ML & AI: Super-Intelligence - Are we afraid?: Part-ten
by
Dr. RGS Asthana
Senior Member IEEE
Figure 1: Computer versus human performance [27]
Summary
ML and AI are developing fast and moving towards Artificial General Intelligence (AGI) from Artificial Narrow Intelligence (ANI) level. This will be fully accomplished by say around 2040. Then it will progress from AGI to Artificial Super Intelligence (ASI) very quickly and then to Singularity really fast.
It will then be more intelligent than the human race and may give option to humans to co-exist. The development of part human and part robot will become possible as two may merge. We briefly discuss Robotics and human race risk of extinction or survival.
Prerequisite
Read article [1] to [15]
Keywords
Machine Learning (ML) Tools, Artificial Intelligence (AI), Neural Networks, Internet of Things (IoT), Data Science (DS), Deep Mind, IBM’s Watson
Prelude
ML/AI is good as it iterates and improves its performance in each iteration, particularly, in playing board games. Think of a scenario when an AI based solution is placed against an AI based solution only or a program plays against a program. In this scenario we find new strategies getting developed which we call ‘super-intelligence.’ This does happen in more or less all scenarios and we do reach super-intelligence.
Elon Musk- a very future-oriented tech entrepreneur - is vocal about his fear of a world take-over by artificial intelligence (AI).
Some of the most popular sci-fi movies — 2001: A Space Odyssey, The Terminator, The Matrix, Transcendence, Ex Machina, Robot, Transformers, and many others — were created assuming AI will progress in time to a level at which humanity will lose control on its own creations, leading to the end of entire civilization.
A few quotes given below illustrate the perils of AI as well as comments of an AI expert saying that these comments may not be correct:
‘If AI power is broadly distributed to the degree that we can link AI power to each individual’s will — you would have your AI agent, everybody would have their AI agent — then if somebody did try to something really terrible, then the collective will of others could overcome that bad actor.’
—
‘I think we should be very careful about artificial intelligence. If I had to guess at what our biggest existential threat is, it’s probably that. So we need to be very careful.’
—
‘I think we need to be very careful in how we adopt artificial intelligence and that we make sure that researchers don’t get carried away,” he stated. “Sometimes what will happen is a scientist will get so engrossed in their work that they don’t really realize the ramifications of what they’re doing.’
—
Elon Musk
‘I think the development of full artificial intelligence could spell the end of the human race.’
—
‘Once humans develop artificial intelligence, it will take off on its own and redesign itself at an ever-increasing rate.’
—
‘A super intelligent AI will be extremely good at accomplishing its goals, and if those goals aren't aligned with ours, we're in trouble.’
—
Stephen Hawking
“I don’t think it’s very helpful for other people who are incredible in their domains commenting on something they actually know very little about, but because they are quite big celebrities now, more than just scientists or businessmen, it gets picked up a lot,”
—
Hassabis, CEO of Deep Mind (bought by Google in 2014)
Hassabis, says further that he has criticized the likes of Microsoft founder Bill Gates and SpaceX and Tesla founder Elon Musk for their comments on AI.
Are we really scared of super-intelligent AI? You should be, says neuroscientist and philosopher Sam Harris [16] -- and not just in some theoretical way. We're going to build superhuman machines, says Harris, but we haven't yet faced the problems associated with creating something that may treat us the way we treat ants. What happens when machines surpass humans in general intelligence? Will artificial agents save or destroy us [17]?
Three key factors which are responsible for development are: processing power, availability of data and development of ML/AI algorithms. Further, processing power is increasing day by day and is available on call and we need to pay by minutes; thanks to cloud technology. Today, people have smart phones which have a very powerful computer. Vast amount of structured as well as unstructured data is also available in form of Big Data. So the missing link seems only ML/AI Algorithms/ programs.
Artificial Intelligence
Think of a state when machines do achieve Insect and Mouse level intelligence followed by Chimp level and then eventually human level and then comes the super-intelligence followed by Singularity level (see figure 1 & 2). Where are we today? Experts do have different view on this but one thing, in general, seems true that we have not yet reached the super-intelligence but in some areas we are almost there.
So let’s discuss some important things. First, do not think of robots. A robot is only an ampule for AI and may represent the human equivalent form — but the AI program may reside in the computer inside the robot. AI is the brain, and the robot is its body—if it can be said that it has a body.
