Disclaimer : Before the reader goes any further in this article, it should be noted that it has little technical value. Rather, it is an expectation in which we mix the notions of blockchain that could be linked with one or more artificial intelligences. Without pride, this article, far from being scientific, proposes keys of innovation for a synergy between these two disruptive technologies. Have a good reading!
Just as the Internet has changed trade, law and research over the past thirty years, blockchains are shaping the future of business, logistics and finance. Its close link with Artificial Intelligence (AI), which is now emerging, and the growing interest of governments in it justify the throng of projects that have revolved around the blockchain and the development of Initial Coins Offering since 2017. It has been ten years now that the GAFAs have installed AIs within their business model and they are already planning to place AIs at the heart of a new revolution that, if not industrial, will be digital.
To be sure, it is enough to look at the investments specific to each of the technologies. Between 2013 and 2018, the amount of global investment in the development of Artificial Intelligence amounts to 126.3 billion dollars (source: Statista). Also, a Deloitte study suggests that 27% of the organizations surveyed worldwide are willing to invest between $1 million and $5 million in the development of a blockchain proof-of-concept demonstration that would improve their value proposition.
Although distinct, these two processes, in our opinion, present interesting points of synergy. While in recent surveys, each sector perceives them as competitors.
But then, is it possible to integrate one or more AIs into a Blockchain ? Is there a mutual relationship? If not, can this mutuality be created? What mechanisms are involved and how does all this work? These are all questions that we will try to answer in this article. Before going any further, it is important to remember what a blockchain, an AI is and how these two technologies can find points of convergence.
A blockchain is a database accessible to people all over the world and operating without a trusted third party. It uses advanced cryptographic techniques to reward those who process and secure this data (in the case of Proof-of-Work). Since the Ethereum blockchain, this database has become programmable with the implementation of Smart-Contracts, decentralized applications (Dapps), and decentralized organizations (DAOs).
What is most likely to be implemented with an AI are smart-contracts or autonomous programs that automatically execute, thanks to the blockchain, actions predefined by the stakeholders. These contracts are extremely effective when it comes to betting, reimbursements or simple payments between wallets (cryptocurrencies portfolios).
Artificial intelligence (AI) is when an algorithm reproduces the recognition and adaptation capabilities of a human being. In other words, AI can perform simple tasks previously performed by humans. It is confronted with a database that it integrates in order to make the right decisions when carrying out these tasks. For this algorithm to be effective, it must therefore be trained and confronted with a swarm of similar data to find connection points, recognize them, and why not produce standards later. As a result, the masters of these techniques are obviously the digital giants (GAFAM, BATX, NATU) thanks to the mass of data collected from their users.
These two technologies, although radically different in appearance, can in the event of unity improve the paradigm of all. One produces secure data that is accessible through the transactions issued, the other uses data for recognition, optimization or creation. So, let's see how these two tools can be used together ?
The use of relevant data, a trust builder for AI
As mentioned above, for an AI to be effective, it must be assigned to a specific task and have a significant amount of data to operate. However, all data on the blockchain has been validated by the network, is no longer editable and is de facto effective, unless a 51% attack or a smart contract breaking occurs. Thus, all the data entered on the blockchain is relevant and secure, which by extension would secure the algorithm that uses it.
By extension, an entrepreneur could store his data securely on a blockchain and exchange them on "data marketplaces". It is quite possible to imagine a secondary data market in which companies would obtain customer’s data. Our contractor would be paid for data that would not cost him anything to produce.
The AI linked to this data could potentially use it as a predictive model, reuse smart-contracts metadata and eventually produce standards. This would be called automatic production of intelligent contracts: the AI could provide programming frameworks that each developer could use. The renewal of smart contracts would be done as soon as a soft fork, i. e. an update on a public blockchain, takes place. The soft fork will be the opportunity for the AI to optimize the protocol, particularly through its cost, execution time and the renewal of stakeholders. Thus, new intelligent contracts could be updated or even created when the old ones are considered obsolete by most of the network.
Artificial Intelligence, the oracle to rule them all
For a smart contract to be effective, a specific situation must be specified. However, some of these contracts require information external to the blockchain to be activated. Indeed, the blockchain is blinded by information that does not concern it. A third party must therefore add the missing information so that the smart contract can be activated. This person is called the "oracle", which induces a new trusted third party.
