AI has seen unprecedented growth in the past two years, the majority of that being in the past 12 months since OpenAI launched GPT 3.5. The technology has progressed to a stage that was - up until recent years - presumed to be years out.

AI is being utilized by almost every sector. As the technology develops and the understanding of its potential grows, the implementations will be seen in every aspect of every business in one form or another.

Due to the computational power required to train AI models and the costs involved in acquiring and running them, the industry's barriers to entry are high. If AI moves forward as is, the largest companies in the world (or extremely well-funded ones) will control all aspects of AI implementations. Microsoft, OpenAI, SentinelOne, and Darktrace are just a few examples of the entities that will control the future of AI.

Tensereum has a solution to this: decentralization. Alone often just buzzwords, ‘AI’ and ‘decentralization’ are technologies and ideologies that, although important alone, are imperative (and unstoppable) when combined.

Decentralization is a concept that has been introduced centuries before, with the term first being used by the French in the 1820s. 200 years later, the importance of decentralization is arguably much higher—with the technological revolution we are currently amidst, data decentralization has never been as vital as it is right now.

There have been many attempts and iterations of decentralization, with torrents being one of the first to achieve mass adoption. Although they achieved decentralization to the masses, the extent of decentralization wasn’t particularly great—and when it came to verifying the ability to trust, it was next to none.

Blockchain is the technology that has revolutionized decentralization. The adoption of Bitcoin, the first technology to implement a fully decentralized blockchain, has proven to the world that the technology not only exists - it works.

Blockchain, although considered complicated by many, is a relatively simple technology once the concept is understood.Blockchain participants can be divided into three: users, nodes, and miners.

Users - Users create a transaction by compiling data correctly and broadcasting it to the blockchain network. Nodes - Each blockchain has its own set of rules. Every participating node can be considered a ‘referee’, ensuring that any given data to the network follows these rules. If the nodes agree the data from the user is abiding by the rules, it is passed on to the miners.

Miners - Once Nodes have verified data, it is passed on to miners, who will produce a block - usually in turn for reward. These blocks contain all the verified data that nodes have passed on to them; they are available to anyone to see and verify themselves in real-time, and historically.

Although relatively simplistic compared to many technologies, the ability to secure the decentralization of a protocol has never been as complete as it is with blockchain technology. Bitcoin may have been the protocol that introduced the world to blockchain, but there are other equally viable use cases.

The current status quo of AI is that of a fully centralized and closed source nature, this leaves the technology open to manipulation, censorship, and a multitude of regulation that could be enforced either by nation states or the entities leading the way in AI technology progression.

The potential of AI has been clearly demonstrated with OpenAI’s GPT models, Midjourney’s generative graphics being practically undetectable, and now OpenAI are launching Sora - generative AI video content. The “future” of AI isn’t coming soon, it’s here.

The technology is impressive, but as with most technology - it will not all being built altruistically.

The founder of OpenAI has launched a product called ‘WorldCoin’ from a company he founded Tools For Humanity. The product requires Iris scans from participants and has led to international controversy over the privacy of the data collected. Spain has enforced a ban on the product amid privacy risks.

Altman has also caused controversy by the suffering OpenAI has caused to many Kenyan nationals whose job it was to go through hundreds of images and texts depicting abhorrent acts, including graphic sexual violence, as he trained the GPT models. Kenya has now banned all Altman’s companies from operating in the country with accusations of exploitation.

The general population are seemingly blind to these facts, but is this the type of person that the world wants in control of potentially the most powerful technology known to man?

The past decade has demonstrated that content online can easily manipulate the general population. The West often speaks of Russian propaganda (a well-known fact) but is frequently oblivious that propaganda is an international issue and a huge industry. From US elections to opinions on conflicts in the Middle East to the disgraced firm Cambridge Analytica and its influence on Brexit, media manipulation has been developing and influencing opinions more than ever.

AI will only enhance this ability as it continues. With closed-source models, it is completely impossible to verify the facts, images, and videos produced. Data sets can be controlled, falsified, and censored, creating a distorted vision of the truth to fit a narrative.