Figure 2: AI and Exponential Growth of Computing [33]
Secondly, you’ve probably heard the term “singularity” or “technological singularity”- it is reached when our technology’s intelligence exceeds our own and normal laws will no longer apply.
AI is, generally, classified in three types [4].
Artificial Narrow Intelligence (ANI) and its application areas
These systems basically were handcrafted and worked basically for the purpose they were designed. This involves specialization in one particular thing i.e. narrow AI (e.g. beating the world champion in chess or Google’s RankBrain [19, 20], and the machine learning algorithms that are currently running on Google like Google Translate, IBM’s Watson ML feature on Amazon that tells you products that are “recommended for you” and self-driving cars). Your smart phone uses lot of ANI e.g. map app, music recommendations say from Pandora, check tomorrow’s weather, talk to digital assistant like SIRI on iPhone, or dozens of other everyday activities, you’re using ANI. Your email spam filter is a classic type of ANI. Amazon product you locate and you get “recommended for you” product on a different site, or Facebook’s friend recommendation. Google Translate is another classic ANI system allowing you to speak a sentence in one language and have the phone spit out the same sentence in another. When your plane lands, the gate selection there is no HITL. Same is true about determining the price of your air ticket. The world’s best Checkers, Chess, Scrabble, Backgammon, and Othello players are now all ANI systems. Google search is another example of ANI system and same is true for Facebook’s Newsfeed. Military, manufacturing, and finance (algorithmic high-frequency AI traders account for more than half of equity shares traded on US markets), and in expert systems like those that help doctors make diagnoses and most famously, IBM’s Watson [14] are examples of ANI.
From ANI to Artificial General Intelligence (AGI)
It refers to a computer that is as smart as a human across the board — If AI can perform like a human, we consider it AGI. If an AI is going to be as intelligent as the human brain, one crucial thing has to happen— the AI [21] “needs to equal the brain’s raw computing capacity. One way to express this capacity is in the total calculations per second the brain could manage.”
Think of a system which learns to drive by watching the human driver? It is difficult to believe but is true [22]. The result seems to match the responses you’d expect from a human driver. But what if one day it did something unexpected—gone down looking at a tree or crashed, or did not move at a green light or hit a pedestrian crossing road at zebra line? This is a burning issue with an AI system it does right thing but there is no explanation why and how it arrives at its decision. But this should not happen—unless we find ways of making techniques like deep learning more plausible to their creators and answerable to their users. Otherwise it will be hard to predict when failures might occur—and it’s unavoidable they will.
The term ‘Deep learning’ is derived from “deep” neural nets which are built by layering many networks on top of each other [7]. Due to the increasing power and falling price of computer servers and advent of cloud computing, machines with enough processing power are now available as well as are capable to run these networks. Now you don’t need to own infra but democratization of data and only pay for actual use by minute as server- less environment is becoming common way of processing data today.
Complex ML methods, that could fully automate the decision-making process, are being used for investments or financial decisions, or the military decisions, or medical decisions, although this process is altogether unclear or hidden from the people. This is the cause of concern many people have expressed about AI and ML [1].
How we can get AGI soon that we know of. The way is obvious from the following quote [21]:
“We’d build a computer whose major skills would include doing research on AI and coding changes into itself — allowing it to not only learn but to improve its own architecture. We’d teach computers to be computer scientists so they could bootstrap their own development.” This is equivalent to fielding AI against AI to get better performance than human.
There are two areas where we may already have got to AGI level, viz., military and healthcare. As, US in Iraq and Afghanistan war has used stealth aircrafts and drones which had human–in–the-loop (HITL) capability only for ‘kill’ command but even this was not necessary. Computer can be very good at analyzing images even better than human being, particularly, medical images even today e.g. tumor from images. A doctor is used for finally selecting the images but it is also not necessary. HITL is required till neural net weights are set i.e. training is complete but becomes voluntary after that.
As per Donald Knuth, “AI has by now succeeded in doing essentially everything that requires ‘thinking’ but has failed to do most of what people and animals do ‘without thinking.’
When you want to reach a thing, the muscles, tendons, and bones in your shoulder, elbow, and wrist perform a long series of operations, in co-ordination with your eyes, to let you move your hand in a straight line through three dimensions in an easy way as you have perfected this process or software (if we can call it) in your brain. It also recognizes the object and puts pressure according to softness of the object. However, when a computer or Robot wishes to do the same thing it needs to do this the robot not only needs to solve complicated mathematical equations but also recognize the object and then decide on the grip to pick it.
