View
2.716
Download
1
Category
Preview:
DESCRIPTION
A lecture given at the second LAST festival (www.lastfestival.com) by Piero Scaruffi on Artificial intelligence and the Singularity - History, Trends and Reality Check
Citation preview
Artificial Intelligence
and the Singularity
piero scaruffi
www.scaruffi.com
October 2014
"The person who says it cannot be done should not
interrupt the person doing it" (Chinese proverb)
Table of Contents
• The future of mind
– Knowledge-based systems
– Connectionist systems
– ...and their cultural background
• The future of body
– Robots
– Implants
– Synthetic Biology
2
• The Singularity?
– Reality Check
– Accelerating Progress?
– Non-human Intelligence
– Human Intelligence
– A Critique of the Turing Test
4
A Brief History of Logic
George Boole's "The Laws Of Thought" (1854): the
laws of logic “are” the laws of thought
Propositional logic and predicate logic: true/false!
5
A Brief History of Logic
Axiomatization of Thought:
Gottlob Frege's "Foundations of Arithmetic"
(1884)
Giuseppe Peano's "Arithmetices Principia
Nova Methodo Exposita" (1889)
Bertrand Russell's "Principia Mathematica"
(1903)
A Brief History of Logic
• David Hilbert (1928)
– Entscheidungsproblem problem: the
mechanical procedure for proving
mathematical theorems
– An algorithm, not a formula
– Mathematics = blind manipulation of
symbols
– Formal system = a set of axioms and a
set of inference rules
1912 –1954
Alan Turing
8
The Cultural Context
• 1910-1950 Everything changed:
– Everyday life
– The foundations of science
– The concept of art
– The geopolitical order
9
• Electricity
• Regriferator
• Automobile
• Airlane
• Telegraph
• Telephone
• Phonograph
• Camera
• Cinema
• Radio
• Typewriter
• Calculator
• Skyscraper
• Plastic
The 1910s
10
You are a
formula Everything
is relative
You are and
you are not
You are just
a reflex You are a
probability
Everything is
uncertain
Everything is
moving away
from you
11
The emancipation of the dissonance
History is a
nightmare from
which I am trying
to awake.
12
There will always
be something you
cannot prove
Your mind
creates reality
Truth is an
opinion
Life and machines
obey the same
laws of nature
Everything is
information
Everything
comes from just
one point
Mind is a
symbol
processor
13
Cultural Context
• Bottom line:
– Nonconformism
– Anxiety
– Noise
– Freedom
14
Cultural Context
• World War II (1939-45)
• The Holocaust
• Hiroshima
• Disintegration of the British Empire
• Rise of the USA and Soviet Union
15
Alan Turing
• Hilbert’s challenge (1928): an algorithm
capable of solving all the mathematical
problems
• Turing Machine (1936): a machine whose
behavior is determined by a sequence of
symbols and whose behavior determines the
sequence of symbols
• A universal Turing machine (UTM) is a
Turing machine that can simulate an arbitrary
Turing machine
Alan Turing
• Alan Turing (1936)
– Universal Turing Machine: a Turing
machine that is able to simulate any other
Turing machine
– The universal Turing machine reads the
description of the specific Turing
machine to be simulated
Turing Machine
17
Alan Turing
(BTW, the halting problem is undecidable, i.e. Hilbert’s
Entscheidungsproblem is impossible)
Alan Turing
• Turing machines in nature: the ribosome,
which translates RNA into proteins – Genetic alphabet: nucleotides ("bases"): A, C, G, U
– The bases are combined in groups of 3 to form "codons“
– RNA is composed of a string of nucleotides ("bases") according to
certain rules
– There are special carrier molecules ("tRNA") that are attached to
specific aminoacids (proteins)
– The start codon encodes the aminoacid Methionine
– A codon is matched with a specific tRNA
– The new aminoacid is attached to the protein
– The tape then advances 3 bases to the next codon, and the process
repeats
– The protein keeps growing
– When the “stop” codon is encountered, the ribosome dissociates
from the mRNA
19
Alan Turing
• World War II:
– Breaking the Enigma code (Bombe)
– Turing worked at Bletchley Park where the
Colossus was built but it was not a universal
Turing machine (not general purpose)
Replica of the Bombe
20
The Turing Century
• Can you name any achievement of the last
50 years (from the Moon landing to animal
cloning) that would have happened even
without programmable computers?
