An impressive feature of Artificial Intelligence (A.I.) is the technology’s ability to provide computational power to create cognition in machines. Yet A.I. critics today have become concerned that many artificial intelligence projects are centrally controlled and therefore producing “Narrow A.I.”
Unlike human cognition, narrow A.I. is not conscious or driven by emotion. Rather, narrow A.I. operates within a pre-determined, pre-defined range, even if it appears to be much more sophisticated than that. Virtual assistants like Google’s Siri and Amazon’s Alexa exhibit examples of narrow A.I. While these A.I.-based systems are able to communicate with users and answer questions, these machines are nowhere close to having human-like intelligence.
According to Arif Khan, V.P. of Marketing at SingularityNET, centrally controlled A.I. projects led by large tech companies have resulted in the creation of narrow data sets, which could be harmful for the future of artificial intelligence.
Let’s say Facebook wants to develop A.I. algorithms. These data sets will never be disclosed to a competitor, as it is in Facebook’s best interest to keep this information private. In turn, Facebook would never have access to data sets from their competitors. Moreover, the data sets that Facebook builds upon would contain private information, most likely to be used for their own benefits to drive shareholder value,” Khan explained. “Yet if Facebook created an A.I. for optimizing shareholder value, users would get a pigeonholed view of society. For example, humans might share fake news, resulting in Facebook getting very narrow applications. If you just have narrow data sets, narrow machines will be created that might not benefit the overall good of society.”
For example, in 2016 Microsoft launched an A.I. based Twitter bot named “Tay” as an experiment in “controversial understanding.” Microsoft noted that the more users chatted with Tay, the smarter the bot became, learning to engage with humans through “casual and playful” conversation.
Unfortunately, conversations with Tay took a turn for the worse when Twitter users started tweeting racist remarks at the bot. Tay, being an A.I.-based project controlled by a central authority, started to repeat these sentiments back to users. This is an example of the impact narrow algorithms can have on society.
Creating Diverse Data Sets With Blockchain Technology
Unlike artificial intelligence based-projects, blockchain technology creates decentralized, transparent networks that can be accessed by anyone, around the world. While blockchain technology is the ledger that powers cryptocurrencies, like Bitcoin, blockchain networks are now being applied to a number of industries to create decentralization. For example, SinguarlityNET is specifically focused on using blockchain technology to encourage a broader distribution of data and algorithms, helping ensure the future development of artificial intelligence and the creation of “decentralized A.I.”
SingularityNET combines blockchain and A.I. to create smarter, decentralized A.I. Blockchain networks that can host diverse data sets. By creating an API of API’s on the blockchain, it would allow for the intercommunication of A.I. agents. As a result, diverse data sets would be created,” Khan said. “Think of the blockchain as this wide horizontal layer that stretches across all different cultures, nationalities and geographies. Everyone has access to this horizontal data set and can interact with this layer of technology, which allows people to upload and work with very diverse data sets. Unlike centrally controlled data sets, data sets on a blockchain networks are not firewalled or controlled by a central authority.”
Hanson Robotics is planning to use SingularityNET to improve the intelligence of their humanoid robot, Sophia. Unlike Amazon’s Alexa, which answers questions approved only by Amazon, Sophia will be able to reach out to other artificial intelligence providers for answers to questions. This is a much more flexible solution, as Sophia will not be controlled by a central authority.
Blockchain technology will make it easy, and in many cases profitable, for A.I. developers to provide their code as services that other products can access. For a product like Sophia, it will increase the scope of third-party A.I. services that can be accessed in real-time to help do things like answer questions and interpret situations. The SingularityNET team is also creating some uniquely powerful A.I., like the OpenCog integrated AGI system, that will help robots like Sophia understand the world in human-like ways,” said Ben Goertzel, chief scientist at Hanson Robotics.
According to Goertzel, as more A.I. services are placed on blockchain networks, these services will begin interacting with each other and enhancing each others’ intelligence. In turn, this will create a dramatic increase in the breadth of A.I. applications available in the world, as well as the general intelligence and creativity of the A.I.s out there.
Moreover, open marketplaces for shared data, such as Ocean Protocol, will allow anyone the ability to set up a marketplace for any kind of data. Users of the data will pay to access these sources with cryptocurrency. The marketplaces built on Ocean Protocol will allow data to be accessed by all participants, ensuring that no central player can control or exploit the data. The overall goal of this project is to decentralize access to data, ensuring that narrow data sets are not produced.
Businesses can turn their data from cost centers to profit centers. They can also buy data that hasn’t been available until now, to compete with Google, Amazon and Facebook. And because the Ocean Protocol network rewards data with clear provenance, power is pushed back to the data source, which in many cases is the individual,” Bruce Pon, Founder Ocean Protocol, told me. “Data has immense value, but no one shares it because everyone’s scared to lose control. By creating a decentralized network where anyone can share safely, while keeping control and privacy, a new Data Economy can emerge.”
Decentralized Data Depends On Us
Although the combination of blockchain and A.I is still nascent, Khan believes that the real challenge facing the adoption of decentralized A.I. is getting people to understand how their data is currently being used. Apple’s CEO, Tim Cook, even stated in October that people don’t understand how their own data is being “weaponized” against them.
Having others decide how data is being sold in order to to create profits for businesses demonstrates that data is being weaponized against us. Blockchain allows us to cryptographically protect our data and have it used in the ways we see fit. This also let’s us monetize data personally if we choose to, without having our personal information compromised. This is important to understand in order to combat narrow A.I. and create diverse data sets in the future.”