However, in a blockchain a part of the network can contest the information of the oracle. The AI, in relation to the relevant information, could include the information necessary for the performance of the contract, as soon as it is in its possession. As a result, the consensus around the information would be automatic since it comes from one or more automatic databases and certified by the developers.
Nevertheless, this system has some limitations, if the information source unfortunately provides a "fake news" through a hack. If this situation were to occur, then the information on the blockchain would be distorted and the execution of smart contract would not take place, would be delayed, or worse would have paid a third party not involved in the process. This creates a loss of trust in the protocol and an inevitable drop in the price of the cryptocurrencies issued on this blockchain.
In the distant future, the end of mining and the beginning of a financial system between machines
Currently, miners bring computing power to the network to be designated validators and can receive the reward of currently 12.5 bitcoins per validated block. This is the Proof-of-Work consensus algorithm. For some, a huge amount of money has been invested in graphics processors that act as a calculator to have a probability of being designated as a transaction validator and being paid in cryptoassets.
It has long been criticized by traditional media; this algorithm is extremely energy consuming. Today, according to the Bitcoin Consumption Energy index, the miner's network would consume only 73.123 TWH per year for the bitcoin blockchain, either Austria's electricity consumption or 34.73 Mt CO², or a carbon footprint comparable to that of Denmark. These figures quickly show that electricity consumption and carbon dioxide emissions from the Bitcoin network are unsustainable, especially in a world suffering from climate change.
In this situation, an AI designated as the only validator node and performing the transactions may be a potential solution. It is possible to create a whole ecosystem in which the intelligent system would replace miners, to validate transactions taking place on blockchain. The algorithm would be the only wealth generator according to the protocol, since it is the only validator. Environmental considerations would be less important since only a computer would have the capacity to validate all the data emitted on the main net.
We can quite imagine that it would remunerate all the nodes of the network that hosts part of its algorithm within its storage capacities. In technical terms, "archival node" or "full node" would be directly remunerated according to their storage capacity and time of storage. As a result, motivation would no longer come from mining but from data recording. It should be noted that the solution may seem outdated given the market already occupied by companies specializing in "cloud" solutions, which would not hesitate to connect to the network.
Having a real financial action, thanks to its cryptoassets, on the other nodes, the collaboration between machines would be increased. Through, the automatic creation of smart contracts, the financing of decentralized organizations that would strengthen the members of the network. A new paradigm would emerge that would largely favour individual initiatives rather than the gross calculation force and associated energy costs.
But then, if this synergy were to appear and everyone could host an AI and participate in the network, could we talk about decentralized artificial intelligence? In this case, the IA and Blockchain pair would first make it possible to optimize the data over the long term thanks to the information stored and given by the blockchain. Secondly, it would also allow users to connect in a secure and transparent way and more optimized while saving time with a deep learning of the algorithm. In the long term, some members, such as miners, could be excluded from the transaction security process.
Even if none of these scenarios are verified in the future, to hide behind the complementarity of blockchains and intelligent digital agents seems to be a potential loss of revenue, especially for companies in their permanent quest of value creation. Their combination will enable not only digital players, but also small specialised companies, to always have optimal use of a large amount of data, which is essential for the proper conduct of their economic activity.
The integration of an AI into a blockchain would further expand the possibilities for development and growth. Since technology is not a goal, the objective would be to ultimately create a near-perfect user experience in which the user would be paid for their direct contribution to the network. In all markets, the first companies to link these innovations will be able to ensure their sustainability.
Today the art market represents more than 14.58 billion euros in transactions. Most works of art have an extrinsic value, which is mainly based on the certificate of authenticity provided by third parties: the artist himself, but also gallery owners, art dealers, auctioneers, museum curators, and private experts. However, according to Bloomberg, nearly 10% of the value of transactions on the art market comes from forgeries.
As we’ve seen last week, a growing narrative within the ecosystem is pushing the idea that Proof-Of-Work is a bad system that will inevitably fail in time. The most common proposed alternative is Proof-Of-Stake or PoS [...]