21 21
Electronic Brains
• 1941: Konrad Zuse's Z3 programmable
electromechanical computer, the first Turing-
complete machine
• 1943: Tommy Flowers and others build the
Colossus, the world's first programmable digital
electronic computer
22 22
Electronic Brains
• 1944: Howard Aiken of IBM unveils the first computer programmed by punched paper tape, the electromechanical Harvard Mark I
• 1945: John Von Neumann designs a computer that holds its own instructions, the "stored-program architecture"
23
Electronic Brains
1945: John Von Neumann's computer architecture
Control unit:
•reads an instruction from memory
•interprets/executes the instruction
•signals the other components what to do
•Separation of instructions and data (although
both are sequences of 0s and 1s)
•Sequential processing
24 24
Electronic Brains
1946: The first non-military computer, ENIAC, or
"Electronic Numerical Integrator and Computer",
is unveiled, built by John Mauchly and Presper
Eckert at the University of Pennsylvania
25 25
Electronic Brains
Computation
26 26
Electronic Brains
• Apr 1949: The Manchester Mark 1, the first
stored-program electronic computer
• May 1949: Cambridge's EDSAC, the second
stored-program electronic computer
• Aug 1949: Philadelphia's EDVAC, the third
stored-program electronic computer
• 1950: The Pilot ACE computer
27 27
Electronic Brains
• May 1950: The first stored-program electronic computer to be deployed in the USA, the SEAC, and the first to use semiconductors instead of vacuum tubes
• Feb 1951: The Ferranti Mark 1, the first commercial computer, an evolution of the EDSAC
• 1952: A Univac 1 correctly predicts that Eisenhower would win the elections
28 28
Electronic Brains
(Computer History Museum, Mountain View)
• Goldstine and Eckert with the electronics
needed to store a single decimal digit
29 29
Electronic Brains
Computer programmers of 1951: Patsy Simmers
(holding an ENIAC board) Gail Taylor (holding an
EDVAC board), Milly Beck (holding an ORDVAC
board), Norma Stec (holding a BRLESC-I board)
30 30
Electronic Brains
1954: IBM's first “mass-produced” computer, the 650
(1,800 units sold - $200-400,000 each)
31 31
Electronic Brains
• USA/ Semiconductors
– 1947: AT&T's Bell Labs invent the transistor (William Shockley, John Bardeen, Walter Brattain)
– 1949: The USA files an antitrust lawsuit against AT&T
– 1952: AT&T's symposium on the transistor, open to everybody
– 1954: Texas Instruments introduces the first commercial transistor
– 1954: The first transistor radio (“Regency”)
32 32
Electronic Brains
• USA/ Semiconductors
– 1961: Texas Instruments introduces the first commercial integrated circuit
– Military and space applications use the integrated circuit
– 1965: Gordon Moore predicts that the processing power of computers will double every 18 months
– 1971: Intel invents the microprocessor
– Universities are irrelevant in semiconductor progress because the manufacturing process is too costly
– Universities are crucial for progress in computers
Jack Kilby’s I.C.
