History has demonstrated innumerable times that great discoveries and inventions do not have an insular existence. By extension, this statement also covers the rapidly evolving field of technology, where each year seems to come with its own wave of pioneers ready to challenge the status quo with their innovative solutions and approaches.
One of these pioneers is Humans.ai, a company that has managed to fuse two of the most innovative and in-vogue technologies from the market, blockchain and artificial intelligence, to usher in a new era of trust. Although the intersection of blockchain technology with artificial intelligence seems like a natural step in the evolution of both technologies, Humans.ai is the first company to successfully deliver a clear use case in the form of synthetic media that leverages the unique properties and benefits of both technologies.
Now, with the technology developed in-house by Humans.ai, people can encapsulate their digital DNA in an AI NFT, enabling them to benefit from one of the most advanced instruments of our time, artificial intelligence, without needing to worry about the complex inner workings of these technologies.
At the centre of the Humans.ai ecosystem is the Proof-of-Human (PoH) mechanism, a complex blockchain-based governance, consensus and verification system that ensures that every AI NFT is backed by a human decision. Through PoH, humans leverage their biometric data in order to prove that a specific AI is still under close biological supervision. Humans.ai empowers people to participate in the governance and management of any AI, essentially making sure that the objective of the AI is aligned with the human’s objective.
AI NFTs an intelligent asset class
Before taking a closer look at the Proof-of-Human concept and its role in the Humans.ai ecosystem, it’s important that we first outline what an AI NFT is. A non-fungible token, simply known as NFT is a cryptographic asset that rests on top of a blockchain network.
Its distinguishing characteristics are its unique metadata and identification codes which make each NFT one of a kind. As such, unlike cryptocurrencies, NFTs cannot be traded or exchanged at equivalency. Even so, due to their inherent properties, NFTs can be an ideal medium to represent real-world items like artwork or real estate deeds and even people’s identities. But if we place the acronym AI alongside NFT, we open up a whole new window of opportunities and technological possibilities. AI NFTs are a type of machine learning model that also acts as a new asset class capable of encapsulating neural networks or a digital genome that can be represented by a person’s voice, face or other biometric data.
How is an AI NFT created?
Imagine a researcher who develops an algorithm that takes voice inputs and builds a program that converts written text to audio that resembles the initial voice input. To ensure the security of AI networks, Humans.ai uses blockchain and a suite of other technologies like cryptography and containers. The AI network is encapsulated in a container, a structure similar to a virtual machine that acts as a method of packaging an application so it can be run with isolated dependencies, along with its libraries and algorithm, so the application runs quickly and reliably from one computing environment to another. When the container receives a compatible input, it will produce an AI model.
If you want to digitalize your voice, you would use a similar AI network and feed it your unique digital genome, respectively your voice to produce an AI NFT.
The process through which an AI NFT is created in the Humans.ai ecosystem is called minting. Once an AI NFT is minted, it will contain two components:
A descriptive component — the biometric data alongside a set of governance rules outlined by the creator/owner of the AI NFT, as well other data that confirms the existence of the NFT inside the blockchain
A computational component — a type of container structure that encapsulates an AI network. By employing a novel form of zero-knowledge cryptography known as Zero-Knowledge Succinct Non-Interactive Argument of Knowledge (zk-SNARKs) we are able to ensure that no entity can execute the AI component from the container without the digital signature of the owner of the AI NFT.
Proof-of-Human, the centrepiece of the ecosystem
Proof-of-Human is an integral component of the Humans.ai blockchain that acts as a governance, consensus and verification system, which helps ensure that behind every AI NFT there is a human decision. Simply put, PoH is a complex and versatile mechanism that ensures three key functions related to AI NFTs: governance, consensus and verification. People can interact directly with the Humans.ai blockchain and the PoH mechanism via smartwatch or smartphone through specialized applications.
To ensure the governance of an AI NFT, Humans.ai utilizes Proof-of-Human to confirm that behind every request made towards an AI model there have been multiple digital signatures of real people. The digital signatures in question are generated through a mix of different data such as private keys and biometric data (fingerprints, heartbeats, retina scans, face recognition etc.)
Proof-of-Human ensures that requests made towards the AI NFT fall in line with the governance rules outlined inside the NFT by its creator/owner. For example, let’s say that a person called Alice has encapsulated her voice inside an AI NFT. To protect how her voice is used by third parties (a third party can be a media company or any person or entity that wishes to utilize the AI NFT to generate content), Alice can inscribe a set of rules inside her AI NFT that outline how her biometric data can be utilized. For example, Alice can stipulate that her AI NFT cannot be utilized to produce obscene content, hate speech, or anything politically incorrect. The rules written inside the AI NFT can always be modified by its owner in case they wish to rectify them or if they want to include additional rules.
