If you’re even remotely interested in science fiction then surely you’ve seen at least one movie or read one book about an Artificial Intelligence (AI) growing in knowledge and power until it surpasses humankind and enslaves us to its own will.
Possibly more extreme an outcome than how reality will unfold, but this future view is still based in some fairly real concerns. First of all, an AI is designed to learn, grow, develop, and advance. If allowed to do so unchecked, what could it eventually be capable of? This isn’t even getting into the notion of self-awareness. Without getting overly metaphysical, an AI could still advance and by simply employing the core jobs it was given it would attempt to grow and learn further.
“Artificial Intelligence is a way of making a computer, a computer-controlled robot, or a software think intelligently, in the similar manner the intelligent humans think.” (Tutorials Point, 2019)
As we get better at making these systems think intelligently and like humans think, will they start making decisions to further advance those protocols they were developed with? Perhaps beyond the original intent of those that programmed the AI.
So, maybe this is a scare thought after all. I mean, how can we trust an AI that keeps wanting to grow and advance? Will we eventually lose control of the system? Perhaps it never reaches self-awareness or sentience, but I don’t think we even need to worry about those ideas at first. Before we even touch that idea, what happens when the AI becomes autonomous? Able to make decisions on its own volition and able to fully support itself. It doesn’t take self-awareness to determine money is an important commodity to acquire resources. Or how to use the internet to make money. Or how to use that money to purchase better, faster server hardware. This is no different than learning how to play a game. Many modern games are built around the concepts of learning how to acquire resources faster and conquer, even dominate, other players. And AI’s have become very good at playing games. “The Alphabet division just finished a StarCraft II Demonstration where its AlphaStar AI agent successfully defeated professional gamers 10-1.” (Li, A 2019)
AI has come a long way already.
As with most things in life, AI is not all bad. Actually, it isn’t good or bad, it’s how we use it or misuse it. AI also has some very impressive applications that could, and do, benefit our life today. The application of AI can be beneficial in various areas. Narrowing in on even one way can show the potential benefits we could see from AI. Healthcare is an area AI can really do a lot in. AI-assisted robotic surgery, which could lead to a 21% reduction in a patient's hospital stay. Virtual nursing assistants could save the healthcare industry $20 billion every year. (Marr, B 2018)
So even with the appreciable risks, AI holds great merit for the potential positive outcomes it could foster. But there is still that pesky trust issue. How do we trust an AI? A potential answer to the trust issue may surprise you. It did me. Blockchain.
Blockchain isn’t just the latest buzz word in the tech industry. It’s actually becoming a pretty big player.
“Although it is presently synonymous with the finance sector, there are endless uses for blockchain technology including transportation, voting, education, charity, and law enforcement, and this is because blockchain can be applied to any sector in need of an absolute, publicly accessible form of data to confirm actions.” (Vilner, Y 2018)
Let’s get a working definition of what blockchain is. “A blockchain is a digital record of transactions. The name comes from its structure, in which individual records, called blocks, are linked together in single list, called a chain.” (TechTerms 2018)
But how can blockchain help us trust AI more?
Trust is something we build with a person or a business through a history of favorable interactions. What if we could build that history with an AI? What if the AI were forced to produce an audit trail of all its prior transactions? We could then see its past behavior to better trust it will act in a manner we expect for this transaction. Blockchain affords us such a tool. (Halunen, K 2018)
How might this work? I mean, who is going to want to sift through blockchain transactions for every single AI interaction we have to make? Because one day soon that will likely be many transactions each day.
What if we only really end up needing to trust one AI? A personal assistant AI perhaps. An AI that we perform a thorough setup procedure with which ensures this AI understands our core values and trust model. We then use this personal assistant AI to complete transactions with other AIs knowing that our personal assistant AI will ensure the transaction conforms to our preset parameters. If our AI doesn’t trust the other AI after reviewing the blockchain transactions we get notified as to why and can then make an informed decision as to if we want to engage in that transaction or not.
What types of transactions could these be?
Just about anything!
You want to purchase gas to put into your car. You could pull out your credit card and swipe it at the pump. But you aren’t sure the pump has been compromised with an illegal card reader add-on. So instead you pull up your AI. Your AI requests a full blockchain from the gas station AI and immediately can see if the transaction you are about to engage in is with the gas station or some other party.
You are about to purchase your first house and have heard how house titles can be forged. Rather than trusting the paperwork is legit and your money is going where it should, you bring up your AI which requests a full blockchain on all purchases and all documents pertaining to the house from the time it was built. Your AI validates the authenticity of each transaction and document on the chain confirming the title is valid and provides you with a full financial history of your house.
Your doctor recommends a series of procedures for an ailment you reported. You are unsure if you will be able to afford all of them but don’t want to risk your health by ignoring the issue. Using your AI to communicate with your doctor’s office AI and your medical insurance company AI, you see the amount your insurance company would pay for each recommended procedure as well as what you would be expected to pay. Working further with your doctor’s office, AI speculative outcomes could be provided based on likely treatment options the doctor’s office AI could recommend along with potential costs involved.
There are far more opportunities for AI and blockchain to transform the world we interact with. While ignoring the risks of AI would be asking for trouble, dismissing AI altogether would be irresponsible when considering the vast-reaching improvements that could be had. Yes, the course we chart into an AI future needs to be a careful one but if we approach with care, we have an opportunity to do great good in the world we know.
Victor’s love for technology started with episodes of Star Trek and learning to code on a TRS-80 micro computer. Several computers, later he has now developed software professionally for over 20 years. His hope of technology offering a better way of life, tempered with the effects of technological advances on culture, resonate in his writing as he builds new future worlds for us to explore.
***Check out Victor’s latest work at www.victorillian.com
Li, A (2019). “DeepMind’s AlphaStar AI wins 10-1 against professional StarCraft II players” 9to5google.com/2019/01/24/deepmind-alphastar-wins-starcraft-ii
Halunen, K (2018). “Using blockchains to increase trust in AI” vttblog.com/2018/02/19/using-blockchains-to-increase-trust-in-ai
Marr, B (2018). “How Is AI Used In Healthcare - 5 Powerful Real-World Examples That Show The Latest Advances” forbes.com/sites/bernardmarr/2018/07/27/how-is-ai-used-in-healthcare-5-powerful-real-world-examples-that-show-the-latest-advances/#63022d455dfb
TechTerms (2018). “Blockchain” techterms.com/definition/blockchain
Tutorials Point (2019). “Artificial Intelligence Tutorial” tutorialspoint.com/artificial_intelligence
Vilner, Y (2018). “Trust, Security And Efficiency - How Blockchain Really Intersects With Artificial Intelligence” forbes.com/sites/yoavvilner/2018/08/05/trust-security-and-efficiency-how-blockchain-really-intersects-with-artificial-intelligence/#47b695992c9