When we ponder on the appearance of an artificial superintelligence, we usually base our guesses on the wrong factors. AI has already become superior in image recognition; it is able to process colossal volumes of data. However, these and other technological wonders will not necessarily pave the way to the appearance of superintelligence superior to human mind. Why is that so?
Artificial superintelligence is a concept of an object that transcends the capabilities of bright, gifted minds. That can be an arguable statement though: Nick Bostrom — a prominent contemporary philosopher, defines superintelligence as “an intellect that is much smarter than the best human brains in practically every field, including scientific creativity, general wisdom and social skills”.
If we consider the developmental achievements of Google Deepmind, we will see that their most prominent creation — AlphaGo (the first computer program to defeat a human Go champion), barely fits the definition. Most modern “intelligent” services are nothing short of a trick dog and are able to complete only the tasks they were initially intended to complete. IBM Watson, on the other hand, can take up various human parts — it can be a trivia show expert or a medic diagnosing cancer. That is much closer to the Bostrom’s definition but still pretty far from the true multitasking capabilities.
There is an important nuance. We cannot properly assess the concept of superintelligence if we try to understand the natural intelligence linearly. How do we usually measure our intelligence? We apply the intelligence quotient (IQ) and evaluate our approximate effectiveness in the completion of intelligent tasks. We expect the superintelligence to be superior in the completion of the tasks available to humans. Suppose it possesses an extraordinary intellect, mind, intuition, understanding; ability to learn, think and imagine. Does that mean that it exceeded the humanity’s level of mind while the animals stay on the lower level still?
The answer is no. The measuring scale has one problem — it is not linear. It does not intend the existence of a “ladder” with stairs as levels of intelligence. The intelligence is not monodimensional (single-dimensional if you wish). Competing with a squirrel in memorizing the acorns stashes is pointless. In a similar manner, the existing AI gradually “transcend” our intelligence in strictly limited dimensions. The IBM Watson’s RAM comprises over 15 terabytes whereas the human memory is capable of holding around 1 million gigabytes of data. The comparative speed of human neurons’ interaction, however, is only 200 Hz. Compare that to your smartphone’s CPU’s clock rate (frequency) and you will see that the phone is “faster” by 7x. But, can our intelligence be measured based on other criteria?
There is no single opinion on this question among the renowned AI experts. From the point of view of biology — French neurobiologists, in their research for Science magazine, compare the existing AI with the human subconsciousness. We are able to execute complex swift actions and rapidly recognize familiar faces in the crowd without the excessive thinking all thanks to the subconscious. The majority of modern artificial intelligence systems are able to imitate such skills and even operate with them better than the human does having no conscience whatsoever. Filtration of the important data for long-term planning as well as biographical data processing and ability to correct mistakes remain unattainable for the existing systems though. However, the achievements of the deep machine learning enable neural networks to cope with some of those tasks with an impossible for the human brain speed.
A science fiction concept — an appearance of the superintelligence capable of successfully solving the conscious and subconscious problems, gradually becomes a prophecy come true. Sociologist Robert Merton formed it as a kind of a prediction which either expressly or by implication affects the reality in such way that, eventually, it turns out inevitably true. It doesn’t matter that fantasts and futurologists were wrong about the time of appearance and some capabilities of the existing AI. The approaching superintelligence becomes more real with every successful operation of the contemporary intelligent systems.
From the point of view of the futurologists and computer systems developers, we are to come across yet another phase of AI development before the superintelligence appears — transition from narrow focus specialized tasks to general purpose AI. Such kind of the artificial intelligence is best described and represented in cinema and sci-fi literature: an assistant capable of realization, planning, problem solving, abstract thinking, understanding of complex notions, and fast learning based on the first-hand experience.
So, what other problems are there to solve? Fei Fei Li — Google’s main expert in cloud technologies, believes that without the deep comprehension of the context, the transition from the “weak” AI to the “strong” one won’t happen. Machines lack humanity, so fast and flexible learning is impossible without the understanding of the context.
If we want to create the technologies that would make our lives more productive and better and the world safer, we need to enrich the machines with a certain level of human communication and cooperation skills. Blockchain can be a great solution in providing the AI of the future with such means. Blockchain systems, providing data storage that is transparent and reliable at the same time, will become a protocol of interaction with a massive of human knowledge. The inability of side interruption guarantees consistency of data based on which the AI of the future will operate; a capability of processing mathematical algorithms independently will allow to exclude or at least constrict the human involvement.
All controlled AI bots (including IoT devices and any other digital software services) will be able to share the unique data or knowledge with each other directly without the risks of falsification. The independence and decentralization of such system will become its advantage in the aspect of integrity.
Additionally, a “strong” AI, in order to develop rapidly, must feature proper vision and language systems. Modern voice assistants still suffer lack in context comprehension, but the works are in progress. Chatbots are able to keep up with conversation and understand the overall concept of user’s question.
Before the veritable superintelligence (capable of succeeding in any sphere of human activity) appears, we must prepare ourselves for incorporation of the strong AI in the spheres of robotic science, data analytics (juridical and medical); in commerce and other susceptible to data interaction businesses. The AI will invade numerous spheres before taking up a part of an autonomous car driver. We need to learn to interact with it today.
Chatbots and voice interfaces are to become our general communication channels featuring the smart systems of the future. From the UX perspective, the end user won’t see a big difference between interaction with a single superintelligence or with a complex system of intelligent agents cooperating for mutual benefit.
The concepts for the further development of personal digital assistants lie on the surface and competitiveness rate is pretty high on the market. The users are ready to interact with various bots that would consider your preferences, always remember the latest actions and purchases and be able to rapidly process all requests within the user-friendly interface.
From the business perspective, one must not forget about the economic side of things when preparing for the transition to the next stage of AI development. The closed systems will not benefit from integration with the upcoming intellectual aggregators that use blockchain for data storage and interaction with various services. That can be fixed though: let your users interact with the business via the trusted chatbot. The BRN.AI company works on the development of a dedicated cloud platform for collection, analysis, and processing of data from various popular messaging services’ chatbots.
They are not the only figure on the market of blockchain-projects that is ready to incorporate the superintelligence of the future. Large technological enterprises strive to create their own ecosystems developing the personal assistants as proprietary solutions: Siri, Google Assistant, Viv, Bixby for mobile devices; Google Home, Alexa, Apple Homepod for home IoT devices; and the powerful specialized services like IBM Watson and Google Deepmind projects for a broad audience of users.
According to the journalistic investigation by New York Times, despite the huge salaries in the sphere, the artificial intelligence niche lacks in experts and the situation doesn’t seem to improve. The shortage of qualified employees that would handle the AI-related tasks makes it difficult for traditional businesses to build their own intelligent systems.
The companies that are not ready to “fight” for employment of the AI experts may find the solution in the utilization of a certain scalable platform. The one which will allow to flexibly solve the current tasks and be able to interact with a global platform in the future.
Employing one or another cloud platform today, you will prepare your business for the superintelligence incorporation and save your customers from the “future shock” predicted by the futurologist Alvin Toffler. Become a partner of the BRAIN platform and get your business ready for the HIGH EXIT.