AI in Blockchain — BlockShow Asia 2017
Does AI need Blockchain?
In order to answer this question, I will first define what are the main fields in AI. It is mostly divided into two categories. Soft AI focuses on highly skilled AI in very narrow fields, while hard AI has the ability to transfer and adapt skills that are efficient in a specific environment, to unknown environments. We are not really at this stage for the time being, even though some practical examples have already been presented.
Since hard AI has to transfer and transform inherent skills, its ability to analyse environments is key. For this purpose, the implementation of blockchain technology could be useful in recording external inputs and time-stamping them in order to provide clean and easy-to- use datasets. Blockchain is indeed an efficient tool to manage IoT or digital interfaces in order to structure data and its properties.
What kind of problems can Blockchain solve in AI?
After defining the two main categories of AI, it is also interesting to analyse current trends. We are coming from AI acting as curators to AI acting as orchestrators (“AI is the new UI”, WIRED, April 2017). It means AI is not anymore about analysing data, but also about analysing results/impacts and iterating the process in order to guide and manage. To feed this trend, having access to past events that are ordered and immutable is key. Blockchain-based solutions are here again a strong and efficient asset.
Another trend that further reinforces this point: we are in a sharing economy. Having access to devices or services is more important than owning them. For AI, being able to use “on-demand” tools, like iExec’s cloud computing will transform the way developers have to anticipate their needs in resources and accelerate the deployment of AI algorithms.
Blockchain for AI or AI for blockchain?
Thanks to blockchain, the tokenization of societies is a growing trend. Tokens could/should be digital assets with trading and usage values. The usages could be framing, powering or feeding an AI.
As AI is a data-based technology, blockchain provides the possibility to bring power to data owners. The transparency on how data could be used is also a strong point in creating a bridge between blockchain and AI.
Let’s finish on a less technical approach: evolution of complex societies from a philosophical standpoint. Blockchain is about avoiding centralization by distributing power thanks to the mix of network and data, i.e. less power to third parties.
AI is going towards having overall analysis on complex situations, going from a curator to a pseudo-omniscient orchestrator. We could easily imagine that one day the third parties managing institutions and countries could be helped and guided by egoless AI advisors. AI and Blockchain are tools to distribute responsibilities and guide decision to the overall best.