Synthetic media, also known as AI-generated media, generative media, personalized media, or deep fakes has come a long way since it first hit the tech scene in the early 2000s, slowly consolidating itself as one of the most popular and important applications of AI technology. To some extent, this seems to fit the bill, but if we take a closer look at synthetic media, or however you choose to label this technology, you may spot something significant. Synthetic media manages to resonate with an integral part of ourselves that truly makes us human — our latent desire to express ourselves through the act of creation.
Humans.ai believes that synthetic media is a direct expression of our interest and desire to create something original, representing at the same time a new medium of expression that will challenge the current top-down media creation paradigm by reducing the gap between idea and execution.
Synthetic media is a catch-all term that describes the creation of artificially generated content in multiple modalities like text, image, audio, and video. Humans.ai has made it its mission to develop a multifaceted and inclusive ecosystem that streamlines the creation of AI. With our technology, anyone can turn their ideas into reality through the power of artificial intelligence. Furthermore, by pairing AI and blockchain together, we manage to address some of the most pressing issues concerning the ethical creation and use of artificial intelligence and synthetic media while also providing in-depth traceability and ownership.
Synthetic media under the hood
The increase in computing power in the last twenty years has created fertile ground for the development of synthetic media, to the point in which it has become a common occurrence in our daily lives. As the name implies, synthetic media refers to media that has been produced artificially in a digital environment, instead of the traditional way on canvas or in stone by a master of his craft alongside a couple of apprentices. With synthetic media, anybody can don a painter’s apron and let their imagination run wild through the power of AI.
When discussing synthetic media, one important aspect is the modal part. Think of it as the medium in which the media propagates. It is directly related to our basic senses, vision, hearing, tactile, and so on. In its early days, synthetic media usually had a mono-modal approach like an artificially created image or a piece of audio made by a computer. But as progress was made in this field, synthetic media shifted toward a cross-media approach in which people can move from one modality to another.
For example, to make a synthetic video, you would need to combine multiple modalities like audio and video. But it’s easier said than done because the technology used to achieve these feats requires expert-level knowledge to develop and utilize. This is where Humans.ai steps in. We want to take all of the complexities inherent to synthetic media and artificial intelligence and hide them behind an intuitive UI to streamline the creation process and make synthetic media more inclusive.
With our technology, aspiring content creators don’t need to learn the mapping between one modality to another, as our AI NFTs, a combination of AI technology with blockchain will act as building blocks that will facilitate the transition between different modalities, enabling anyone regardless of their technical background to get started with synthetic media. This isn’t to say that you are limited to transition only from different modalities. We could also have an image-to-image translation, a process through which you take an image from one domain and transform it in a way to retain the characteristics of a different image from another domain. We can achieve this by training our in-house deep learning models to associate text with images and create a latent embedded space in which we could do this translation from different modalities.
Synthetic media with a human touch
Human.ai’s approach to synthetic media follows the K.I.S.S. principle, no, not K.I.S.S. the band, K.I.S.S. as in “keep it simple stupid”, meaning that we strive to simplify the design of our ecosystem and tools to make your interaction with our technology as simple as possible. Once you get your hands on one of our unique AI NFTs, you can start tinkering with your synthetic media project. Depending on what you want to do, you will need different types of AI NFTs. If you want to generate a synthetic piece of audio, an AI NFT that encapsulates a voice inside it will get the job done, but if you want to do something a little bit more complex, like a digital avatar, you will also need an AI NFT that encapsulates an image or a video. It all depends on your idea and how you want to give it a heartbeat.
Regardless of what you want to create on the Humans.ai platform, everything starts from a text input. For example, if you want to generate a slogan for your business, all you need to do is write it in our UI, and the gears will start spinning to give the audio output. If you want to take things one step further, you can attach that audio to a digital avatar to give a body to your voice. An advantage to this text-based approach is that text comes inherently in a particular language, meaning that you can have a digital avatar talk in that particular language.
We manage to achieve this through an intermediary step during which we convert your text input into phonemes, units of sound that distinguish one word from another in a particular language. Each language has its unique particularities and some are phonetic like Italian, Romanian and Arabic, while others are less consistently phonetic like English and French. This technique also enables us to do the mapping between phonemes and how the lips of an avatar should move to render a hyper-realistic model.
When it comes to technology, it’s difficult to state that you are truly the first to develop something, as parallel development is a common occurrence in the field. Of course, this does not mean that we cannot put our signature on this technology and try to innovate and improve things. This is best reflected by our inclusive approach to synthetic media and AI technology and how we strive to make it seamless and user friendly.
One major restriction present in existing synthetic media solutions is that users are forced to stay in a standard background or wear a certain type of clothes or maintain a rigid body posture that seems unnatural. We are working to circumvent these types of technological limitations to be able to support different environments and changes in head poses. We call this robustness to the environment the ability to allow our users to be in their natural environments without imposing any constraints. Current synthetic media tools require the users to stay in a fixed position accompanied by a simple background. At Humans.ai we are working to enable people to manifest themselves in their familiar environment without restrictions.
