As macro trends, the hype cycles behind blockchain and artificial intelligence can be reduced to a single dichotomy: decentralization versus centralization. Blockchain systems are distributed by design - hence the calls for "decentralization" - whereas AI, especially given the controversies surrounding data centers and corporate overreach, is assumed to be centralized. While there are exceptions (centralized "blockchain" systems, locally run AI models), both technologies usually find their stride when they do what they do best - exist as true distributed systems (blockchain) or leverage the world's processing power and knowledge to deliver fast results (AI).
At teia.cafe, we recently branded ourselves with a new slogan "No Databases, No Logins" since it's the most obvious thing that sets us apart from everyone else. Wanting to double-down on the idea that a blockchain ledger is essentially just a "shared database", we have focused all of our efforts over the last few years building applications and features without any reliance on traditional databases or logins - you sync your wallet when you need us, then un-sync when you're done. No data passes through anything that wasn't already there - essentially guaranteeing that your privacy and ownership are kept intact.
Those are the sort of exact issues that people are concerned with in AI projects right now, since in most cases you really have no idea what the models are doing with your information once you send it off into their data centers for research and transformation. So we started asking ourselves what could happen if we combined the verification layer of the blockchain with the processing power of AI? The possibility of the emergence of the "hyper-consensus layer" as a new paradigm in the near future.
The Hyper-Consensus Layer
Before diving in, we have to first clarify what the hyper-consensus layer is not - ever since the hype cycle in 21' there have been numerous "blockchain" projects that have emerged where the rush to get into the markets quickly caused a lot of projects to cut corners on the whole "decentralization" thing - when OpenSea first launched, they didn't even bother with things like IPFS or on-chain minting and stored things on traditional servers, including the media. Only later did they migrate to IPFS and on-chain minting after people (including several high profile celebrities) complained about it - but at that point, they already lost people's trust and the markets were tanking, never recovering from its all-time highs, even now.
Despite these poor market conditions, the reason why we have chosen TEIA as our community to work with over the years is the fact that they have stuck to their guns and are using blockchain/NFT/crypto tech as intended, and we believe they will get rewarded for it eventually. The broader Tezos ecosystem - which gained its reputation for the sophistication of its governance models and is verifiably one of the most decentralized communities out there due to its low barriers of entry - also helps to reinforce the narrative that they are "serious" about the whole decentralization idea, even in practice. (They were one of the few places to defend royalties payments for NFT artists during the hype cycle - something that is essential for professional applications in the entertainment/arts industries.)
The strength of AI models, on the other hand, comes from the consolidation of knowledge in it of itself - while locally run AI models do have their place in certain contexts, they're unable to compete with the raw power and speed of large corporate models that are often way ahead of the curve in terms of processing power and performance.
If we want to combine the two technologies together, we want to focus on the respective strengths of each and arrive at powerful results we wouldn't be able to get otherwise - a blockchain system that is "open" as possible, combining it with the latest AI models that show the most promise for solving real problems at scale. Here are some reasons why combining the two streams makes a lot of sense:
- As a data layer that is inherently validated, blockchain systems are ideal for storing information, especially sensitive and detailed information that needs to be reviewed by multiple parties. In a way, AI models already do this by amalgamating existing datasets into its responses, but the models will run more efficiently if it pulls its information from data systems that are already "clean".
- The two industries already share a common resource, "tokens". Cryptocurrency systems inherently generate tokens through its economic models, while AI systems consume tokens as a measurement of use. With AI models struggling to find a path towards profitability right now, being able to regulate the token usage of these projects could create a "self-sustaining" loop.
- In the last 5 years, many new DAOs (Decentralized Autonomous Organizations) have sprung up (including the TEIA DAO, which we are members of) as an attempt to create governing or regulatory bodies around certain professions or activities - but these organizations often have trouble finding things to regulate. AI models give these groups and organizations something to collectively design and moderate, in a transparent way that is also digitally native.
In the future, you might imagine a scenario where you go to a crypto exchange to convert your cash into AI tokens—which you can either burn for your own use, or hold as an investor and stake for yield. The token's price and issuance rate (i.e., inflation) are adjusted by demand through DAO votes by stakeholders, which in turn respond to the burn rate of the tokens themselves. (Finally, crypto tokens can actually be used for something.) A lot of things need to line up for this to work, but when they do, a picture of a "mature" model for both industries starts to emerge: crypto starts to look more like real money, while AI systems gain the accountability they need to be taken seriously.
Art NFTs? Why Now?
Well, if you made it this far, you probably realized that we have pretty big ambitions about what these technologies can do, beyond art, NFTs, or even crypto itself. So why be interested in NFTs at this point in time? Why be interested in art?
While the combination of blockchain and AI tech seems like more or less an inevitability at this point (it makes too much sense - someone will do it eventually), it turns traditional conceptualization of these issues on its head and the process of getting there is more likely than not going to be very messy. There are going to have to be lots of experimentations - probably for years - before something coherent and workable comes to the fray.
The crypto industry went through several phases since I joined in 2013 - the blockchain itself, proof-of-work/proof-of-stake, smart contracts, NFTs, memecoins, and now the trend seems to be in tokenization of real-world assets, none of which has quite hit its "home run" just yet. That's where art comes into play - for better worse, NFTs are still arguably the only "real" product in the Web3 space right now - as a representation of something that you can actually buy/sell/own. And this is an ideal place to figure out the nuts and bolts of how these models should operate in a relatively low-risk environment - compared to, say, tokenized uranium (yes that's a real Tezos product) or real-estate where people are more likely to fall back to known habits because the scrutiny around these assets is so high.
As far as innovations go, historically speaking, always start as "play". That "play" usually manifests itself first in art, music, entertainment, and games. The concepts are born there, gets in people's minds, then eventually become common practice later when the idea matures over time.
So the "hyper-consensus layer", what is it exactly? Well, it's basically the idea of people using decentralized systems like DAO and multi-signature smart contracts to regulate the decision-making power of models like AI and algorithms as a whole...AND using their own token funds to pay for everything, including itself. Sounds like science fiction right now, but I don't think people realize how close we really are to making something like this happen - the pieces are already there, after all.
In a nutshell, that is what TEIA is currently wired up for - it has art, it has community - it has governance, it has its own economy and procedures that make it the only one of its kind. If we continue to build on top of it, something will happen - I know this, for sure.