Artificial Intelligence, The Future

Overview of Artificial Intelligence (AI)

• Definition of AI is a moving target
• Winston, 1977:  "AI is the study of ideas which enable computers to do the things that make people seem intelligent"
• Winston, 2000:  "AI is the study of the computations that make it possible to perceive, reason, and act"
• Distinguishing the brain from the mind
  • the brain is like hardware
  • the mind is like software
  • Thus the brain is the hardware that the mind "runs" on
• Two approaches to AI summarized by:
• Computer structured to imitate the structures of the brain. Software simulates human mental processes.
• Computers exhibit intelligent behavior (external) regardless of their internal struture or processes.
• Turing test for machine intelligence:
  • Human converses via keyboard with remote human or machine
  • Human determines whether he/she is conversing with a machine or with another human
  • If after a period the person cannot determine whether the other is human or machine and it turns out to be a machine, the machine is considered intelligent.
  • Turing test is based on machine behavior, not on its structure
• Searle's Chinese Room argument:
  • English Person receives question written in Chinese
  • Person looks up pattern of question's characters in reference text, which yields characters representing answer
  • Person issues answer in Chinese.
  • Person does not know a word of Chinese
  • Is this system intelligent?
• Computers are structured very differently than our brains are structured
  • The brain consists of billions of slow and simple processors called neurons (introduction to neurons)
  • Computers consist of one or a small number of fast and complex processors called CPUs
  • Neurons are highly interconnected and work in parallel.
• Computers and humans have vastly different capabilities
  • computers do some things easily that are difficult for humans: complicated math, remembering lots of detailed data
  • computers have difficulty doing things that people do easily: visual recognition, speaking, understanding speech
• Some practical applications of AI
• expert systems: knowledge base contains facts and rules to draw inferences in deep but narrow knowledge realm. Originally used for medical diagnosis and searching for mineral deposits. Many decision-making applications. Rules ("if...then...") are used to draw inferences based on the current set of facts.
• neural networks: imitate structure of brain. Through training, patterns in the brain's network are formed. Helps machine recognize similar patterns Common applications include speech recognition and generation, and robotics.
• The CYC (pronounced "psych") project
  • CYC attempts to make expert systems and knowledge bases more widely usable
  • Computers have knowledge that is very deep but narrow (see "Savant Syndrome")
  • CYC provides rules and facts about common sense knowledge
  • Example: if you are in the classroom, so is your head.
  • Example: after a person dies, (s)he remains dead
  • See www.cyc.com
  • CYC knowledge base can be visualized as a pyramid. See www.cyc.com/cyc/technology/whatiscyc_dir/maptest
• The DARPA Grand Challenge and Urban Challenge
• DARPA is the same DoD agency that funded original development of the Internet
• Competition for autonomous vehicles
• Quote from DARPA Urban Challenge web page: "An autonomous ground vehicle is a vehicle that navigates and drives entirely on its own with no human driver and no remote control. Through the use of various sensors and positioning systems, the vehicle determines all the characteristics of its environment required to enable it to carry out the task it has been assigned."
• 2007 Urban Challenge, run on an adandoned Air Force base in California, offered $2,000,000 first prize which was won by a vehicle from Carnegie Mellon University
• The Sixth Sense project
  • MIT Media Lab (same folks who developed Scratch)
  • A wearable device to enhance interactions between the real world and the cyber world
  • Includes visual sensing, projection, and communication system
  • Pattie Maes, MIT professor, directs the Fluid Interfaces Group
  • Pranav Mistry, MIT grad student, spearheads the Sixth Sense project at the Fluid Interfaces Group
  • View this YouTube video of their show-stopping presentation at the 2009 TED conference (also see www.ted.com)

