Blockchain is a decentralised ledger that records transactions securely and transparently. Best known for its use in cryptocurrencies its potential applications go far beyond that. One area where blockchain could be usefully applied is making AI more trustworthy, by improving its reliability and transparency. To understand how this might work we also need to understand how both AI and blockchains work.
In simple terms, Artificial Intelligence (AI) is a type of technology that enables computers to process data and learn from the results, to perform tasks typically requiring human intelligence. Tasks like problem-solving, predicting the future and learning how to improve itself.
The AI training process requires data, lots of data - big data.
Data is used to train AI by feeding it large amounts of information that it can use to learn patterns and make predictions. The process is known as machine learning, and it involves using models - a combination of algorithms (a set of calculations designed for a specific purpose) and computations that provide predictable outputs to reflect the real world.
These models recognise patterns in data and make decisions based on that information. The more data fed into the algorithm, the more accurate its predictions become. This is why large tech companies like Google and Facebook invest heavily in collecting and analysing data, as it allows them to improve the accuracy of their AI systems.
This is where blockchain technology can play an essential role - making machine learning and the models that power it more reliable and transparent and helping avoid bias, privacy and accuracy issues in the data.
By their very nature blockchain systems support the kind of clarity and certainty we need if we’re to continue trusting the AI systems we increasingly work with and encounter.
Blockchain is a digital ledger technology used to record data and information in a secure and transparent way. It is a decentralised system that allows multiple parties to access the same information, without the need for a central authority.
To further bolster trust each block in the chain contains a record of transactions, and once a block is added to the chain, it cannot be altered or deleted. This makes blockchain a highly secure and tamper-proof system for recording and verifying data. Some advanced blockchain platforms (like our very own FALKOR SI) also support the verification of the people and systems that use that data.
One of the biggest challenges with AI is ensuring that the data it uses is accurate and unbiased. This is where blockchain can play a pivotal role. By using blockchain to store and verify data, AI developers can ensure that the data used by AI is accurate and has not been tampered with. This can eliminate bias and ensure that AI makes decisions based on accurate and reliable data.
AI systems are used in various highly sensitive settings from cancer diagnosis to probation rulings. Decisions informed by AI can therefore be life-changing. By their own admission, the creators of AI don’t always fully understand how their systems work, or are unwilling to share how the models AI use come to their conclusions.
This lack of transparency is a major obstacle to gaining trust in AI services. The model is the heart of every machine learning system as it provides a set of rules for how it treats data and how it adapts itself
By using blockchain, the decision-making process of an AI model can be recorded and made transparent. This means that people could see precisely how the AI arrived at its decision, making it easier to trust and rely on.
By their very nature machine learning models are subject to change. Systems like Falkor SI can enable developers to ensure each change is tracked, improving our understanding of how AI systems really work by being able to show how models evolve over time.
This transparency could help increase both their use and development. For example, in healthcare, allowing models that work to be shared between healthcare providers, without sharing confidential patient data, could help spread useful new diagnostic and analysis techniques.
Data’s important role in modern healthcare reveals another critical role for blockchain in AI - helping to ensure the privacy and security of data used in AI services. AI systems often rely on sensitive or protected data, such as personal information or intellectual property. Using blockchain to create a secure and decentralised system will ensure these types of sensitive data are protected from unauthorised access or manipulation.
Blockchain’s a promising potential solution for trust. But it’s not everything.
Behind data-informed AI decisions lies human stories. What blockchain can’t address is the need for business leaders, technical experts and domain leaders to apply critical thinking to the systems their organisations use, particularly when there’s potential for harm. This means not relying on the people that build and implement AI systems to manage their impact, but dedicating time to understanding the consequences of their use - both good and bad.