It is possible that we may achieve true AGI level around 2040.
AGI to Artificial Super Intelligence (ASI)
How soon AI will reach human-level general intelligence? As per a survey of hundreds of scientists about when they believed we’d be more likely than not to have reached AGI was 2040 and from AGI to ASI happens really quickly.
In our world, smart means a 130 IQ and stupid means an 85 IQ—but we don’t know what to call say an IQ of 13,000.
We know with intelligence comes power. An ASI will be the most powerful form of life on Earth, and all living things, including humans may become subordinate to it. If our small and slow brains were able to invent Wi-Fi, then something with IQ of 1,000 or 1 billion should have no problem controlling the positioning of each and every atom in the world, at any time [33]. Also possible is end of all life on Earth.
Oxford Prof. Nick Bostrom [23] defines super-intelligence as “an intellect that is much smarter than the best human brains in practically every field, including scientific creativity, general wisdom and social skills.
ASI ranges from a computer that’s just a little smarter than a human to one that’s trillions of times smarter—across the board. ASI is the key reason which makes AI an interesting subject and why the words “immortality” and “extinction” are required when ASI is under discussion.
It’s very likely that ASI or AI [24] may achieve a level of intelligence which makes AI smarter than all of humanity combined — it will be something entirely different than intelligent machines we are comfortable with. The goals to ASI systems are given by its creators —e.g., your GPS’s goal is to give you the most efficient driving directions and Watson Computer’s aim [25] is to understand questions and then answer them accurately using its NLP interface – which is capable of answering questions posed in natural language. The main aim of an ASI system is to fulfill its assigned goals.
As of now, humans have conquered the lowest caliber of AI—ANI—in many ways, and it’s everywhere. The AI revolution is the road from ANI, through AGI, to ASI—a road we may or may not be extinct before ASI comes. The human brain does have a size restriction as it has to reside in a certain cavity or a given space, the neurons can communicate at 100 meters per second and can fire 200 times a second but computers do not have any such restrictions and do operate at gigahertz range and signals move at speed of light. The internet connectivity will increase with time and will have billions [15] of devices and systems connected.
A key distinction is the difference between speed super-intelligence and quality super-intelligence. Think of a machine that thinks like a human, except a million times quicker, which means it could figure out in five minutes what would take a human a decade. What makes humans intellectually capable than chimps isn’t a difference in thinking speed—it’s that human brains contain a number of sophisticated cognitive modules that enable things like complex linguistic representations or long term planning or abstract reasoning, that chimps’ brains do not. If we speed up a chimp’s brain by thousands of times wouldn’t bring him to our level—even with a decade’s time, he wouldn’t be able to figure out how to use a set of custom tools to assemble an intricate model, something a human could knock out in a few hours.
Singularity
The technological singularity (simply, referred to as the singularity) is the premise [26] that the invention of ASI will cause huge growth, initiating immeasurable ups and downs to human civilization. According to this premise, an upgradable intelligent agent such as a computer running software-based AGI would enter a "runaway reaction" of self-improvement cycles, with each new and more intelligent generation appearing more and more rapidly, causing an intelligence explosion and resulting in a powerful super-intelligence that would, qualitatively, far surpass all human intelligence.
Superintelligence: Paths, Dangers, Strategies is a 2014 book [23, 32] by the Swedish philosopher Nick Bostrom from the University of Oxford. It argues that if machine brains surpass human brains in general intelligence, then this new super-intelligence could replace humans as the dominant life form on Earth. Sufficiently intelligent machines could improve their own capabilities faster than human computer scientists, and the outcome could be an existential catastrophe for humans.
The ML/AI Singularity will not happen till 2040 advances before that will make fundamental changes that will allow us to live for a long time. The first one to come will be biotech [30]. Then Nano-tech singularity [31] and finally, we may achieve singularity due development in ML/AI tech [26] around 2060+ and figure 3 shows the status of human race thereafter the worst and better possible scenarios.
Figure 3: What Happens to human after ASI [34]?
The following quote seems very fascinating to me:
“The merging of mind and machine has already begun with cochlear and retinal implants, coming neural implants, and gene editing [41].”
Knowledge is only one byproduct of numerous feedback loops that run the five physical senses that make us human.
