Currently, a hot button topic in the technology industry is artificial intelligence. Intellectuals have told us that it is the next significant threat to society as we know it. Created by humans, people predict that the machines and the algorithms will become so advanced that they will no longer need humans, and that will be the end of our civilization. That could be entirely true, but, we are not far enough along to make that determination. When it comes to artificial intelligence, there is however one element that today could shape the future of where the technology ends up, and that is who controls it. I believe that the only way we will be able to integrate artificial intelligence into our lives peacefully is if we implement decentralized controls.
Today, there are three different types of AI computing (Nonveiller, 2018):
• Centralized Environment, where virtually all the work is done by one machine. The environment has many “helper” machines, organizing data and running smaller calculations, but all the decision making is handled by one centralized system.
• Distributed Computing, where one uniform system handles the computing at multiple locations
• Decentralized Computing, where various computers handle varying data and datatype to perform individual tasks
Logically, it would make sense that the most common and widely-accepted method would be Decentralized computing, as security could be easily implemented and controlled; if one part of the system were to go down, the entire enterprise would not be lost, and it would also decrease the likelihood of a hacker taking over the whole system.
But that is not the case. As explained by Jesus Rodriguez, most AI solutions today are composed of Centralized Environments. This is because of the nature of the tech industry today, where few companies hold the most power. Companies such as Google and Facebook are collecting more data and revenue than most companies; therefore they can invest more in the technology and use more robust data sets to create the machine learning algorithms (Rodriguez 2018, “Everything…”). This gives larger companies even more leverage in the marketplace, as they will have unprecedented power and influence in society.
Centralized Environments lead to many ethical questions as well. If there is a monopoly on the Artificial Intelligence industry, how can we ensure that what we consent to give up privacy-wise is being upheld? How can we ensure that companies are acting morally when there is not a system of checks and balances? How can we prevent the systems and algorithms from being inherently bias in favor of the corporations who create the algorithm? On a broader scale, how can we promote growth in an industry with such high barriers to entry, primarily the cost of collecting and analyzing data (Rodriguez 2018 “AI…)?
Currently, there are few decentralized platforms for Artificial Intelligence to operate. One of the most common is SingularityNET, which uses blockchain and cryptocurrency to allow users access to opensource AI agents. Their founder and CEO, Dr. Ben Goertzel, was recently on The Joe Rogan Experience to talk on a variety of topics regarding AI. In this clip, Dr. Goertzel discussions about the role of blockchain in SingularityNET. He describes blockchain as blocks of stored information that are stored in many places to help prevent corruption and theft, or decentralized control. The platform uses Ethereum, which is a blockchain-based currency that also allows smart contracts, as the method of payment for these open source agents. The reason that Dr. Goertzel set it up this way was that decentralized control was the top priority for him, and he didn’t want this “global brain to be controlled by the global elite.” I’ve included the entire episode at the bottom of this post; it is worth a listen.
In conclusion, there is no stopping artificial intelligence. It has already arrived in its infancy in the form automation, where machines are set to complete a particular task until an anomaly is detected, which then leads the computer to run through various scenarios and execute the next-best alternative. AI should not be viewed as a threat, and more as a tool that, when in the wrong hands, can become a threat. The concern that many individuals have, such as Jesus Rodriguez and Dr. Goertzel, is that centralized computing will give individuals unparalleled control in society, which should scare all of us.
Harrison Heim's MIS 304 Site 