A word-obsessed, 25-year-old grad student who more notably squats 475.
AI and Blockchain are among some of the most influential drivers of innovation today — spreading unabated and boasting a distinct degree of technological complexity and multi-dimensional business implications. The collaborative applicability between both technologies has yet to be fully realized but is beginning to spur profound changes in numerous aspects of our lives, from financial transactions and autonomous vehicles to engaging assistants and granting us newfound ownership of our data — the ball has only begun rolling.
Blockchain is hindered by issues of security, scalability, and efficiency while AI remains plagued by concerns of trustworthiness, explainability, and privacy. Integration and a harmonious companionship could elicit unlimited innovation potential and profound downstream effects in the most unassuming niches of society.
These terms are often thrown around as synonymous but there are distinct differences. Machine learning, not the first to reach the conceptual stage but the first to be developed, is an algorithm that parses data, learns from that data, and then applies newfound knowledge to make informed decisions.
Spotify incorporates machine learning models including collaborative filtering and natural language processing to cross-analyze user behaviour, associated text, and audio to determine what songs it should recommend to you next. Machine learning algorithms contrast your musical preferences with other listeners who have similar taste, producing reliable recommendations. Although often lazily referred to as AI, machine learning is an entirely separate protocol.
Deep learning is a subset of machine learning with some defining differences. They are often used interchangeably and hold similar functions but while barebones machine learning models will become better over time, they still require a helping hand from a human overseer. If a machine learning model produces a faulty or inaccurate response, an engineer will need to fix the error and adjust the protocol, an often costly process that can be exponentially cumbersome.
Deep learning has seen application in the medical research realm, with the technology being employed to detect cancer cells. A medical research team at UCLA developed a next-generation microscope that produces a high-dimensional data set in a readable format so their deep learning application can learn and subsequently accurately identify cancer cells.
With a deep learning model, the primary and key differentiation is that the algorithm can assess the accuracy of predictions autonomously through its neural network. In essence, a deep learning model is designed to perpetually analyze data in a style that resembles the human thought process. This is achieved through a complex layered structure of algorithms known as an “artificial neural network.” The design is inspired by the yet to be understood structure of the brain and is the closest we have come to imitating the clump of cells between our ears.
The field of AI research was founded at a small workshop at Dartmouth College in 1956. They had millions of dollars in funding and wildly optimistic claims of developing an electronic brain that would mirror that of our own — within a generation.
Unfortunately, they didn’t produce, and the project eventually fell through leading into the “AI winter, a period characterized by overzealous and unremitted promises which led to a dramatic drop in investments in the sector.” Seven years later, the Japanese government injected billions of dollars into AI research which also never amounted to anything substantial.
What was the problem? The goal of AI research is to mimic the brain in both design and function which requires an in-depth understanding modern neuroscientists do not have. Both projects underestimated the complexity of the brain and how much they actually understood at the time. It was only recently, following the conception of machine learning, and then quickly after, deep learning, that new funding and optimism were injected into this research niche.
Blockchain, on the other hand, can be summarized succinctly. Blockchain technology offers automated payment capabilities within the inexorably linked cryptocurrency realm and produces a shared ledger of data, transactions, and logs, in a decentralized public or private manner. Blockchains smart contract functionality enables autonomous governance producing a trustless system. The alliance of these groundbreaking modalities (AI and Blockchain) would provide unlimited innovation potential.
You may be wondering why AI research is suddenly in vogue again, especially after the tapering off of the sector following the AI winter. The answer is data.
Data is priceless in today’s connected economy, so much so, that social media platforms are founded on revenue models solely based on selling your information. The adage “if something is free, you are the product,” is consistent with most major social media platforms running the show today. Access to big data helps firms better understand consumers, potentially increasing profit by up to 23%, identify gaps in markets enabling better product positioning, and level up their marketing efforts through analytics analysis.
In addition to the proliferation of social media, one of the primary drivers of the recent boom in AI research has been the race to achieve level five autonomy in vehicles. Dozens of specialized startups, boasting proprietary AI technology, have been attracting huge capital investments, competing in a voracious market to try and come out on top. Just as quickly as these firms pop up, they are swallowed up by major auto companies in the race for a fully autonomous vehicle. With huge potential revenue for both boutique firms, racing to design a groundbreaking AI protocol, or a manufacturer that is either trying to optimize their own software or acquire it, there is huge money sloshing around in the autonomous vehicle industry — inspiring rapid innovation.
