Using Blockchain to Unlock Artificial Intelligence
Artificial intelligence is paving the way for the next leap in human advancement by bringing about a sea ofchange in the way we interpret data across technological, financial, medical, and commerce sectors.
The writing is on the wall in seemingly mundane ways: cruise control equipped vehicles jumped several generations and are now fully autonomous and self-driving. Siri no longer simply recognizes your voice, she adapts to it along with your speech patterns and is getting better at listening to you and making suggestions. Amazon, Netflix, Pandora, and Googleʼs predictive search are all anticipating your next purchase, film or song choice, and in the case of predictive search results, sometimes even your next thought.
Data is the New Oil
These everyday instances are moments during which we casually interact with artificial intelligence, but its progress hinges on our interacting with technology in ways that create stores of data for AI to interpret.
Competing forces such as Google and Amazon monopolize data ownership and its analysis. This creates artificial data scarcity and limits the ability for smaller tech outfits, medical research facilities, universities, individuals, and others who would otherwise progress the field of applied AI from accessing data and computational resources.
A recent technological innovation called blockchain may be exactly what AI has been waiting for in order to unlock the vast stores of centrally controlled data. Blockchain is a cryptographically secured, decentralized, and global ledger which allows people who do not know each other to trustfully share data on an immutable record of events.
How can blockchain unlock the data bottleneck and bring AI to life?
Oceans of Data and the Problem of Trust
Data streams created by people on a constant basis are locked away in data silos which are isolated ponds of potentially life-saving, world- changing information and make up small bits of the massive ocean of data weʼre unwittingly floating atop. But because automotive, social media, telecommunications, and e-commerce companies are centralized and compete with each other, the data they accumulate from their users is, firstly, not owned by the users themselves and secondly, not shared publicly which creates the aforementioned silos.
Our data oceans are doubling in size every 60 days. To put this into
perspective Eric Schmidt, former CEO of Google and Alphabet Inc., stated during a 2010 conference that every two days there was more data being generated than in all of human history up to 2003.
The main driving force behind such a quantum leap in data production was, and remains, user-generated data, and these types of data are increasing with unstoppable velocity.
Text messages, status updates, tweets, Google searches, documents, Youtube videos, financial and electronic health records, voice messages, shared photos, the list of user-generated data goes on. But until recently, such crude data has been largely uninterpretable, leaving the vast trove of human-generated data untapped.
Unstructured data contains many nuances that traditional computing systems are unable to glean information from. An important element in most user-generated data is context such as emotion or history between participants in an exchange of messages.
Analyzing and interpreting such context or the content of other data like photos, video, and medical records take an extraordinary amount of computational resources that, so far, only large corporations have at their disposal. The reward for accurately interpreting unstructured data sets and creating hypotheses from them is the development of predictive technologies and autonomous systems that have the power to reshape society as we know it.
The gold rush to collect, own, and analyze as much data as possible has led to a cold war-esque race in which each of the major technological players has set up a distributed computing outfit, such as Amazonʼs AWS, Googleʼs DeepMind, and IBMʼs Watson.
The perpetual news cycle of hacked personal data and data sold to the highest bidder in closed backroom deals, however, creates serious doubt over whether anyoneʼs personal data should be owned or stored by centralized services. The closed nature of centralized AI means you will simply have to trust that the algorithms and data-sets being used to come to decisions that affect your life, such as when youʼre riding in an autonomous vehicle, are good.
Bringing Together the Data Haves and AI Haves on the Blockchain
Combining AI with blockchain fundamentally changes the current AI milieu in several key ways:
Data is moved away from centralized black boxes like Facebook and onto open-source blockchains by the commodification of data, giving users ownership over their data and allowing them to be paid for it in a blockchainʼs native currency.
The current centralized data model means both the data and AI
algorithms used to interpret them belong to the same company. By moving user-owned data onto the blockchain, algorithms become the most important part of the equation and lead to the creation of marketplaces that join projects offering AI with users offering their data-sets.
Blockchain-based AI is inherently trustworthy due to the immutable, public nature of the blockchain ledger. Every event that takes place is transcribed onto the ledger in such a way that it can always be traced.
There are several notable blockchain projects already doing significant work in the AI space.
1. AI Crypto — Sungjae Lee, CEO of the blockchain-based marketplace AI Crypto, believes the monopolization of AI by global giants Google and Amazon have led to a stifling of the field. According to Lee, the AI Crypto Ecosystem will provide a “…global shelter for AI engineers, scientists, and small start-up companies,” by providing a marketplace wherein computational resources, user data, and AI models can be directly exchanged between participants with services compensated using AIC, the native token of the AI Crypto Ecosystem.
2. Ocean Protocol — Ocean Protocol is another notable effort in the blockchain-based AI realm. Centralized data exchanges “…lack fair and flexible pricing mechanisms, data providers lose control over their assets, and there is a lack of transparency over how data is used.” By
their own measure, 16 ZB of data was generated globally but only 1% of it was analyzed. They aim to change that by providing a middle ground between the data haves and the AI haves, unlocking what they estimate to be a trillion-dollar data sharing market.
3. SingularityNet — SingularityNet offers an AlaaS (AI-as-a-service) marketplace wherein owners of AI models offer their algorithms for rent to those with data-sets theyʼre looking to analyze. Singularityʼs inbuilt search function allows for users with data-sets to then find similar models to the ones they rented or are interested in, allowing for a dataset to be pored over by several AI models or for results from similar analyses to be compared.
Notably, in all three of the aforementioned blockchain-based AI solutions, users always retain ownership of the data sets they offer.
Blockchainʼs promise to restore ownership of data back into the hands of users is a major factor in itʼs potential to disrupt AI amongst many other verticals. As the move to commodify data and shift ownership back to users becomes more enticing, the true potential of blockchain- based AI will blossom in earnest.