Buddhists also recognize a sixth sense which is the subjective experience of the mind. Chemical processes in the mind/body create feelings in the body, and so on. ML/AI tools that can enhance consciousness, particularly, when self-awareness surpasses biology (human beings) we reach ASI and then Singularity. Consciousness like the Universe expands also evolves. The following quote is interesting:
“The greater shock may be that Singularity has already been occurring throughout history without the aid of technology. In fact, enlightenment is true Singularity [41].”
Your brilliant future here, now, is Singularity as holistic self-awareness in the now moment that you are mind, body, and spirit capable of miracles. Death once conquered, immortality becomes your new reality.
Robotics Advancement
An Android is a humanoid robot or synthetic organism designed to look and act like a human, especially one with a body having a flesh-like appearance. Although initially, Androids was a science fiction but recent advancements in robotics technology made it feasible to develop functional and realistic humanoid robots [37]. A cyborg is a being with both organic and bio-mechatronic body parts [38]. The Borg characters (from Star Trek or Star Wars movies) may be depicted almost similar to humans. Cyborgs (robots) in military forms are depicted with built-in weapons, among other thing.
Shadow Robot Co. in England [39] assembled the bionic man from prosthetic body parts and artificial organs donated by laboratories around the world. We quote Mr. Walker based on what he told live-science below [39]:
"Our job was to take the delivery of a large collection of body parts — organs, limbs, eyes, heads — and over a frantic six weeks, turn those parts into a bionic man. You put a prosthetic part on a human who is missing that part, we had no human; we built a human for the prosthetic parts to occupy."
A Fembot is a humanoid robot that is gendered feminine. Fembots appear in science fiction film and art. With advancements in robotics, realistic humanoid are also emerging in real-life robot design.
If in the future, where man and machine merge (see figure 4) then how do we classify robots created with human or organic parts as still human or call them a whole new class [36]. Both are part human, part machine? They just have different starting points. I guess an argument can be made as to whether or not a "brain" can be transferred as of now? '
Figure 4: A modified Robot with merging of human and Robot
Credit: collective-evolution.com
What happens when computers become more intelligent than human?
“How long it may take until the first machine reaches super-intelligence [27]? Not amazingly, views vary, people have almost diagonally opposite view and this is a topic of intense debate among scientists and thinkers (see figure 1).
A survey result for achieving AGI level is as follows:
Median optimistic year (10% likelihood) → 2022
Median realistic year (50% likelihood) → 2040
Median pessimistic year (90% likelihood) → 2075
Median realistic year (50% likelihood) → 2040
Median pessimistic year (90% likelihood) → 2075
AGI to ASI transition may happen in less than twenty years or even earlier depending on how much autonomy computers may have (see figure 5)?
Figure 5: ANI to AGI and AGI to ASI: expert predictions [34]
ASI to Singularity transition may happen really very fast as human may lose control over development. This may be very optimistic scenario but level of autonomy may be much higher even at AGI level.
Facebook recently [28] shut down AI bots based app where bots started ignoring the set programmed instructions and began speaking in a language unintelligible to humans instead of English which was supposed to be set method for communication. It is feared that Facebook was able to close the App but once AGI/ASI is achieved human may not be able to close the app as Ai will know how to manage the app and even the Internet connectivity.
Mechanical structure including stiffness of humanoid robot is an important task in the development of humanoid robot [29].
Way forward
Super-intelligence may mean end of supremacy of human race as you will have machines having more intelligence than human. Here is the cause of fear - will machines take over control and will start ruling the human race. Human may find it difficult to shut down machines. Are we ready for such a scenario? The answer to this question for some may be yes but majority will say no.
Figure 6: Immortality v/s Extinction
The advent of ASI will, for the first time, open up the possibility for a species to land on the immortality side [34] of the balance beam. However, an unimaginably dramatic impact [see figure 6] that it’s likely to knock the human race off the beam, in one direction or the other. Super-intelligence may be needed to solve difficult problems, particularly, in healthcare areas. The key problems being how to prevent aging and solve old age issues like Alzheimer and Dementia. It is hoped that super-intelligent machines may think differently and find some workable solution. But can we develop AGI for the limited purpose. Secondly, can we control it? Are we ready with political, social and ethical issues? The answer at present is probably/definitely no. If we develop say a AGI technology for welfare of human race in Silicon Valley, are we ready to share technology with china and North Korea? The answer is yet no.
