Can Blockchain Help ML and AI?
It can be difficult not to get caught up in the hype surrounding blockchain and cryptocurrencies, such as the notoriously volatile Bitcoin. “They will completely revolutionize the world!” (or so we’re told). Such lofty claims usually turn out to be wrong, but could the combination of blockchain and machine learning/artificial intelligence actually generate real value for real people?
Shahin Khan, CxO for the Silicon Valley consulting firm OrionX had it right when he compared blockchain to an Ingmar Bergman movie during a session on blockchain at Tabor Communications‘ recent Advanced Scale Forum (ASF) summit in Austin, Texas.
“You know it’s important. You just don’t get it,” Khan told the ASF audience last month. “Blockchain is very much like that. You know it’s important. You know something is going on here. But to put your arms and mind around it and figure out what all is happening is really complex.”
That complexity of blockchain has spurred an untold number of dubious business plans and questionable initial coin offerings (ICOs), some of which have the icky air of a get-rich-quick scheme and can trigger Bergman-esque fears of intellectual inferiority among outsiders who are left scratching their heads and wondering what’s going on.
While there are legitimate concerns about the legality of cryptocurrencies built on blockchain technologies — such as the price manipulation of Bitcoin in 2017 that was recently alleged by researchers – there is also a lot of upside to the technology that should not be ignored.
According to Khan, who seems to take a mostly positive view of blockchain, the early troubles will be worked out. “Digitalization of money is going to happen. It is inevitable,” Khan said. “The question is really what does that mean.”
Blockchain has potential uses beyond being the technological foundation for cybercurrencies. And in fact, there are some interesting ways that people are looking to use blockchain to improve AI, to enhance machine learning and deep learning, and to bolster the increasingly data-based world that we all live in.
For starters, as a distributed, encrypted, and immutable ledger of transactions, blockchain has the potential to improve the security and privacy of data. And since data is the fuel powering AI and machine learning models, it doesn’t take a logical leap to conclude that blockchain could be the mechanism by which more data is unleashed, whereby it’s promptly gobbled up by hungry deep learning models.
Blockchain can also be used to track models, which is a big concern for those looking to put machine learning into production – particularly in regulated industries or companies that fall under reultations like GDPR. The data strategist Francesco Corea says in a Medium article that blockchain could be used to help AI explain itself by providing a clear audit trail of the model.
Corea also sees blockchain contributing to a world where data, models, and AI applications can be bought and sold in online marketplaces. The prospect of blockchain-verified data and model sharing could “provide a more fluid integration that lowers the barrier to entry for smaller players and shrinks the competitive advantage of tech giants,” he writes.
A new crop of startups has popped up that are looking at interesting ways to utilize blockchain with AI. Many of these startups have similar business plans that hope to use blockchain to give people control over their data in the hopes that they’ll share more data with machine learning algorithms and AI apps.
- TraneAi, which wants to use the blockchain to streamline data tasks, such as the tagging of training data, in a crowdsourced manner. Those who help with the AI training would work with the company’s Transaction Protocol for Artificial Intelligence (TPAI), and be compensated with TPAI tokens.
- SingularityNET, whose goal is to create a “full-stack” blockchain-powered AI solution that can democratize access to machine learning capabilities. “The AI code is there, free and open source, on the Internet right now,” says Ben Goertzel, the company’s CEO and Chief Scientist, in a video on the company’s website. “It’s just hard to find and hard to use. So SingularityNET can complete that connection between AI developer and AI users.”
- OpenMinded, which is using a variety of technologies, including federated learning and blockchain, to create an anonymous and secure grid for the training of machine learning models. The OpenMinded grid give data scientists and developers access to data supplied by “miners,” who help train the models and are rewarded.
- BurstIQ, whose goal is to facilitate the sharing of healthcare data among businesses, researchers, and users. It uses blockchain to maintain the security and privacy of the data, and to adhere with regulations like HIPAA and GDPR. “The BurstIQ platform allows businesses to get more from their own data, share it with the right stakeholders at the right time, and access new data streams through a global network of people, partners and products,” the company says.
- And Neureal, who says that it’s using blockchain to build a peer-to-peer framework to harness idle computing power for big data analytics. The company, which was founded in 2014, boasts that its network is 1 million times more powerful than the world’s biggest supercomputer. It even claims that neither Facebook nor Google can “harness the massive and raw amounts of data needed to surpass what Neural’s architecture allows.”
As these startups’ business models show, the potential combinations of blockchain and AI are limited only by our imaginations (and perhaps also by our marketing budgets). The technology clearly is rapidly evolving, and the potential for disruption is high.
A Gartner recently stated that technologies such as blockchain, AI, Internet of Things (IoT) and 3D printing “will have a transformative impact on how work gets done.” “Blockchain holds the promise of fundamentally changing the way commercial interactions occur,” said Gartner Fellow and vice president David Furlonger.
However, the analyst group also noted the scarcity of real-world blockchain deployments. Blockchain evolution will take time, according to Rajesh Kandaswamy, a Gartner vice president. . There will be three phases, and we’re still in the first phase, he says.