Artificial Intelligence and Blockchain have become two powerful trends in recent years. While they are not usually related to one another, Blockchain could actually make Artificial Intelligence not only safer, but also much, much better.
Artificial intelligence (AI) has been received with both wonder and fear. The rise of the intelligent robots and our dystopian future is what mid-twentieth century science fiction novels are made of; and if recent reports about AI becoming more aggressive is anything to go by, the fears of George Orwell, Bill Gates and Elon Musk are not too far off the mark. But another new technological trend could be our saviour. Could Blockchain combat AI aggression?
Setting the scene
Can we really trust the AI’s analysis? We simply do not know if the system generated decisions are detrimental or not. Additionally, on the emotional level, there also remains a fear that AI will get too smart in the future and overturn its masters. AI is any task performed by a program or a machine that demonstrates at least some forms of behaviours associated with human intelligence. This may include such things as learning, reasoning, problem solving, perception, motion, and manipulation.
The exponential advancement of AI is happening largely due to the availability of big data. The organisation of these large swathes of data into structured components allows computers to process and synthesise to create advanced AI.
With blockchain we can wield AI better. There is a plethora of information out there discussing the intricacies of blockchain, but simply put, blockchain is an immutable distributed database, that is structurally based on trust and transparency. Its value, and perhaps its weakness too, has been publicly demonstrated in the financial sector, but blockchain application is as wide as it is varied.
There are two core elements to blockchain to keep in mind. It tracks and records all transactions within its network; and, each blockchain network is distributed across thousands of computers and nodes at the one time. Every change, amendment or addition is considered data added to the network. Before these can be made the existing data is compared across all the other computers on the network, ensuring its integrity.
Wherever there is a need for an absolute, publicly accessible database to confirm actions, blockchain can be applied.
Decentralised control. Decentralised decision-making
A key feature of blockchain is its decentralised network of nodes. Using mathematical logic, these nodes work together to solve complex problems and ensure secure ‘transactions’. Essentially, these transactions are just decisions confirmed across the nodes. This parallels the way AI determines the best course of action. When an AI system makes a decision, it will assess all possible alternatives through its alternating decision branches, ultimately resulting in the most mathematically logical decision. Now combine the two, where each decision AI branch is replaced with nodes across a decentralised network which must be confirmed by all parties, and you can see how decisions will systematically undergo logic and sense checks to improve decision making.
Trust is a key feature of blockchain and arises from its public and immutable keeping of records. Each transaction or decision made to a blockchain needs to be verified before being applied to the blockchain. When applied to AI, users can safeguard over who makes changes to that record, effectively democratising the decision process. User also stand to have a clearer understanding of the AI systems’ decision matrixes through clear and transparent decision records. This transparent system, will also aid bots trust each other, increasing inter-machine exchanges, and allowing them to share data and coordinate decisions.
Auditing AI decisions
AI decisions, such as a bot determining fraudulent financial transactions for example, are systematically recorded one data point at a time, and usually stored on a single, static location. However, blockchain is fundamentally a transparent ledger. It enables records that are kept in it to be fully transparent and traceable, making auditing more accurate. This will ensure confidence in the record, clearly show any tampering from the time the information was recorded and when the audit process takes place, and ensure the right decisions are being made by the AI system.
Big big data
AI systems improve as more quality data is fed into them. Logically, traditional static central storage systems can be limited by space, money and individual control. Being a decentralised network, blockchain stands to offer much larger data sets to AI systems that are not centrally controlled. Furthermore, this data is significantly safer not being stored in a single location; aside from natural physical points of failure, the distributed network means it is harder for malicious non-network members to hack and alter or destroy the AI system.
As masses of data fuels the advancement of AI, so too our fears of an aggressive AI system. We are still in the early stages of both technologies, with more room for evolution. Blockchain may offer a way for us humans to gain clarity on how AI operates. By combining both technologies, we can generate greater benefit to our society in the near future.
DataVLT is an affordable, on-demand analytics platform secured by blockchain technology. It is designed to simplify the complexity of data science. Backed by artificial intelligence and machine learning capabilities, DataVLT empowers enterprises to make meaningful sense of their big data and scale cost efficiently. Essentially, it is an end-to-end data/information management platform.
Learn more at www.datavlt.com