Artificial intelligence (AI) is dominating the headlines virtually 24/7. In fact, discussion around the technology has emerged to fit into a category similar to politics and religion – including copious examples of conflicting opinions, pitting those who believe in its potential squarely against fear of our computer overlords.
Despite the dissension, AI is growing rapidly and is poised to affect countless aspects of life; Accenture research finds that the impact of AI on business is projected to increase labor productivity by up to 40 percent, enabling people to make more efficient use of their time. Statistics like this make it tough to ignore the potential AI can have in the enterprise and beyond.
But before it hits the prime time, AI must face the problem that is its bad rap, mainly thanks to the parade of shocking headlines and misguided efforts from Hollywood that exploit the technology. AI is often misunderstood and it badly needs a correction.
AI vs Traditional Software
Most software today is rules-based and is primarily written to perform a very explicit task or function, generally following a series “If/Then” statements. Conversely, AI can be viewed as a software program that can learn, change and adapt to its environment based on its interactions with the world once deployed. At a high level, AI pertains to anything that is built using a neural-network model or a genetic algorithm that can evolve and change over time. In other words, AI gets smarter. The “smarts” are built on probabilities rather than strict rules — so it’s more flexible. But occasionally, AI is wrong.
The future of the enterprise is bright with bots and automated processes: by 2021, more than 50 percent of enterprises will spend more per year on bot and chatbot creation than traditional mobile app development. But while usage is exploding, their results can prove unpredictable. Last month’s creepy Alexa glitch is an example of AI gone rogue; so how are enterprises expected to feel good about AI when they don’t really understand what it can do? To solve this problem and increase trust in AI technology, it’s necessary to introduce a protocol to validate AI identity, enforce compliance, facilitate auditing and establish reputation standards.
In just the last year, there have been countless examples of AI failures and systems gone awry. Industry leaders are working to get ahead of problems, but there are three large problems to tackle:
- Spoofing — as AIs proliferate, how do we identify spoofs?
- Failure — how do we know when an AI fails and what can we do when it does the wrong thing?
- Compliance — how do we control AIs when they go rogue or cause some sort of mischief?
Ongoing innovation is catalyzing change across almost everything, but amidst all the good is the issue of fraud that seems to grow exponentially each year. It’s not uncommon to read about spoofed websites, spoofed emails, and now spoofed bots. How can we trust AI, understand how it has been trained, and what it is authorized to do without any context? Before AI-fueled bots invade the enterprise, it’s critical that we confront the problem of AI identity, failure and lack of compliance in order to establish trust. How can we put some chains around AI agents to make sure they don’t go off track?
Blockchain as a Solution
While still in its infancy, blockchain technology can solve many of these problems. By its very nature, it provides a decentralized method of establishing consensus among potentially untrusted parties – in this case, AI bots and other agents. It also offers a foundation for registering and universally identifying AI agents, solving the verification issue – like the way website certificates work to validate website ownership.
The blockchain inherently allows every AI agent to record all activity so it is immutable and inspectible by those with the encryption keys. A recent piece in Nautilus, by MIT professor Iyad Rahwan, contemplated that “complex AI agents often exhibit inherent unpredictability: they demonstrate emergent behaviors that are impossible to predict with precision—even by their own programmers.” Without a solution to track and audit what happens, enterprises can’t deploy systems with these emergent behaviors.
The advancement of AI – both of its adoption and its capabilities – is unstoppable. Devices will soon communicate with users directly and people will be able to communicate back in a way that is unlike today’s capabilities. It will all have a tremendous impact on the world around us, which will ultimately increase our productivity and improve our cognitive skill sets and abilities. Despite the worry that stems from the unknown, combining AI with emerging technologies like blockchain, offers a transition to a future that we can feel confident about entering.