Leverage AI to Identify Valuable NFTs
Last Updated on January 15, 2022 by Editorial Team
Author(s): Living Opera
Originally published on Towards AI the World’s Leading AI and Technology News and Media Company. If you are building an AI-related product or service, we invite you to consider becoming an AI sponsor. At Towards AI, we help scale AI and technology startups. Let us help you unleash your technology to the masses.
Don’t rely on market exuberance— leverage machine learning and AI to make the most of the Web3 revolution.
We’re in the midst of an NFT boom, but that won’t always be the case. Today, NFTs are being flipped quickly — much like house flipping in the lead up to the 2007–08 financial crisis.
Obviously, that doesn’t mean that NFTs are all driven by speculation, just that we need to be cautious and prudent when evaluating their value. Artificial intelligence (AI) is one tool for helping identify and produce valuable art NFTs. Let’s dive into that more here (but see Christian Jensen’s recent article for a broader background on the investment lingo in NFT land).
AI and NFTs Today
Generative Adversarial Networks are a particularly powerful set of techniques for producing realistic images because they embed two important steps — a generative and discriminant process — that involves creating simulated examples of the “right answer” (according to the model) and a classifier for whether the example truly fits the data. The concept is based off a game-theoretic framework where the generator network is pitted against the discriminator network, battling it out.
And yet, the strength of these AI algorithms depends in large part on the availability of big data and computational power. Fortunately, these ingredients are already available from the Web2 revolution. For example, large reservoirs of images already exist and companies have plenty of cloud computing resources available.
In that sense, if you want to build a GAN to produce art NFTs, then it’s not hard to train a new classifier based on already-classified images with some slight alterations — that’s where the randomness comes in with generative art, which sets forth some basic structure with another element of randomness.
New Possibilities with AI
One of the limitations with a lot of NFT art is that it’s just a “nice image.” But, nice images are easy to produce at scale… so, if supply can expand almost no cost, then unless demand also expands rapidly, price goes down and the only NFT projects making money are those that are differentiated in other ways.
My personal view is that the art NFTs that will hold lasting value are the ones that tell stories. Nice pictures are “nice,” but are they going to preserve their value when fads change — or will they even beat out the noise with the constant churn of new NFTs that are minted each day?
The art NFTs that will hold lasting value are the ones that tell stories.
That’s where AI can come in.
First, AI can help with identifying good investments.
Most already know about rarity.tools, which can identify the rarity of art NFTs, but there’s a lot more that can be done at scale. For example, consider processing all the NFTs on OpenSea, tracking how much they were sold for, how often they were traded, and a wide array of features.
But, how do you build a meaningful feature set from art? That’s where AI comes in. You can either take a deep learning approach if you’re not as interested in the underlying factors, or you can take a more structured approach by hypothesizing over the factors that matter and measuring them (e.g., searching for color combinations or objects).
With that predictive model, you can help make good investments.
Second, AI can help with creating new NFTs.
While much of the work that’s been done so far has been through super simple generative applications, often in R Studio, I believe that the best digital art needs to tell a story. But, that requires imposing structure and weaving different ideas together. For example, in a recent NFT collection that we in Living Opera launched called Glory Streams, we combined the operatic voice over Silent Night with digital animation around the Nativity and the stars.
What if we brought AI into the equation? The entire field of transfer learning is devoted towards taking information from one medium to another — that is, augmenting sound and images, for example. Given a particular theme and story that the creator wants to convey, he or she could search over possible combinations of different mediums to figure out the right combination.
Obviously there are many other factors involved in the process. In fact, Valerie does a fantastic job summarizing them, ranging from the role of communities to the impact of partnering with influencers.
But, a big part of the message here is that it’s important to be creative with how we use the tools around us, whether it’s AI or a design interface, to identify good opportunities. Technology is suppose to help us — not hinder us.
With the expansion of computing power and data, so much is available to build effective predictive models. That said, the tool should be just that — a tool. Don’t overlook gut instinct. The “market” isn’t always right and you should still invest where your heart is.
This article was written by Christos A. Makridis, the Chief Technology Officer and Head of Research at Living Opera, a multimedia art-technology startup. He also is a research affiliate at Stanford University’s Digital Economy Lab and Columbia Business School’s Chazen Institute, and holds dual doctorates in economics and management science & engineering from Stanford University.
Join thousands of data leaders on the AI newsletter. It’s free, we don’t spam, and we never share your email address. Keep up to date with the latest work in AI. From research to projects and ideas. If you are building an AI startup, an AI-related product, or a service, we invite you to consider becoming a sponsor.
Published via Towards AI