There are discussions between neuroscientists, cognitive experts, and philosophers about whether the human brain can be created or reconstructed. Current breakthroughs and discoveries in the science of the brain are steadily paving the way for the time when the artificial brain will be recreated from scratch. Some people assume that this is beyond the scope of the possible, the second ones are busy with the ways of its creation, the third have been fruitfully working on the task for a long time. In the article we will consider questions about the development of artificial intelligence, its prospects, as well as about large companies and projects in this area.
Key Points
The artificial brain is correlated with a robotic machine, which is not inferior to humans in terms of intelligence, creativity and consciousness. In the entire history of mankind, the task has not been completely solved, but the futurists say that this is a matter of time. Given current trends in neurobiology, computing and nanotechnology, it is predicted that artificial intelligence and the brain will appear in the 21st century, possibly by 2050.
Scientists are considering several ways to create artificial intelligence. In the first case, large-scale biologically realistic simulations of the human brain are carried out on supercomputers. In the second case, scientists are trying to create massively parallel neuromorphic computing devices that are easily modeled on neural tissue.
Human consciousness in the aspect of the most interesting puzzles of science and metaphysics is considered the most complex and most achievable. To such conclusions come by reverse engineering the human brain.
Machine learning
Machine learning underlies the strategy of developing “artificial intelligence”; for this, human brain cells are comprehensively studied. This type of training has great potential: its platform includes algorithms, development tools, APIs, and model deployment. Computers have the ability to learn without explicit programming. Innovative companies Amazon, Google and Microsoft are actively using machine learning.
Deep learning platforms
Deep learning is part of machine learning. It is based on how the human brain works, and relies on artificial neural network (ANN) algorithms through which information flows. Robots can “learn” from the input data and the results. Deep learning is a promising trend in artificial intelligence, combined with large amounts of information. It has proven itself in recognizing patterns and classification. Deep Instinct, Fluid AI, MathWorks, Ersatz Labs, Sentient Technologies, Peltarion, and Saffron Technology are examples of companies that are pioneers in this area of intelligence.
Natural Language Processing
Neuro-linguistic programming (NLP) is located on the border between computer and human language and is an artificial intelligence technology. Computer programs can understand human spoken or written language. Amazon Alexa, Apple Siri, Microsoft Cortana, and Google Assistant software use NLP to understand user questions and provide answers to them. This type of programming is widely used in economic transactions and in the field of customer service.
Natural language generation
NLG software is used to convert all types of data into human-readable text, this is achieved through the study of the brain. It is an underrated technology with applications such as business intelligence reporting automation, product descriptions, and financial reports. Thanks to the technology, it becomes possible to create custom content with predictable additional costs. Structured data is converted to text at high speeds, up to several pages per second. Interesting players in this market are Automated Insights, Lucidworks, Attivio, SAS, Narrative Science, Digital Reasoning, Yseop and Cambridge Semantics.
Virtual agents
In terms of artificial intelligence technologies, the terms “virtual agent” and “virtual assistant” are not used interchangeably. Some people try to identify the difference between the concepts, and they do it.
Virtual Assistant is a kind of personal online assistant. Virtual agents are often presented in the form of computer AI characters, conducting an intelligent conversation with users. They can answer questions, and their main advantage is that clients can receive assistance 24 hours a day.
Speech recognition
Speech identification is the ability of a program to understand and analyze words and phrases of a spoken language, as well as convert them into data using the built-in artificial brain algorithm. Speech recognition is used by the company for call routing, voice dialing, voice search and speech-to-text processing. One of the drawbacks is that the program can confuse words due to differences in pronunciation and background noise. Speech recognition software is increasingly being installed on mobile devices. Nuance Communications, OpenText, Verint Systems, and NICE are developing in this area.
Hardware with integrated AI
Devices with built-in AI, chips and graphic processors (GPU) are widely used. Google built artificial intelligence into its hardware, taking as a basis the development of the human brain institute. The impact of AI integration with software goes far beyond consumer applications such as entertainment and games. This is a new type of technology that will be used to promote deep learning. Such developments are carried out by Google, IBM, Intel, Nvidia, Allluviate and Cray.
Decision management
Managing business solutions in innovative products (for example, a robot with artificial intelligence) covers all aspects of the design and regulation of automated systems. It is necessary for organizations to manage the interaction between employees, customers and suppliers.
Decision management improves the alternative choice process, it uses all possible information for better preference, while emphasizing maneuverability, consistency, and accuracy of decision-making. Decision management takes into account time constraints and known risks.
Banking, insurance, and financial services organizations integrate current decision-making software into their customer service processes.