Anyone can become a validator
How does the system manage to ensure that the rules stipulated by the owner of the AI NFT are respected? This is where the consensus, validation and governance components of Proof-of-Human come to light. Validators are the ones who analyze requests sent to an AI NFT to determine if they fall in line with the rules outlined by the owner of the AI NFT. Anyone can become a validator in the Humans.ai ecosystem as long as they stake a certain amount of $HEART tokens.
The rules embedded inside the AI NFT can also stipulate that only validators who have a certain threshold of staked tokens can act as validators for the respective AI NFT or how many validators are necessary to approve or reject a request. For their work, validators are rewarded with $HEART tokens, but if they consistently approve requests that deviate from the rules set within the AI NFT, they will be sanctioned and lose their validator privileges.
What are the advantages and disadvantages of Proof-of-Human?
The advantages of Proof-of-Human are straightforward. Through the sum of its parts and overarching functionality within the Humans.ai ecosystem, Proof-of-Human acts as an ethically compliant mechanism, that ensures that all the actors who send a request to the AI NFT, the validators, the owner of the AI NFT as well as the AI NFT itself play fair under the supervision of real humans.
As a solution that leverages blockchain technology to ensure and support part of its core functionalities and overall security, it’s only a matter of time until people start to raise a series of questions regarding its energy consumption. And rightly so, because blockchain has managed to assert itself as an energy and computational demanding technology. But this is true only for blockchains that utilize Proof of Work (PoW) to achieve consensus.
Humans.ai doesn’t use PoW in its blockchain, relying instead on a mechanism that resembles Proof of Stake (PoS) which is far more energy-efficient and a greener alternative. That takes care of the blockchain part, but what about the execution of the AI component?
It’s well known that building and executing AI networks and products requires, besides computational power, a lot of specialized resources. Although at a first glance this may seem like a disadvantage, Humans.ai circumvents this potential barrier by providing AI creators with an entire ecosystem and infrastructure that removes the need to worry about additional overhead. As such, they don’t need to worry about questions like where is the AI model going to be executed, how will it be executed, is it up, is it down, what servers and additional hardware do they need and so on. The only thing AI creators need to make sure is that their AI model plays fair, it’s available to the other members of the network and that the rules embedded within the AI satisfy their needs. The only resource that AI creators and people who want to send requests to an AI NFT are $HEART tokens, the governing currency of the Humans.ai ecosystem.
As we all know, no system is perfect, and some tradeoffs are always present, but in the case of Proof-of-Human, its disadvantages do not stem from a technological shortcoming but rather from the human factor and how validators can interpret some of the rules written inside an AI NFT. Depending on the complexity of the rules imposed by the owner of an AI NFT (for example, he/she can impose that all validators need to unanimously agree on a request for it to be accepted) there may be a delay between a person’s request and the validation of said request.
Proof-of-Human in action
The main use case of the AI technology developed by Humans.ai focuses on the synthetic media sphere, namely text to speech/video translations through the use of an AI NFT. But what does the process look like?
Let’s say that Morgan Freeman decides one day to mint an AI NFT that encapsulates his unmistakable voice. If a company decides to use Morgan Freeman’s voice to advertise one of their products, they would need to formulate a request towards the AI NFT that holds the biometric data, respectively the voice of the actor. This is when Proof-of-Human comes into action.
Validators start to examine the request to see if it falls in line with the rules stipulated by its owners and begin to vote whether to accept or decline the request. Depending on the number of characters of the request (the actual lines that will be transposed into Morgan Freeman’s voice) the $HEART costs can increase or decrease. Pricing is delineated by the creator of the AI NFT together with the validation rules inside the NFT. After the request is processed, the $HEART tokens paid by the entity who formulated the request is split between the creator of the AI NFT, the people who hold the computing nodes and the validators.
Conclusion
In traditional blockchain systems, consensus mechanisms are designed to solve the Byzantine Generals’ Problem, an allegory that describes an inherent flaw of distributed systems where actors do not know or trust each other. Simply put, it is a mechanism through which unknown members reach a common decision.
To an extent, Proof-of-Human can be seen as a consensus mechanism for AI NFTs, but that would be an incomplete perspective on the capability of this mechanism because Proof-of-Human is also designed to act as a validation and governance mechanism. The decision to mint or not mint a product represents the consensus part of Proof-of-Human, but how the decision is taken by validators extends into the validation part of the mechanism itself.
Lastly, Proof-of-Human acts as a governance mechanism because every validator has a direct stake in the system, enabling them to participate directly in the management of the ecosystem. To conclude, Proof-of-Human is a governance, validation and consensus mechanism designed to support AI NFTs.
