Blockchain Technology & Artificial Intelligence — Unmasking the Mystery at the Heart of AI
Can Blockchain Demystify AI’s Black Box Problem and Help Accelerate Mass Adoption?
Peanut butter and chocolate, Mick and Keith, Batman and Robin — great partnerships — successful marriages so to speak. Something magical happened when their paths merged, forging culinary, music and comic book history. When two technologies collide, the result is groundbreaking innovation — healthcare and robotics, supply chain and distributed ledger technology, digital technology, and photography, 3D printing & healthcare. In business for an innovative idea to be implemented on a large scale, it has to solve a specific need and it needs to be able to be replicated at a reasonable cost.
Blockchain Technology & AI
Though the field of artificial intelligence was born in the 1950s, it didn’t really find mainstream popularity until the 1990s and early 2000s. Even today, adoption of AI is still slow outside of certain industries. For every Siri, Alexa or business intelligence algorithm that makes it to market, countless others struggle to find funding due to the extreme costs or acceptance due to wariness about the possibilities of AI gone awry.
“If you’re not concerned about AI safety, you should be. Vastly more risk than North Korea.” — Elon Musk
What is slowing the adoption of artificial intelligence?
Widespread use of new technology typically filters down through the marketplace via niche consumers, military development and business use.
- Military applications: The needs of the military (including programs such as NASA, which falls under the Air Force) often drive technological development, and the military frequently is an early adopter of new tech.
- Business applications: As technologies mature, businesses examine whether (and how) the tech could improve their bottom line.
- Entertainment applications: Movies, games and even media production equipment frequently look to developing technologies for ways to thrill audiences and stand out from the competition.
- Consumer applications: Consumer technology lags behind other applications since many producers wait for technology to mature and drop in price before it becomes commonplace.
Each layer ultimately slows the AI adoption rate, especially if the new technology requires significant investments of time or money to implement.
Of course, the filtering of AI tech isn’t the only reason that adoption can be slow. Perhaps the biggest of these reasons is the “black box” effect or problem.
According to Wikipedia — In science, computing, and engineering, a black box is a device, system or object which can be viewed in terms of its inputs and outputs (or transfer characteristics), without any knowledge of its internal workings. Its implementation is “opaque” (black). Almost anything might be referred to as a black box: a transistor, an algorithm, or the human brain.
AI learns by “training”. It churns through existing data, analyzes it and makes adjustments to settings within the program based on that analysis. These adjustments are fine-tuned over multiple training sessions or passes, and some AIs analyze the same data hundreds or even thousands of times as part of their training.
The analysis differs slightly with each pass, however, and the final settings that the AI will use don’t necessarily directly correlate with the input or any specific training pass. This results in AIs that produce “eerily” accurate outputs based on new data. But it’s difficult and sometimes impossible to explain just how the AI learned to come to a specific conclusion. That’s why they call it a black box… we know the inputs and the outputs but exactly how and why it came up with the output or conclusion can’t be fully explained.
From driverless cars to medical diagnosis, implementing AI could save lives but if something goes ‘awry’, who or what is held accountable when we don’t know how an AI came to a specific conclusion, a life-altering decision or diagnosis? If we don’t know how — can we trust it?
What is Blockchain Technology?
“The blockchain is an incorruptible digital ledger of economic transactions that can be programmed to record not just financial transactions but virtually everything of value.”
Don & Alex Tapscott, authors Blockchain Revolution (2016)
Using Blockchain Technology to Unmask AI’s ‘Black Box’ Problem
Incorporating blockchain into AIs could greatly help demystify the black box. The results of each training session could be stored as immutable data within the blockchain, alongside details about how the weights or conclusions were calculated. This would allow scientists to securely access every point of an AI’s training, providing valuable insight and aid in debugging when future AI experiments take unexpected turns.
This could not only accelerate the adoption of artificial intelligence but may also improve the AI and machine learning fields as a whole. Incorporating blockchain technology into next-generation AI models could make them more accurate, less mysterious and more useful to the countless industries that benefit from artificial intelligence.
The Future of Blockchain-Enabled AI
Providing access to exactly how an AI was trained is only one way that blockchain technologies could improve AI adoption. Some data scientists are already using the decentralized nature of blockchain to spread out data processing and distribution to different nodes, providing processing speeds on par with large supercomputers at a fraction of the cost.
There are other options as well. Blockchain variants could track AI-powered transactions, provide incentives for contributors to AI algorithms and even improve security for sensitive AI systems.
While blockchain and AI were once thought to be on opposite ends of the computing spectrum, the future may see them come together to create something incredible; a technological partnership of epic Batman and Robin proportions.