Machine Intelligence in the Future

• Built an operable computer as a teenager in the 1960s
• Has developed assistive technologies such as an early reading machine and voice recognition software
• Wrote the book The Age of Spiritual Machines
• Developed the theory of singularity : ever-accelerating rates of technological growth will lead to machine intelligence exceeding human intelligence by 2030, "the singularity".
• We cannot predict what will happen beyond 2030 because it will be beyond our comprehension

• Kurzweil argues, YES, apparent consciousness can be achieved by mastering human emotion and passing the Turing test
• Kurzweil also argues that consciousness is a synonym for subjectivity and having subjective experience (from a debate on machine consciousness)
• Computer scientist David Gelernter says MAYBE but we can't get there through software. Consciousness requires the "presence of mental states that are strictly private, with no visible functions or consequences." (same debate)
• Gelernter also says that a person can mimic the execution of a computer program because it knows how a computer works. But a computer cannot in the same sense mimic the operations of the brain because we do not know how the brain works.
• If you download your brain to a computer chip, will it be conscious?

• Cellular scale sensors and processors can be embedded in our bodies
• One such design is for an artificial red blood cell that contains oxygen. Such cells could be injected into a heart attack victim to quickly get oxygen to the brain, for instance.
• Nanorobots programmed to sense and destroy bacteria and other pathogens could be injected into the bloodstream
• Surgical nanorobots can be inserted into the body to take pictures and transmit them to surgeons, to make tiny surgical cuts of diseased tissue, and perform other surgical functions
• Is there a difference between using devices such as the artificial blood cells to heal a diseased body and using them to enhance a healthy one?
• Do we become superhuman, or Frankenstein's monster?

• Suppose "computer scientists succeed in developing intelligent machines that can do all things better than human beings can do them"
• "Either of two cases might occur. The machines might be permitted to make all of their own decisions without human oversight, or else human control over the machines might be retained."
• In the first case, "it is impossible to guess how such machines might behave."
• The "human race might easily permit itself to drift into a position of such dependence on the machines that it would have no practical choice but to accept all of the machines' decisions."
• "As society and the problems that face it become more and more complex and machines become more and more intelligent, people will let machines make more of their decisions for them, simply because machine-made decisions will bring better results than man-made ones."
• "People won't be able to just turn the machines off, because they will be so dependent on them that turning them off would amount to suicide."

In his essay "Why the Future Doesn't Need Us", Bill Joy, the founder of Sun Microsystems, writes

• "Accustomed to living with almost routine scientific breakthroughs, we have yet to come to terms with the fact that the most compelling 21st-century technologies - robotics, genetic engineering, and nanotechnology - pose a different threat than the technologies that have come before."
• "Each of these technologies also offers untold promise: The vision of near immortality that Kurzweil sees in his robot dreams drives us forward; genetic engineering may soon provide treatments, if not outright cures, for most diseases; and nanotechnology and nanomedicine can address yet more ills. Together they could significantly extend our average life span and improve the quality of our lives. Yet, with each of these technologies, a sequence of small, individually sensible advances leads to an accumulation of great power and, concomitantly, great danger." (italics mine)
• "The 21st-century technologies...are so powerful that they can spawn whole new classes of accidents and abuses. Most dangerously, for the first time, these accidents and abuses are widely within the reach of individuals or small groups. They will not require large facilities or rare raw materials. Knowledge alone will enable the use of them."
• "I think it is no exaggeration to say we are on the cusp of the further perfection of extreme evil, an evil whose possibility spreads well beyond that which weapons of mass destruction bequeathed to the nation-states, on to a surprising and terrible empowerment of extreme individuals."

Interestingly, there is a common thread among five of the persons identified in this section: Bill Joy was originally inspired to write this Wired magazine article by a barroom conversation with Ray Kurzweil (Singularity) and John Searle (Chinese Room). Kurzweil was about to publish his seminal book The Age of Spiritual Machines, which contained the above passages from Theodore Kaczynski. Joy repeats them in his article. David Gelernter was one of Kaczynski's bombing victims. Gelernter recovered (although not completely) and is still professor of computer science at Yale.