Google's Deepmind AlphaGo victory over traditional Go champions was a stunning exhibition of how far machine learning has come. It's use of unorthodox and persistent winning algorithms showed the advancement of mathematical logic.
Progress is always a two edge sword. The benefits of technology can be used to improve or decrease its advantages to the environment and society. The increase in the scope of available soft and hard power impacts on privacy and destruction. The threat of an AGI killer robot is much more unlikely than the insidious drip drip of central control bleeding into politics and democracy. We should not create things because we can but because it has merit. The jump from AI to AGI is a leap that may never happen. Even if a sort of advanced AI was invented, the product would always be a shadow of human intellect. If the extremely unlikely event happened, it would not be in the form of a robot as the necessary cognitive power could never be encapsulated in such a small vehicle. The power storage requirements would also be a stumbling block to a fully autonomous, go anywhere machine. The possible answer would be some kind of massive distributed cognitive cloud type ‘intelligence’ that could communicate with androids and a dispersed power supply system to boost local mechanicals.
Learning and other algorithmic based solutions have received a lot of media attention which stretches from the helpful to the ridiculous. The development so far has been two dimensional and uses the speed of the chip and unlimited resources to back up an apparent front end of encroachment on human abilities. For those of us who have studied both machine learning and neuroscience it is apparent that humans and machines are very different entities. The brain is a collection of highly adapted cells which are modified to carry out specific tasks and integrated throughout the brain and body. The synthetic approach so far has been to use brute force to overcome the weaknesses of artificial design. The latest developments have sought to use a quasi neuronal type of inherent weighting to simulate the human brain. The question should be as always, is what are we trying to do here? If we are trying to out compete the human brain then we better get ready for a long wait. The idea of Artificial General Intelligence being the ultimate aim begs quite a few questions. There are billions of people on this Earth who already far exceed the abilities of A.I. and the money should be going to creating a better world for the present life forms and not trying to throw another one in the mix. The advantages of technology can be found everywhere and complements humanity and should be used for the good of all. Co-operation is the name of the game and if something gives an unfair advantage to a limited few or further enriches the elite then it is time to think again.
Neurons are cells within the nervous system that transmit information to other nerve cells, muscle, or gland cells. Most neurons have a cell body, an axon, and dendrites. The cell body contains the nucleus and cytoplasm. The axon extends from the cell body and often gives rise to many smaller branches before ending at nerve terminals. Dendrites extend from the neuron cell body and receive messages from other neurons. Synapses are the contact points where one neuron communicates with another. The dendrites are covered with synapses formed by the ends of axons from other neurons. The pyramidal neuron (above), is a type of multipolar neuron found in areas of the brain including the cerebral cortex, the hippocampus, and the amygdala. Pyramidal neurons are the primary excitation units of the mammalian prefrontal cortex and the corticospinal tract. During infancy neurons throughout the brain fight for survival and only approximately still exist a few years later. Numbers are not important in cognition but connections is the name of the game. This pruning of the ones with the least advantage are removed by the glial cells to produce an efficient dense mesh. Dendrites and spikes are constantly produced by neurons and neighbours produce hormones to attract these protrusions. It is estimated that up to 8000-1000 can be present. They can be considered the work horse of cognition.This and an estimated 80 billion neurons in the brain and 2 billion in the gastric system.
Top four AGI contenders to date. All claim to be striving to create it for the advancement of humanity and not personal wealth. We will see if these ideals last if their goal is actually successful.
Based in: San Francisco, California
Mission: Ensure that Artificial General Intelligence benefits all of humanity
Goal: Be the first to create AGI, not for the purpose of domination of profit, but for the safety of society and to be distributed to the world equally.
Founder(s): Elon Musk, Sam Altman and others.
Investors: Microsoft — $1 billion
Recent achievement: Created Open AI Five, a team of 5 AI bots that defeated the world’s Dota 2 champions and a robot hand capable of solving a Rubik’s cube.
Methodology to AGI: Larger Neural-network models
Based in: London
Mission: Research and build safe AI systems that learn how to solve problems and advance scientific discovery for all.
Goal: “Solve intelligence” by fusing novel techniques from machine learning and systems neuroscience to construct powerful general-purpose learning algorithms.
Founder(s): Demis Hassabis
Investors: Founders Fund and Horizons Ventures
Recent achievement: AlphaGo, the AI system that beat the best human player in the game Go in 2016.
Methodology to AGI: Neuroscience — looking deeper into the human brain and drawing inspiration from there to create sophisticated algorithms.
Based in: Mountain View, California
Mission: Make machines intelligent and improve people’s lives
Goal: Create A.I with the intelligence of a human child.
Founder(s): Andrew Ng and Jeff Dean
Investors: Under parent company Google
Recent achievement: Inventing Transformers(a neural-network for Natural Language processing) and TensorFlow
Methodology to AGI: Supervised learning, and other neural-network structures.
