It’s been an exciting first few months building Lore, and we’re so excited to share an update on where we’ve been, what we’ve learned, and our ambitious new mission to build a spaceship for human creativity.
Lore is a decentralized social network for generative art and storytelling, and a community on a mission to learn more about our humanity by exploring the deep, creative spaces of artificial intelligence together. If you are new to our mission, you can check out our crowdfund essay, follow us out on Twitter, or join us on Discord. You can also check out this video demo, which is live now at makelore.xyz.
Our pilot season of Lore has attracted a small, but passionate community of over a hundred storytellers, minting nearly a thousand NFT’s on the Polygon blockchain to tell beautifully illustrated stories spanning across a robot dreamscape, a collective exploration of the meaning of love, a shared journal, and a branching fairy tale among many others.
Along the way, we’ve gotten a lot of love, like:
We were also honored recently to be among the 35 finalists for Seed Club’s SC04 Cohort. While we didn’t make the final cut, the application process was a great opportunity to hone in on our community’s call to adventure:
We believe that the current explosive moment in artificial intelligence represents an opportunity for humanity to express ourselves and understand each other as never before and that the blockchain acts as a perfect vehicle to help coordinate our exploration.
The best metaphor we can think of for the technology that will allow us to harness these forces and navigate this unfolding new territory is a 🚀 spaceship 💨. The rest of this essay will focus on describing the space we see unfolding, explaining what we’ve learned about it so far, and outlining our mission to explore it.
So get in, fren, we’re going star cruising.
In our pilot season, we’ve observed four important principles about the unfolding space of AI and Web3, which have motivated us to pursue this vision:
The pace of innovation in artificial intelligence is unlike anything the world has ever seen, and will rapidly reshape our culture in ways we never thought to be possible. What we do today will contribute to outcomes beyond the reaches of our wildest dreams.
In our crowdfund essay, we analogized the current moment to the period just before the explosion of computer graphics, where Pixar built a world-changing company by betting that technology would improve.
Consider the progress in AI journal publications through 2020 as more evidence that something very big is happening:
Remarkably, the progress that we’ve seen in the past few months makes us more confident than ever in this diagnosis. Here’s a look at one of our earliest story blocks compared to a block minted during one of our recent weekly story jams (they are on Wednesdays at 2PM ET in Discord if you’re curious!), showing just how far the state of the art in generative imagery has changed since we started (it’s hard to keep up!)
This pace is only increasing. In December, OpenAI published a paper announcing GLIDE, a new model using diffusion (the AI method we’re currently using) to produce photo-realistic images. While the the complete training set for GLIDE is not yet publicly available, it’s fair to say that Lore images will become increasingly life-like in the coming months:
There were a staggering number of other innovations in AI in 2021 that are out of the scope of this essay which you can explore in this blog post.
Our ability to create such evocative imagery is enabled by a rapidly evolving computational understanding of human language. This understanding is captured most thoroughly in large-scale language transformers like GPT-3, which enable us to easily accomplish previously unfathomable things, like summarize an entire novel programmatically, or have a machine write code for you. We use GPT-3 to power our 🪄 magic-wand ✨, which automagically suggests sentences to make your story blocks more interesting.
GPT-3 and other transformers are neural networks that take large corpuses of text data (from the internet, books, and other sources) and create mathematical understandings of the connections between concepts. The connections are known as parameters, and GPT-3 has 180 billion of them, allowing for its staggering and human-like ability to understand and produce language. Its successor, GPT-4, is expected to use new advancements in computer hardware to achieve 100 trillion parameters.
We don’t know exactly what these advancements will unlock, but they seem likely to create richer relationships between text and other mediums like sight, sound, and motion. This makes us extremely optimistic for a future of infinite creativity.
In the words of GPT-3, who kindly offered to finish this section:
🪄This is the future of how we will interact with language: through machines that can understand the nuance and subtlety of human expression. ✨
For the first time in history, blockchain technology offers every human the potential to express themselves in a way that is simultaneously sovereign, traceable, accessible, and immutable, enabling a community-owned public commons for the expression of thought.
Hidden deep in the annals of Polygon is an unfinished, AI-illustrated copy of the Tao Te Ching, minted inconspicuously in the “favorite quotes” branch of Lore’s community journal.
Around halfway through the Tao’s second verse, after “note and voice make the music together,” the story branches, with one branch continuing with the poem, and the other, authored by Grammy-award winning producer and prolific Lore mischief maker Trapchat, goes down a wild rabbit-hole of crypto-music-related AI imagery he calls “Blockchain Beats.”
Nobody gave Trapchat permission to branch off of that verse, and nobody has permission to remove it from the chain. His unique expressions now live permanently in the collective consciousness for anyone to find and add their own own take. In fact, you can do that now, for free, in just a few minutes (but, of course, we didn’t need to give you permission).
Even if Lore doesn’t make it (relax, WAGMI) and our servers shut down, everything necessary to reassemble this delightful train of thought will remain on the Polygon blockchain for a future, very confused, alien anthropologist to analyze, or a next-generation Lin-Manuel Miranda to create a Taoist crypto-pop masterpiece.