Intel 4004
33
Electronic Brains
The future of your brain is coming faster
than your brain can think…
34 34
Electronic Brains
Software
• 1958: Jim Backus (at IBM) invents the FORTRAN
programming language, the first machine-
independent language
• 1964: IBM introduces the first "operating system" for
computers (the OS/360)
• 1968: The Arpanet is established based on Baran’s
idea (four nodes: UCLA, Stanford Research
Institute, UCSB, University of Utah)
• 1969: the Unix operating system is born
35 35
Electronic Brains
Democratizing technology
• Antitrust policies contribute to the rapid diffusion
of intellectual property throughout the computer
and semiconductor industries
• 1956: IBM and AT&T settle antitrust suits by
licensing their technologies to competitors
• 1969: The “unbundling” of software by IBM
creates the software industry
Cybernetics
The Steam Engine
• Biggest impact on daily life since the printing press
• Inventors are ordinary people, not academics
• The automation of manufacturing begins in
Lancashire, not at a university
James Watt (1776)
Cybernetics
Norbert Wiener (1947)
• Bridge between machines and nature,
between "artificial" systems and natural
systems
• Feedback, by sending back the output as
input, helps control the proper functioning of
the machine
• A control system is realized by a loop of
action and feedback
• A control system is capable of achieving a
"goal", is capable of "purposeful" behavior
• Living organisms are control systems
38
The Turing Test
1950: Alan Turing's "Computing Machinery and Intelligence"
(the "Turing Test")
Can machines think?
39
The Turing Test
The Turing Test (1950)
• Hide a human in a room and a machine in another
room and type them questions: if you cannot find
out which one is which based on their answers,
then the machine is intelligent
40
The Turing Test
The “Turing point”: a computer can be said to be intelligent if its
answers are indistinguishable from the answers of a human
being
? ?
41
Artificial Intelligence
1954: Demonstration of a machine-
translation system by Leon Dostert's team
at Georgetown University and Cuthbert
Hurd's team at IBM
1956: Dartmouth conference on Artificial
Intelligence
Artificial Intelligence (1956): the discipline of
building machines that are as intelligent
as humans
42
Artificial Intelligence
1956: Allen Newell and Herbert Simon
demonstrate the "Logic Theorist“, the first
A.I. program, that uses “heuristics” (rules of
thumb) and proves 38 of the 52 theorems
in Whitehead’s and Russell’s “Principia
Mathematica”
1957: “General Problem Solver” (1957): a
generalization of the Logic Theorist but
now a model of human cognition
43
Artificial Intelligence
1957: Noam Chomsky's "Syntactic Structures"
S stands for Sentence, NP for Noun Phrase, VP for Verb Phrase, Det for Determiner,
Aux for Auxiliary (verb), N for Noun, and V for Verb stem
44
Artificial Intelligence
1957: Frank Rosenblatt's Perceptron, the first artificial neural network
45
Artificial Intelligence
1959: John McCarthy's "Programs with
Common Sense" focuses on knowledge
representation
1959: Arthur Samuel's Checkers, the world's
first self-learning program
1960: Hilary Putnam's Computational
Functionalism
1962: Joseph Engelberger deploys the
industrial robot Unimate at General Motors
46
Artificial Intelligence
1963 Irving John Good speculates about "ultraintelligent machines" (the "singularity")
1964: IBM's "Shoebox" for speech recognition
1965: Ed Feigenbaum's Dendral expert system: domain-specific knowledge
47
Artificial Intelligence
1966: Ross Quillian's semantic networks
48
Artificial Intelligence
1966: Joe Weizenbaum's Eliza
1967: Zuse suggests that the universe is a computation
1968: Peter Toma founds Systran to commercialize machine-translation systems
49
Artificial Intelligence
1969: Marvin Minsky & Samuel Papert's
"Perceptrons" kill neural networks
1969: Stanford Research Institute's Shakey the
Robot
1972: Bruce Buchanan's MYCIN
•a knowledge base
•a patient database
•a consultation/explanation program
•a knowledge acquisition program
Knowledge is organised as a series of IF THEN rules
50
Artificial Intelligence
1972: Terry Winograd's Shrdlu
51
Artificial Intelligence
1972: Hubert Dreyfus's "What Computers Can't Do"