Based in: —
Mission: Advancing the state-of-the-art of AI
Goal: Creating Human-like AI
Founder(s): Yann LeCun, deep learning pioneer
Investors: Under parent company Facebook
Recent achievement: Inventing PyTorch, a famous programming language and Mask R-CNN, a computer vision algorithm.
Methodology to AGI: Unsupervised and self-supervised learning.
The ability for synthetic systems to beat humans in certain games and other closed rule based endeavours has fanned publicity that they will take over the world. History has shown that disruption works for awhile and then its advantages become the norm and things settle down. One of the more interesting events has been when humans use fairly modest computers to outperform sophisticated machines. What we offer is direction and novel ideas that the synthetics lack. Of course machines might break away from their rule based straight jacket but this could be a long way off and even then we still have millions of years of evolutionary design behind us. Whether we become augmented cyborgs or human/machine teams this seems to be the most advantageous way to use this brave new technology.
If one assumes that human and AI have different strengths then a co-operative arrangement for complex tasks would seem to be the way forward. The Sampson Array Radar can track hundreds of aircraft at once but you would not want it shooting down the ones it thought was a threat. There is a lot of debate about autonomous AI making life and death decisions, some argue that future threats will be too fast for humans to do this and others that this is too important to be left to AI. With the advances in self-determining machines, ethics is becoming more important than agorithms.
The communication between us and intelligent systems is taking many avenues. Some argue that a direct link using some kind of insertion into the brain will allow instant information and commands to be issued. The problem with this is, although the brain may have areas that are concentrated on certain cognisance, every brain is different and all pathways are distributed and connected. Sticking a rod in the brain will not do it. Other interfaces such as a sensor cap, eye tracking etc. are superficial and do not offer the sophistication required. The natural medium is speech which ironically put us back where we started.
The most obvious impact of AI is the consequence of the automation of tasks across a broad range of industries, transformed from manual to digital. Tasks or roles that include a degree of repetition or the consumption and interpretation of vast amounts of data are now delivered and processed by a computer, sometimes not needing the intervention of humans. AI technologies are constructed by mathematical processes that leverage increasing computing power to deliver faster and more accurate models and forecasts of operational systems, or enhanced representations and combinations of large data sets. However, while these advanced technologies can perform some tasks with higher efficiency and accuracy, human expertise still plays a critical role in designing and utilising AI technology. Human intelligence is what shapes the emergence and adoption of artificial intelligence and innovative solutions associated with it. It is human intelligence that seeks to ask ‘why’ and considers ‘what if’ through critical thinking. As engineering design continues to be challenged by complex problems and quality of data, the need for human oversight, expertise and quality assurance is essential in using AI generated outputs. In the age of AI, understanding the function of work beyond merely sustaining a standard of living is even more important. It becomes a reflection of the fundamental human need for participation, co-creation, contribution, and a sense of being needed; and thus, must not be overlooked. So, in some way, even the ordinary and dull tasks at work become valuable and worthwhile, and if it is removed or has been automated, it should be replaced with something that provides the same for human expression and discovery. With robots, AI, and automation taking some of the mundane and manual tasks out of our hands, professionals have more time to focus in thinking, delivering creative and innovative solutions, and actions that are beyond the reach of AI and are squarely in the domain of human intelligence. This may of course leave us totally on AI to do most tasks and leaving us bored with no purpose. When used with a purpose and not for technology’s sake, AI can unlock tons of opportunities for businesses and improve productivity and participation within the organisation. This, in turn, can result in an increase in demand for products and services and drive an economic growth model that delivers and improves the quality of living. Hopefully.
If the brain is so wonderful, who needs AGI? This is the first question that should be asked and in the wild pursuit of technology is usually avoided.
There are many cited advantages to AGI and some are valid and others do not question the morality or need for the enormous resources required. A classic is the ability to fly autonomous aircraft at your enemy and not risk pilots. The counter argument is why are you fighting in the first place. The problem is usually solvable by other means but human ego and distrust hinder efforts. If a true AGI was invented it would need emotions as well as logic. As the object would have to live in the real world, we could end up with an artificial copy of us. Who says it would be any better. If it only half does the job, what is the point of it? The mooted idea that it could be a wonder being without malice or sin is a myth.
This of course will not stop the mechanistic chasing of results. History has shown that ethics are soon dropped when events change. Many reasons might stop AGI and none of them needs to be human intervention. This kind of massive investment needs a stable environment that can be undone by natural and unnatural cataclysms. Also in spite of the vaunted basic income proposal, people like to work. It is part of our psyche to provide for our families and not everybody agrees that having the opportunity to stay at home and create stuff is the answer. Zoos used to throw food to their animals and wondered why they were not happy and even mentally disturbed. When they hid the food and made the animals look for it or put lions near antelope to keep them both on their toes, things picked up. Life is a constant test and you cannot take that away.
There are many algorithms and designs out there but AGI is so far away at present, they are clutching at straws at present.
Copyright © 2020 deepermind.co.uk | Maldwyn Palmer