News outlets have only become common in fact-checking since Trump became president of the US in 2016. With AI developments, it will only become more difficult to verify whether something is genuinely real or a creation to fit the narrative of what is wanted to be perceived as ‘real.’

Although having open source AI distributed by blockchains will not remove the centralized entities, it will allow people to ‘fact-check’ themselves with verifiable data. The importance of this cannot be understated.

AI technology has come under a huge amount of scrutiny in the past twelve months. World leaders and founders of the world’s biggest companies have warned against the negative potential of AI.

[Regarding AI] The risk of something seriously dangerous happening is in the five-year time frame. 10 years at most.

- Elon Musk

Congress should engage with AI to support innovation and safeguards. This is an emerging technology, there are important equities to balance here, and the government is ultimately responsible for that.

- Mark Zuckerberg

[Regarding AI] But there is a case to believe that it may pose a risk on a scale like pandemics and nuclear war, and that’s why, as leaders, we have a responsibility to act to take the steps to protect people, and that’s exactly what we’re doing.

- Rishi Sunak

The fact of the matter is that all these statements are inherently correct. AI can potentially pose a huge risk to humanity. Regulation may be a solution; however, there will always be bad actors who will ignore or evade regulation. The other issue with regulation is that it is often created with the interests of a few in mind, not the interests of the general population. Regulations can threaten innovation and distribution, amongst other aspects of AI development.

Using blockchain technology, AI can operate in a pseudo-unregulated or self-regulated system. Instead of being regulated by nation-state governments, a distributed blockchain AI network can be regulated by the users themselves.

Since it is evident that AI does pose certain risks, regulation is not inherently a bad thing - it is a vital aspect as the technology progresses. The issue with standard regulations is that they are implemented by centralized entities whose interests are often unaligned with those of the general population. Tensereum’s Decentralized Autonomous Organization (DAO) solves this problem.

User 1: Submits AI model trained to push a right-wing political narrative.

User 2: Submits proposal to remove User 1’s model from the blockchain.

All Nodes: Vote A: Remove User 1’s model or B: Keep User 1’s model.

Blockchain: Assuming consensus is reached, either option A or B will be enacted.

Tensereum will move progressively towards complete decentralization with the DAO in place. It is essential that when the chain is in its infancy, the level of decentralization will need to be more substantial to rely on a DAO alone. As decentralization increases, the protocol can move to a completely DAO-governed network with no central entity having control.The DAO will also be responsible for protocol-level changes and development. In the same way that proposals for regulatory decisions are submitted, proposals for network changes can also be made. Having a dual-purpose DAO that governs the regulatory risk and the protocol-level implementations puts Tensereum ahead in solving the issues involved in decentralized AI.

There are very few compelling arguments promoting closed-source AI models. The usual reasoning for closed source is that it keeps software proprietary, which is the case regarding the current AI models hidden beneath a veil of secrecy. Close source models are completely unverifiable since one can only trust the owner that it works how they tell you it does; this centralizes trust, which is what we are trying to avoid. Closed source models also stifle innovation and development since other participants cannot contribute code changes or fork the current state and create their version based on the current code base.

Open-source AI models are a completely different story. They not only allow innovation, they promote it. Having an open-source AI model invites developers to challenge themselves by contributing code to the project or launching their own project forked from an existing one. Equally, they are fully verifiable. There is no need to trust a single company that this model does XYZ; instead, it can be seen and viewed that the model indeed does XYZ. Open-source technology, particularly AI, is a crucial step towards complete verifiability - especially combined with the blockchain aspect of verifying what model is truly being used.

AI is moving forward at an unprecedented (and unexpected) pace. However, although moving forward, it is only sometimes moving in the right direction. Centralization, censorship, and regulation are just a few of the issues already very apparent within the industry and threaten not only innovation and development but also humanity.

Tensereum has a solution that solves not just some but all of the industry's current issues—a fully decentralized AI network distributed via blockchain. Blockchain allows for immutability, meaning no one can stop or close the network down; it also allows for variable data since the nature of a distributed public ledger is transparent. The distributed network via blockchain also allows for self-regulation via a decentralized autonomous organization (DAO).

Built by the community for the community, Tensereum will deliver the future of secure access to AI models without compromise - for everyone and forever.