Blockchain and Artificial Intelligence (AI) have been thrown around as buzzwords and umbrella terms for the past few years, gaining increasing amounts of traction as their potential as disruptive technologies are analyzed and debated. Recently the discussion has been focusing on the advantages of combining the two technologies, where one can remedy the shortcomings of the other.
Currently, artificial intelligence is loosely defined as the ability of a system to execute tasks commonly associated with human intelligence. One subfield of AI, machine learning, approaches this goal by training computer programs to make non-trivial decisions and predictions using datasets.
A growing concern among the public about AI algorithms is when they’re used to automate or assist in important decision-making processes. Understanding how an algorithm has reached a conclusion is increasingly difficult due to the substantial number of variables it has analyzed, utilising some and discarding those it has deemed irrelevant. Currently, decisions made by smart algorithms have to be audited by humans, especially when concerning life-altering processes such as suitability for mortgages or council housing, which can be time-consuming, costly and difficult to understand.
If the decision-making process was uploaded to a secure blockchain, explicitly recording the progression of the algorithm for inspection, the entire process would become easier to audit whilst increasing transparency and the likelihood of spotting biases present in the algorithm or the dataset.
Another valuable application of AI and blockchain technology is utilising the inherently secure nature of the encrypted ledger to store sensitive data collected and analyzed by smart, predictive algorithms. The data content can range from health history to browsing preferences which recently, have become vulnerable to substantial breaches despite continued efforts and capital invested by businesses to keep them secure.
By leveraging the security advantages of today’s blockchain architectures, AI companies can reduce the exposure risks that come with expansive data storage, holding information on the blockchain meaning only encryption keys have to be guarded. Furthermore, researchers are currently working on developing algorithms that can process fully homomorphic encrypted data that could potentially eradicate the risk of a breach.
The collaboration of AI and blockchain technology is still in its infancy, where use cases are more commonly theorized than applied. There are, however, a handful of companies emerging in the space where the two technologies are significant components of their B2B and B2C plans.
Numerai is an open-source hedge fund which incentivizes data scientists around the world to submit AI predictive algorithms for the financial market on the basis that if their model is accurate, they will be rewarded. Based in San Francisco, the fund consults 35,000 data scientists and quantitative analysts who submit models in weekly competitions, staking NMR tokens on the assumption their models will perform well in the live market. If this is the case, Numerai awards them more tokens otherwise, their staked tokens are destroyed.
Richard Craig, founder and CEO of Numerai, became motivated to find a way to encrypt his data set to share with others after his predictions significantly outperformed the market whilst working with an asset management company and he concluded that there would be many people around the world who could produce models that would do even better. He claims that within ten days, a graduate student from Bangalore with a background in neural networks beat his model. In the first month, 10,292 predictions were submitted by Numerai users and their user platform boasts professors and students from Universities like Stanford, Harvard, the Indian Institute of Technology, analysts on Wall Street, and Google employees.
DeepBrain Chain is a decentralized blockchain based platform that aims to alleviate AI companies of the exponential costs rendered from the high usage of computing power when training and running AI algorithms. The financial entry barrier for AI is exceedingly high with large proportions of capital being diverted from development and research to maintenance which in turn slows progress and reduces the likelihood of success.
DeepBrain Chain provides a deep neural network that supplies processing power to AI companies and divisions looking to develop AI products. The platform claims to cut up to 70% of computing costs for companies by paying for idle processing power from others in DBC and redistributing it to those in need. Currently, over 100 companies and hundreds of thousands of users make use of the platform already.
SingluarityNET is striving towards becoming the key decentralized platform for the exchange of AI tools between developers and business users, allowing small divisions to have access to the correct technology at cost effective prices. The platform is powered by their native token, AGI which is tailored to enable transactions, settlements, incentives and governance.
A founding partner of SingularityNET is Hanson Robotics, the creators of Sophia Hanson. Sophia is described to be the ‘most advanced humanoid robot’ as well as the most expensive in terms of her proprietary software, firmware, and hardware. They have utilised the Singularity network to run Sophia’s ‘mind’ where she uses multiple modules to see, hear, and respond in tandem, many of which are available open-source on the network.