I wrote this original article on disintermediation more than 5 years ago (early 2012), and while most of the comments are as true today as then, I wanted to expand a bit now that time and technology have moved on.
The two newest technologies that are now mature enough to act as major fulcrums of further disintermediation on a large scale are AI and blockchain. Both of these have been in development for more than 5 years of course, but these technologies have made a jump in capability, scale and applicability in the last 1-2 years that is changing the entire landscape. Artificial Intelligence (AI) – or perhaps a better term “Augmented Intelligence” – is changing forever the man/machine interface, bringing machine learning to aid human endeavors in a manner that will never be untwined. Blockchain technology is the foundation (originally developed in the arcane mathematical world of cryptography) for digital currencies and other transactions of value.
While the popular term is “AI” or Artificial Intelligence, a better description is “Deep Machine Learning”. Essentially the machine (computer, or rather a whole pile of them…) is given a problem to solve, a set of algorithms to use as a methodology, and a dataset for training. After a number of iterations and tunings, the machine usually refines its response such that the ‘problem’ can be reliably solved accurately and repeatedly. The process, as well as a recently presented theory on how the ‘deep neural networks’ of machine learning operate, is discussed in this excellent article.
The applications for AI are almost unlimited. Some of the popular and original use cases are human voice recognition and pattern recognition tasks that for many years were thought to be too difficult for computers to perform with a high degree of accuracy. Pattern recognition has now improved to the point where a machine can often outperfom a human, and voice recognition is now encapsulated in the Amazon ‘Echo’ device as a home appliance. Many other tasks, particularly ones where the machine assists a human (Augmented Intelligence) by presenting likely possibilities reduced from extremely large and complex datasets, will profoundly change human activity and work. Such examples include medical diagnostics (an AI system can read every journal every written, compare to a history taken by a medical diagnostician, and suggest likely scenarios that could include data the medical professional couldn’t possibly have the time to absorb); fact-checking news stories against many wide-ranging sources; performing financial analysis; writing contracts; etc.
It’s easy to see that many current ‘professions’ will likely be disrupted or disintermediated… corporate law, medical research, scientific testing, pharmaceutical drug trials, manufacturing quality control (AI connected to robotics), and so on. The incredible speed and storage capability of modern computational networks provides the foundation for an ever-increasing usage of AI at a continually falling price. Already apps for mobile devices can scan thousands of images and make suggestions for keywords, mark for collections of similar images, etc. [EyeEm Vision].
Another area where AI is utilized is in autonomous vehicles (self-driving cars). The synthesis of hundreds of inputs from sensors, cameras, etc. are analyzed thousands of times per second in order to safely pilot the vehicle. One of the fundamental powers of AI is the continual learning that takes place. The larger the dataset, the more of a given set of experiences, the better the machine will be at optimizing the best outputs. For instance, every Tesla car gathers massive amounts of data from every drive the car takes, and continually uploads that data to the servers at the factory. The combined experience of how thousands of vehicles respond to varying road and traffic conditions is learned and then shared (downloaded) to every vehicle. So each car in the entire fleet benefits from everything learned by every car. This is impossible to replicate with individual human drivers.
The potential use cases for this new technology is almost unbounded. Some challenging issues likely can only be solved with advanced machine learning. One of these is the (today) seemingly intractable problem of updating and securing a massive IoT (Internet of Things) network. Due to the very low cost, embedded nature, lack of human interface, etc. that is a characteristic of most IoT devices, it’s impossible to “patch” or otherwise update individual sensors or actuators that are discovered to have either functional or security flaws after deployment. By embedding intelligence into the connecting fabric of the network itself that links the IoT devices to nodes or computers that utilize the info, even sub-optimal devices can be ‘corrected’ by the network. Incorrect data can be normalized, attempts at intrusion or deliberate altering of data can be determined and mediated.
The blockchain technology that is often discussed today, usually in the same sentence as Bitcoin or Ethereum, is a foundational platform that allows secure and traceable transactions of value. Essentially each set of transactions is a “block”, and these are distributed widely in an encrypted format for redundancy and security. These transactions are “chained” together, forming the “blockchain”. Since the ‘public ledger’ of these groups of transactions (the blockchains) are impossible to alter, the security of every transaction is ensured. This article explains in more detail. While the initial focus of blockchain technology has been on so-called ‘cryptocurrencies’ there are many other uses for this secure transactional technology. By using the existing internet connectivity, items of value can be securely distributed practically anywhere, to anyone.
One of the most obvious instances of transfer of items of value over the internet is intellectual property: i.e. artistic works such as books, images, movies, etc. Today the wide scale distribution of all of these creative works is handled by a few ‘middlemen’ such as Amazon, iTunes, etc. This introduces two major forms of restriction: the physical bottleneck of client-server networking, where every consumer must pull from a central controlled repository; and the financial bottleneck of unitary control over distribution, with the associated profits and added expense to the consumer.
Even before blockchain, various artists have been exploring making more direct connections with their consumers, taking more control over the distribution of their art, and changing the marketing process and value chain. Interestingly the most successful (particularly in the world of music) are all women: Taylor Swift, Beyoncé, Lady Gaga. Each is now marketing on a direct to fan basis via social media, with followings of millions of consumers. A natural next step will be direct delivery of content to these same users via blockchain – which will have even a large effect on the music industry than iTunes ever did.
SingularDTV is attempting the first ever feature film to be both funded and distributed entirely on a blockchain platform. The world of decentralized distribution is upon us, and will forever change the landscape of intellectual property distribution and monetization. The full effects of this are deep and wide-ranging, and would occupy and entire post… (maybe soon).
In summation, these two notable technologies will continue the democratization of data, originally begun with the printing press, and allow even more users to access information, entertainment and items of value without the constraints of a narrow and inflexible distribution network controlled by a few.