This is a unique and powerful foundation for human expression.
Working in tandem, AI and the blockchain can allow every idea to serve as the seed for an entire universe, blurring the lines between creator and consumer with an infinite “yes, and” that respects each expression’s origin.
Each time an idea is published to a blockchain, it is an open invitation for those who are moved by it to add to it.
While, for most of human history, this type of invitation was only really valid for those with highly-developed skills, AI is rapidly lowering the barrier to entry for such types of creative collaboration. This abundance is deeply embedded into Lore’s core functionality (immediately after writing something, you’re rewarded with an AI illustration), and we believe it will be one of the defining features of the next phase of Web3.
Consider this entry from famous and very expensive NFT photo collection Twin Flames:
The image on the left, of course, is the original photo, while the one on the right uses the photo as an initiation image in VQGAN + CLIP, using the captioned poem as a prompt, creating a new piece of work using the emotion that the original created.
Here’s the same technique we later applied to another photo from the series using Disco Diffusion, riffing on themes of twin-hood and partnership:
This creative “call and response” can go on infinitely, enabling a type of branching, meandering exploration of rich expression that has never before been possible, and opening up rich new avenues of collaboration between creators and their fans.
When we use AI to combine individual ideas, new perspectives emerge that can be greater than the sum of the parts that created them. This principle of emergence provides an incentive for us all to contribute to the collective consciousness, and receive compound interest from what we help create.
One of Lore’s most beautiful first stories was a series of poems by errorgardener.eth about the changing seasons:
While prototyping with Chigozie Nri’s Zooming VQGAN + CLIP Animation pipeline, we were inspired to combine his words with house DJ Nora En Pure’s recent track, “Sign of the Times:”
The resulting work is greater than any of the individual components -- the song, the imagery, or the words, creating an immersive dreamscape which combines them all to stunning effect.
Here’s another example of combining some of the Twin Flames concepts above using Disco Diffusion:
There are many more examples of this type of emergence, and more are popping up every day. By giving AI little bits of ourselves through our expression, we empower it with the tools to create new work that is more than the sum of our parts. By doing this together, we can create art that takes from each of us to produce an original work of beauty that is meaningful to all who contributed.
We don’t know exactly what will happen in the future (that’s part of the fun), but know that the best way to satisfy our adventurous spirit is to help build it.
AI is opening up a strange new space of unlimited creativity, and Lore’s mission is to help us all navigate it together by building the tools to create, remix, combine, and distribute creative works together using the latest advances in AI and blockchain technologies. As described in many of the examples above, our first intended use-case of such tools is to empower fans to collaborate with the artists they love most by building on top of their work, and give artists the tools to curate what resonates with them.
Over time, we believe abundant AI will evolve the function of the artist from a creator of single works to a cultural curator at massive scale, taking general-purpose models and fine-tuning them with their creative spirit for exploration by their communities. The evolution of artist to AI shaman will have profound impacts on society, as AI will enable the scaled marriage of the aesthetic sensibilities of the artist with the lived experiences of her fans. More concretely, this means an overwhelming abundance of new artistic work created by communities armed with the palettes of the artists that best speak for them.
This shift will create massive value for society, and the blockchain will provide the perfect infrastructure to capture it inclusively, providing perfect provenance not only for the resulting work, but also the people and components that contributed to it. In an aspirational AI-blockchain-movie-of-the-future, records of all contributions will exist on-chain, from the model engineering, to the artistic curation, to even a single life story contributed by a community-member to the AI training set (“remember that time we flew to Indonesia on a whim to see Prince and met the Pope?!”). These artifacts will accrue value on their own based on their cultural significance, but will also allow for smart contracts that automatically send royalty payments that reward contribution.
That technology can prevent us from leaving each other behind is one of the most beautiful promises of the blockchain.
It will obviously take some time for all of this to materialize. In the meantime, we are working to start the creation/curation feedback loop as quickly as possible by finding communities of artists and their fans now and equipping them with the tools to get started. Working hand in hand with these communities, our goal in the near future is to establish a positive-sum NFT game that produces beautiful work that people want and allows everyone who contributed to share in the upside.
While we are open to all ideas for projects, we are especially interested in exploring some of the ideas prototyped above, including a community-produced music video with a recording artist, and remixing an existing NFT collection with its creator. If that’s you (or someone you know) please get in touch!
But first things first. In order to build this Creativity Spaceship and send it into orbit, we need help, starting with locating a crew.
Finally, there’s some other important stuff not covered in this note:
If you want to get in touch, join our Discord and say hi, or email us at we@makelore.xyz.
Thank you to the OpenAI team for world-changing innovation, to Katherine Crowson, Remi Durant, @SomnAIDreams, Chigozie Nri, and others for making magic possible with their tools, to every member of the Lore community who has inspired us with their incredible creativity, and to Ana Ugalova and Colin Bartoe for their edits. We love you all!