Autonomous vehicle manufacturers continue to train their vehicles on a daily basis, exposing them to new conditions, incidents, or weather, to refine the protocol and make it safer and more intelligent. This process produces a massive amount of big data enabling developers to train an AI protocol appropriately, and ultimately inspiring technological innovation in AI to better interpret data and bolster its outputs.
With the enormous amount of data being churned out on a daily basis and then siloed away for further analysis, and often for unauthorized monetization purposes, the power is held by platforms that promote huge amounts of UGC (user-generated content) and boast robust storage capabilities.
Unsurprisingly, the social media space is littered with information breaches, including the most recent Facebook scandal which saw 50 million users profiled and targeted, unbeknownst to them, by the firm Cambridge Analytics.
Blockchain intervention could encourage the sharing of data, providing transparency and accountability to all levels of data access — putting the control into the user’s hands. Users will be more confident knowing their data will be used properly and accessed by the appropriate parties. The impact of blockchain intervention in this niche could eradicate data silos (raw data accessible by one party) and have profound positive effects in numerous verticals.
In the medical sector, an agreed-upon blockchain ledger used by the vast majority of a country or globe could elicit huge amounts of anonymous medical history data which could then be analyzed by an AI protocol, possibly discovering common variables among patients or coming to conclusions that may have been overlooked by the human eye. Blockchain intervention in data storage would level the playing field, both for firms and not-for-profit organizations, possibly saving lives and bringing about a paradigm shift in numerous sectors.
Running intensive deep learning or machine learning protocols requires a substantial amount of computing power. Blockchain can enable developers to leverage decentralized computer power to run AI systems. This is not a new idea. Grid computing — utilizing numerous computer resources to reach a common goal — has been around for decades now but offered limited application and did not see widespread adoption.
Today, however, there exists new levels of computing power and data to boot, enabling developers to utilize millions of GPUs owned by gamers to execute and train machine learning algorithms. Gaming computers are often used for a small sliver of time and could, via a decentralized marketplace, sell their computing power to the highest bidder. Smart contracts could facilitate these deals and offer firms, excluding Facebook and Google, a seat at the AI table.
Employing a deep learning protocol that can arrive at sound conclusions and make decisions that have real-world implications such as in the fields of medical research or financial planning, all within black boxes (any AI system whose operations are not visible), can spur well-founded adoption concerns.
If a decision is incorrect, an explanation is necessary, as it could imply a loss of life or economic disaster. For full transparency and accountability, an immutable trail tracking data developments and the intricate behaviour of a deep learning model is essential. Blockchain technology can offer this exact solution. Through this tracking protocol, developers and researchers would gain a new understanding and confidence in their AI models, producing a more reliable, accurate system with enhanced explainability.
Have doubts about your data security online? I don’t blame you. News surrounding data leaks in some of the biggest tech companies in the world and accusations of voter ballot fraud in the recent US elections don’t exactly paint the current data storage systems employed by big firms in the most favourable light. If all of a sudden you stop using traditional centralized social media and video sites like Netflix or Facebook, you would lose that level of personalization you have probably grown accustomed to.
Is there a way to achieve this experience while maintaining privacy? Yes, but it won’t be easy.
Envision a decentralized content provider, such as a social media platform based on a blockchain, leveraging an AI protocol to personalize your content locally, instead of at Facebook. A machine learning algorithm would run on your CPU and enhance your experience without ever pushing information onto you, but pulling it. No data would ever leave the confines of your device, bringing about new levels of privacy and personalization.
AI is the DNA of the fourth industrial revolution while blockchain may help facilitate a transformation of an entire economic system. They are both still in their infancy and despite an ample amount of data that can be applied to develop deep learning and machine learning protocols into hyper-intelligent predictive and efficient computers, keeping up with the rising complexity of the human world is no easy feat.
Recent security incidents with Ethereum, ZCash, Bitcoin Cash, and other cryptocurrencies, may suggest it will take some time before we can enjoy both scalable and secure blockchain applications for real-world use. Nonetheless, I hold an unflagging optimism for the blossoming companionship between these two technologies, and you should too.
Create your free account to unlock your custom reading experience.