Humans.ai removes the barriers to synthetic media development
The traditional art studio is reserved for people who have dedicated their lives to studying and perfecting their techniques and mastering their craft, but the advent of digitalization and synthetic media is paving the way for an alternative to this status quo, one that is more inclusive and open to anyone.
Humans.ai is speeding up this process with its ecosystem composed of an extensive AI Library and Humans Studio, a place where all the pieces of the AI puzzle fall together, waiting to be used by a creative spirit who wants to make their idea a reality. The Humans ecosystem is designed to host an expanding suite of different technologies.
In the first stages of development, the focus is being placed on AI algorithms developed in-house by the Humans.ai team, but as the project progresses, we will enable AI researchers and members of the academic environment to host their technologies and make them available for wide-scale use. This way, AI producers have the possibility to be part of larger projects that otherwise they will not be part of and users will directly benefit from a larger selection of AI algorithms they will have at their disposal.
To make the most out of its expanding array of AI components, Humans.ai has decided to give its ecosystem a plug and play approach, meaning that users can choose what algorithms best suits their needs and combine them seamlessly to create their unique piece of synthetic content. Like with any puzzle, not every piece is compatible with each other, but algorithms will be arranged logically and labelled accordingly to enable anyone, regardless if they are a tech person or not, to play with the components and produce a running piece of media. By bridging the gap between AI producers and people, Humans.ai aims to democratize the creation of synthetic media, while also ensuring that ownership and remuneration are guaranteed fairly.
Let’s not forget that with great power comes great responsibility. So, to ensure that its technology and the technology it hosts aren’t used unethically to produce harmful and offensive content or as a means to push certain agendas through fake news and disinformation, Humans.ai has created the Proof of Human (PoH) mechanism. Acting as the cornerstone of the Humans ecosystem, PoH is a validation, governance and consensus mechanism which ensures that behind every piece of AI created inside the Humans ecosystem is a real human who thoroughly verified and checked to see if that AI respects the Humans.ai standard and the ethical aspects of AI. After all, we aim to use AI to turn humans into superhumans, not supervillains.
Using blockchain technology as a core element of the Humans.ai ecosystem, we become more than an AI and media enabling platform. We take a big step towards achieving trustworthy AI, a vision of the artificial intelligence of the future in which trust is guaranteed during each stage of the development lifecycle of AI, starting from design, deployment and use.
As an environment that encourages AI producers to collaborate and share their work with the rest of the world, Humans.ai leverages blockchain to guarantee ownership and provide accountability through this technology’s intrinsic capability to provide in-depth traceability, transparency, decentralization and immutability.
Paired together with smart contracts, self-executing pieces of code that run on top of the blockchain, Humans.ai ensures everyone will have full ownership of their technology and that they will receive all the benefits and revenues when their technology is used.
Humans.ai brings the innovation from the academic environment to real humans
Currently, synthetic media development is highly opaque and fragmented, giving regular people very little visibility and more importantly, access to these new technologies. There is a big cleavage between synthetic media producers/researchers and regular users.
On one hand, big corporations have huge amounts of capital to invest in the development of synthetic media technologies behind closed doors and mainly for their own products. At that level, profits are the name of the game, which up to a point, isn’t necessarily a bad thing, but once you start to quantify innovation as a financial numbers game, you usually start to stagnate. Research most of the time is taking one step forward and two steps back, and when you suddenly pull the plug on a project because the board members concluded that it’s no longer viable moneywise you waste all of the progress you managed to make.
On the other hand, in the academic environment researchers and students use their brainpower to push technology to its limits and to come up with new and innovative applications for the technology. But there is one glaring problem, many academic groups do not know the value of their technology, or they do not develop technology with a specific end-user in mind. This means that most of the time, their work does not advance past the Proof-of-Concept phase, eventually becoming a well-stacked pile of research papers that collect dust.
Commenting on this issue, Nicu Sebe, Computer Science professor at the University of Trento and Head of AI at Humans underlined the fact that “From my perspective as a professor, researchers are often stopping at the prototype level. In academia, we are not making research having an end-user in mind. With a platform like Humans.ai, AI researchers have the possibility to give visibility to their work and increase people’s interest in this domain. Furthermore, people will be free to use these technologies in much larger projects beyond what an individual could do by himself in an academic room.”
Humans.ai is proposing a different approach to artificial intelligence and synthetic media, one focused more on democratizing these technologies. Our mindset is to make AI development an inclusive process, one in which anyone can contribute, the focus being placed on end-users, namely regular people. Also, besides empowering common users by giving them access to AI, Humans manages to bridge the gap between real-world needs and the pen and paper AI developed in the academic sphere.
Our approach to AI is simple — we see it as a black box in the sense that users don’t need to understand or preoccupy themselves with the complex processes that take place in the background. Users will know the input they need to introduce and what will be the expected output. This way, any human can play in our ecosystem with different AI components and put them together to create their own media.