cover the possible futures of.