Home ExclusivesFeatures

Follow us on Twitter or join our Telegram

The Blockchain: Our Saviour Against A Robot Takeover?

Source: iStock/Menno van Dijk
  • Blockchain might help control every action AI takes.
  • The benefits could also be extended to confirming identity of AI.
  • Decentralized AI would not allow big players to dominate the market.
  • AI-blockchains still can make things worse in certain areas.

It may be the stuff of science fiction, but even some of the most respected minds in science and technology are scared that artificial intelligence might wipe out humanity. However, while virtually every lab involved in designing AI has launched its own ethics group in a bid to play down fears, a more secure defence against social ills may come from the blockchain.

A prime example of how the blockchain will work in this regard comes from Talla. Founded in Boston in 2015, Talla started its career by offering AI-based enterprise tools to businesses, yet at the end of last year it announced an ethereum-based blockchain for artificial intelligences. Dubbed the BotChain and currently undergoing beta testing, it works by issuing digital certificates for every action an AI takes and by linking these certificates into – unsurprisingly enough – a chain of encrypted documents. By doing this, it enables companies to monitor the activities of the autonomous bots they're using and ensure that legal and ethical standards are being upheld.

"As more organizations deploy intelligent robotic processes, the need for BotChain is only going to grow,” predicts the former blockchain CTO at EY, one of the "Big Four" accounting firms, Michael Maloney. “Compliance is incredibly important to large enterprises." His expectations are mirrored by PricewaterhouseCoopers’, another auditor from the “Big Four” technology and strategy leader, Chris Curran, who says, “As bot processes grow within organizations, it makes sense that they would be subject to risk and compliance requirements just like any other workers.”

Yet the benefits offered by the blockchain don't stop with ensuring compliance, but also extend to confirming the identity of any AI a company might want to hire or enlist. As Talla's CEO, Rob May, explains, "Bots are starting to be spoofed. Hackers are using them to steal information and because of this, bot identity is rapidly becoming an issue. BotChain is an identity solution for bots, so that as humans interact with more bots, or bots interact with each other, they can quickly establish trust."

Through the BotChain and AI blockchains like it, manufacturers and suppliers of AI will be able rent out their bots to clients. At the same time, clients will be able to see the histories of the AIs they're interested in, confirming that these have the identities they're presented as having and aren’t therefore a risk.

Decentralized AI

This ability will also be provided by SingularityNET, another AI-blockchain that was launched at the end of the year, this time by Hanson Robotics (of "Sophia the Robot" fame). Yet what Hanson is emphasizing most of all about SingularityNET's benefits is not the operational standards bots will have to abide by, but rather the decentralization created by its blockchain. Its platform will be open to any AI manufacturer who wants to offer AI-as-a-service, while it will also enable individual AIs to communicate and collaborate with each other, thereby pooling their intelligences in a way Hanson hope will eventually lead to the emergence of an artificial general intelligence (AGI).

Hanson is aiming here to avoid a situation where the AI industry is dominated by only a small handful of big players, who perhaps wield artificial intelligence more to their own advantages than to society’s as a whole.

A decentralized AI market means that businesses and other customers will have the freedom to choose AI services that fit with their values and ethics, rather than those of a few big companies. And by making the development of AI more open, platforms such as SingularityNET reduce two risks: that AI will propagate the biases and prejudices of a small group of people based in Silicon Valley, and that the AI industry might serve to enrich only a handful of companies.

SingularityNET isn’t the only AI platform harnessing the decentralized nature of the blockchain. Doc.ai is an AI startup that, last August, launched a chatbot designed to help people make medical decisions and diagnoses. It uses personal data and machine learning to produce its assessments, yet unlike similar apps its AI operates on the ethereum blockchain.

AI-blockchains won't solve everything

Such platforms as Doc.ai and Singularity.Net have come at the right time in history, since the tech industry and its self-appointed guardians have been waiting for a viable solution to AI's ethics problem for years. Aside from the ethics groups that have periodically sprouted up in a bid to allay public fears, there have been repeated calls from researchers and public figures for trustworthy independent AI watchdogs to be established. Similarly, in October the AI Now Institute called for the end to 'black box' opacity in how AIs make decisions and operate. Such platforms as SingularityNOW and the BotChain may not directly ensure complete transparency in how exactly an AI, say, decides whether or not to approve a job applicant. However, by making every action performed by an AI visible and open to scrutiny on these platforms, they'll add much greater pressure on AI companies to make their bots fully transparent.

Of course, it's important to remember that AI-blockchains won't solve every ethical conundrum thrown up by robots, and will perhaps make things worse in certain areas. For instance, by creating large-scale platforms where businesses can make use of AIs, the likes of SingularityNET and Talla may only accelerate the process of automating jobs. Unemployment could potentially increase as a result, and in turn inequalities between different segments of societies would grow. That said, the fact that blockchain-based AI platforms will subject bots to oversight and to regulatory standards should at least guard against the worst dangers of robots harming people, or of a full-blown humanoid takeover.

Follow us on Twitter or join our Telegram

More Stories