Think of a scenario when human race become part human and part robotic. Even today, we have artificial lungs, kidneys, heart and many prosthetics and with time these will become better and move towards perfection. With wish to become immortal when one is sure to die in near future, one may be willing to even upload their brains in a machine and then machine shows human like behavior. This becomes true only when it becomes possible to upload brain in a machine. This looks for fetched now but not impossible.
Musk expressed at the World Government Summit in Dubai [35] that, "Over time I think we will probably see a closer merger of biological intelligence and digital intelligence. It's mostly about the bandwidth, the speed of the connection between your brain and the digital version of yourself. Some high bandwidth interface to the brain will be something that helps achieve a symbiosis between human and machine intelligence and this may solve the control problem and the usefulness problem."
References
[1] Progress and Perils of Artificial Intelligence (AI)
[2] Invited Chapter 6 - Evolutionary Algorithms and Neural Networks, Pages 111-136, R.G.S. Asthana in book, Soft Computing and Intelligent Systems (Theory and Applications), Academic Press Series in Engineering, Edited by:Naresh K. Sinha, Madan M. Gupta and Lotfi A. Zadeh ISBN: 978-0-12-646490-0
[3] Future 2030 by Dr. RGS Asthana, Senior Member IEEE
[4] Machine Learning (ML) and Artificial Intelligence (AI) – Part 1, by Dr. RGS Asthana, Senior Member IEEE
[5] Machine Learning (ML) and Artificial Intelligence (AI) – Part Two, by Dr. RGS Asthana, Senior Member IEEE
[6] Machine Learning (ML) and Artificial Intelligence (AI): Cognitive Services and Robotics – Part Three by Dr. RGS Asthana, Senior Member IEEE
[7] Machine Learning (ML) and Artificial Intelligence (AI): Big Data and 3 D Printing – Part four by Dr. RGS Asthana, Senior Member, IEEE.
[8] Machine Learning (ML) and Artificial Intelligence (AI): Drones and Self-driving Cars– Part Five by, Dr. RGS Asthana, Senior Member IEEE
[9] Machine Learning (ML) and Artificial Intelligence (AI): Healthcare– Part Six by, Dr. RGS Asthana, Senior Member IEEE
[10] Machine Learning (ML) and Artificial Intelligence (AI): Will AI/ML intelligence surpass humans? Part Seven by Dr. RGS Asthana, Senior Member IEEE
[11] Machine Learning (ML) and Artificial Intelligence (AI): Impact of AI/ML in Healthcare: Part-Eight by Dr. RGS Asthana, Senior Member IEEE
[12] Machine Learning (ML) and Artificial Intelligence (AI): Big data & Data Science (DS) and their importance: Part-Nine by Dr. RGS Asthana, Senior Member IEEE
[13] Deep mind website
[14] IBM Watson Website
[15] Internet of Things (IoT)
[16] Can we build AI without losing control over it
ISBN-13: 978-0199678112
[18] The AI Revolution: The Road to Super-intelligence
[19] Artificial intelligence is changing SEO faster than you think
[20] FAQ: All about the Google RankBrain algorithm
[21] AI Revolution
[22] The Dark Secret at the Heart of AI
[23] Professor Nick Bostrom, Director, Oxford Future of Humanity Institute; Director, Oxford Martin Programme on the Impacts of Future Technology; Professor of Applied Ethics
[24] AI Revolution
[25] Watson computer
[26] Technological singularity
[27] When Will The First Machine Become Super-intelligent?
[28] Facebook researchers shut down AI bots that started speaking in a language unintelligible to humans
[29] Mechanical Structure of Humanoid Robot with Human Thinking Behavior Application
[30] Singularity Symposium: Biotech Revolution
[31] Nano Technology Now: Singularity
[32] Superintelligence: Paths, Dangers, Strategies
[33] The AI Revolution: The Road to Super-intelligence
[34] The AI Revolution: Our Immortality or Extinction
[35] Elon Musk: Humans must merge with machines or become irrelevant in AI age
[36] A human with robotic parts is still human, but a robot with human parts, is it human/living?
[37] Android (robot)
[38] Cyborg
[39] 1st Fully Bionic Man Walks, Talks and Breathes
[40] Fembot
[41] Humanity 2.0: The unstoppability of Singularity
No comments:
Post a Comment