Generative AI & Web3 🖼️

This weekend, December 2nd-3rd, is the 20th anniversary of Art Basel–an experience that is considered by many to be “the epicenter of modernized art and prestige.” It’s a time where everyone from renowned artists to athletes to tech executives, from across the world descend upon Miami, the newestweb3 hub, to immerse themselves in a weekend of art exhibits, showcases, musical experiences, and (of course) parties. 

This time last year, NFTs were all the craze. While this will still likely be a core part of exhibits focused on digital art, another form of art has emerged and it’s here to take things to the next level. 

Introducing Generative AI. 

In a previous post, we broke down the role of AI in web3 at a high level. As a refresher, AI is short for artificial intelligence - which is basically a form of programming for computers to complete human tasks, like language processing and even learning. So much so that people, and many movies in Hollywood like Ex Machina or Her, make us wonder if AI will replace humans? Honestly, only time will tell honestly, but as of the past few months, we’re seeing a trend that AI can be used to create content like humans do. 

Welcome, Generative Artificial Intelligence. 

Generative AI is the unsupervised and semi-supervised machine learning algorithms that let computers take existing content (read: text, audio, pictures, and more) to create new content. Imagine feeding your AI images of a few of the world’s most renowned paintings, and it spits out its own original image. And not just a poor man’s attempt at merging the images, but an actual piece of new content that looks pretty legit. 

Let’s explore some of the models that make this a reality.

• DALL-E: is a new AI system that can create realistic images and art from a description in natural language. Check out the search below done by Kendall – you’d think she painted these herself… or not LOL. If you’re interested in creating an account and exploring what you can make, sign up here!

• Stable Diffusion: this similar to DALL-E in the sense that it can generate images from text, but the one key differentiator that’s made it popular to generate AI art is that it’s open source and can be built upon. For example, the YouTube VFX site Corridor Crew showed off an add-on called Dreambooth that allowed them to generate images based on their own personal photos.

• Midjourney: this is another text-to-image generator that uses AI to create images from textual descriptions. The key differentiator between this model and the others is that Midjourney creates art that is more focused on creativity and free imagination, whereas the others generate art that is more realistic.

The cool thing about generative AI models is that you can do more than just make 2D pictures… you can also create 3D art!

Get3D makes it easy for anyone to create 3D content. 

In early 2022, NVIDIA launched Get3D–an AI model that uses only 2D images to generate 3D shapes with high-fidelity textures and complex geometric details. As we’re sure you can guess, the typical 3D workflow requires some serious skill and a lot of time to execute on. You couldn’t just wake up one day and have a 3D masterpiece… until now! Get3D makes 3D content creation so much easier for everyone –from 3D specialists to regular people like us. Not that any of us are regular, but ya know what we’re trying to say lol. 

As we move into a world where augmented reality (AR) and virtual reality (VR) become more mainstream and widely accepted, tools like Get3D have the potential to democratize access and disrupt 3D content creation in the same way that website creation platforms like Squarespace and Wix did for website development! 

Web3 artists are taking full advantage of this creative medium.

Leveraging AI models such as the four listed above, artists can use the algorithms to automatically generate digital content based on a simple text prompt that would typically take a human a while to complete. This allows Web3 creators to create mass collections of unique pieces in a fraction of the time. It’s a win-win for the artist because they reduce the cost that’s required to make a large number of pieces, create pieces that even their own mind may not have imagined using the machine’s creativity, and then they get to sell this art as part of their collection.

…or do they?

This brings up one of the biggest questions related to generative models: Who owns the final product? Is it the person who puts in the text, or the model that creates the output?

Recently, Getty Images announced their decision to ban content created by AI models like Midjourney, DALL-E, and Stable Diffusion because of copyright and ownership uncertainty. Even though big players are being cautious about their approach, there could be an opportunity for blockchain technology to track the owner of certain pieces and reward the original creators. 

Who’s leading the charge in generative AI?

Not everyone is using Generative AI to create art, but companies are definitely using this technology to create new tools for their customers. At the big-name company level, we’re already seeing some moving and shaking:

• Amazon (Polly): a tool that turns text into speech, ranging from a simple algorithm to even having your own unique personalized voice for a brand

• Microsoft Github (CodeAssist): a computer program to assist programmers in finding software to help fill coding gaps

• Amazon (DeepComposer): helps the user turn a short melody into a complete song

And if you thought that was cool, wait till you see what some startups are working on:

• Respeecher: uses voice cloning (yes cloning) technology to for advertisers in entertainment and video games

• Rosebud.ai: helps take an idea and explore what it can look like in full form by taking simple text and build models of humans or even worlds that match the request

• Rephrase.ai and Synthesia: used heavily in the advertising business to create customized/personalized sales pitches with stock models, even celebrities

What does this mean for you?

Did you really see the Mona Lisa in the Louvre? Did Drake use an AI ghostwriter to make songs for his latest album? Did we really write this newsletter? (we will never confirm nor deny this) 

If you’re asking yourselves these questions after reading our newsletter, we don’t blame you. Even in the past couple years, we’ve seen what deep fakes have done in the realms of politics and the justice system so we can’t ignore the risks at hand. But beauty will always fall in the eyes of the beholder. As the Generative AI gets more and more advanced, we will have to start taking a little extra time to see who made certain digital products. At the same time, the range of opportunities we can see from this technology is vast, and that’s exciting! So until then, hopefully you can try out some of these cool Generative AI technologies we mentioned and see for yourself what its future has in store!