Neuromorphic apparatus
SyNAPSE is a DARPA-funded program for the development of neuromorphic microprocessor systems matching the intelligence and physical parameters of the brain. The platform is looking for the answer to the main question: is it possible to create an artificial brain? First, neural networks are tested in simulations on a supercomputer, then the networks are directly built in hardware. In October 2011, a prototype neuromorphic chip containing 256 neurons was demonstrated. Currently, work is underway to create a multi-chip system capable of emulating 1 million peak neurons and 1 billion synapses.
Neural network modeling
The Blue Brain Project (Blue Brain Project) is an attempt to reconstruct the human brain and spinal cord, to recreate it using computer modeling at the molecular level. The project was founded in May 2005 by Henry Markram at the Lausanne State Polytechnic School (EPFL) in Switzerland. The simulation is performed on the IBM Blue Gene supercomputer, hence the name Blue Brain. As of November 2018, modeling is carried out on mesocytes containing about 10 million neurons and 10 billion synapses. A full-scale imitation of the human brain with its 186 billion neurons is scheduled for 2023.
Spaun, a unified network with semantic index architecture, was created by Chris Eliasmith and his colleagues at the Center for Theoretical Neuroscience (CTN) at the University of Waterloo in Canada. As of December 2018, Spaun is the world's largest brain imitation. The model contains 2.5 million neurons, and this is enough to recognize lists of numbers and perform simple calculations.
SpiNNaker is a massive, low-power, neuromorphic supercomputer that is currently being built at the University of Manchester in the UK. With over a million nuclei and a thousand simulated neurons, the machine will be able to simulate one billion neurons. Instead of implementing one specific algorithm, SpiNNaker will become a platform on which you can test different algorithms. Different types of neural networks can be designed and run on a machine, thus simulating different types of neurons and communication patterns. SpiNNaker is an abbreviation derived from the phrase Spi King Nural.
Brain Corporation is a small research company developing new algorithms and microprocessors that underlie the biological nervous system. The company was founded in 2009 by computing neuroscientist Eugene Izhikevich and neuroscientist / entrepreneur Allen Gruber. Their research focuses on the following areas: visual perception, engine management and autonomous navigation. The company's goal is to equip consumer devices, such as mobile phones and household robots, with an artificial nervous system. The study is partly funded by Qualcomm, which is located on the Qualcomm campus in San Diego, California. No specific products have yet been released or announced, but the company continues to grow and has been actively hiring new employees since February 2018.
Related Research
The Google X Lab is a secret lab where Google is experimenting with future technologies. The projects that the company is working on are not publicly available, but it is believed that they are based on robotics and artificial intelligence. Details of the laboratory first appeared in an article in the New York Times in November 2011. The publication says the lab is located in the Bay Area, California. It is well known that the founders of Google are interested in the study of artificial intelligence and invest in this area. In a 2006 company memo, it said Google wanted to create the world's best artificial intelligence research lab.
Russia 2045, known as the 2045 Initiative or Avatar Project, is an ambitious long-term project whose goals are to create robotic avatars by 2020, conduct brain transplantation by 2025 and create an artificial brain by 2035. The program was launched in 2011 by Russian media tycoon Dmitry Itskov. He seeks to create an institution of the human brain with the help of a global network of scientists who work together for the benefit of humanity, and the planned development of technology. A number of Russian scientists have already received from Itskov investments in their research. In addition, Itskov is trying to get additional funding from wealthy individuals, charitable companies, as well as national and international governments.
The next interesting project is the program of Boston University and Hewlett Packard (HP) called Moneta. The HP team, led by Greg Snyder, is creating a neural network platform called Cog Ex Machina that can run on memristor-based GPUs and computers of the future. The Laboratory of Neuromorphology at Boston University, led by Massimiliano Versace, created the Moneta Modular Artificial Brain, powered by Cog Ex Machina. Acronym stands for Modular Neural Exploring Travel Agent.
Time frame
The question inevitably arises of when they will be able to synthesize a digital copy of the brain and spinal cord.
Unfortunately, this will come soon. Kurzweil’s prediction of brain emulation by 2030 seems overly short, since he is only 12 years old. Moreover, its analogies with the project of the human genome were unsatisfactory. In addition, many scientists are probably moving in some dead end directions.
Similarly, Goertzel’s predictions about the success of the rule-based approach over the next decades seem overly optimistic. Although probably not impossible, given his approach to teaching artificial intelligence.
In a likely scenario, the creation of a code or a likeness of the human brain is possible in 50-75 years. Nevertheless, the date is rather difficult to predict, given the error in neuroscience, on the one hand, and the rate of change dynamics, on the other. The year 2050 is a kind of black hole when it comes to forecasts.