1974: Marvin Minsky's Frame (see chapter on “Cognition”)
1975: Roger Schank's Script (see chapter on “Cognition”)
1975: John Holland's Genetic Algorithms
1976: Doug Lenat's AM
1979: Cordell Green's system for automatic programming
1979: Drew McDermott's non-monotonic logic
1979: David Marr's theory of vision
52
Artificial Intelligence
1980: John Searle’s "Chinese Room"
1980: Intellicorp, the first major start-up for
Artificial Intelligence
1982: Japan's Fifth Generation Computer
Systems project
1980s: Second A.I. bubble
53
Artificial Intelligence
1982: John Hopfield’s simulation of annealing
1983: Geoffrey Hinton's and Terry Sejnowski's Boltzmann
machine for unsupervised learning
1985: Judea Pearl's "Bayesian Networks"
1986: Paul Smolensky's Restricted Boltzmann machine
2006: Geoffrey Hinton's Deep Belief Networks
Rummelhart network
Neurons arranged in layers, each neuron
linked to neurons of the neighboring
layers, but no links within the same layer
Requires training with supervision
Hopfield networks
Multidirectional data flow
Total integration between input and output
data
All neurons are linked between themselves
Trained with or without supervision
54
Artificial Intelligence
Genealogy of Intelligent Machines
Hydraulic machines
Steam engines
Cybernetics
Neural networks
Logic
Hilbert
Turing Machine
Computers
Expert Systems
55
Artificial Intelligence 1997: IBM's "Deep Blue" chess machine beats the world's
chess champion, Garry Kasparov
2011: IBM's Watson debuts on a tv show
2014: Vladimir Veselov's and Eugene Demchenko's program Eugene Goostman passes the Turing test
56
Information-based System
Data
Base
Who is the
president of
the USA?
Where is
Rome?
OBAMA
ITALY
57
Knowledge-based System
Know
ledge
Base
Who will the
president of
the USA?
Where is
Atlantis?
X
Y
58
Artificial Intelligence
Artificial
Intelligence
A New Class of
Applications
A New Class of
Technologies
59
Artificial Intelligence
A New Class of
Applications
Expert
Tasks Heuristics Uncertainty
“Complex”
Problem Solving
The algorithm does not exist
A medical encyclopedia is not equivalent to a physician
The algorithm is too complicated
Design a cruise ship
There is an algorithm but it is “useless”
Don’t touch boiling water
The algorithm is not possible
Italy will win the next world cup
60
Artificial Intelligence
A New Class of
Technologies
Non-sequential
Programming
Symbolic
Processing
Knowledge
Engineering
Uncertain
Reasoning
Common Sense
Small minds are concerned with the extraordinary,
great minds with the ordinary"
(Blaise Pascal)
62
Common Sense Types of inference
Induction
Concept formation
Probabilistic reasoning
Abduction
Diagnosis/troubleshooting
Scientific theories
Analogy
Transformation
Derivation
63
Common Sense
• Plausible reasoning
– Quick, efficient response to problems when an
exact solution is not necessary
• Non- monotonic Logic
– Second thoughts: inferences are made
provisionally and can be withdrawn at any time
Common Sense
The Frame Problem
– Classical logic deducts all that is possible from all
that is available
– In the real world the amount of information that is
available is infinite
– It is not possible to represent what does “not”
change in the universe as a result of an action
("ramification problem“)
– Infinite things change, because one can go into
greater and greater detail of description
– The number of preconditions to the execution of
any action is also infinite ("qualification problem“)
Common Sense
Uncertainty
“Maybe i will go shopping”
“I almost won the game”
“This cherry is red”
“Bob is an idiot”
Probability
Probability measures "how often" an event occurs
But we interpret probability as “belief”
Glenn Shafer’s and Stuart Dempster’s “Theory of
Evidence” (1968)
Common Sense
Principle of Incompatibility (Pierre Duhem)
The certainty that a proposition is true
decreases with any increase of its
precision
The power of a vague assertion rests in its
being vague (“I am not tall”)
A very precise assertion is almost never
certain (“I am 1.71cm tall”)
Common Sense
Heuristics
• Knowledge that humans tend to share in a
natural way: rain is wet, lions are dangerous,
most politicians are crooks, carpets get
stained…
• Rules of thumbs
György Polya (1940s): “Heuretics“ - the nature,
power and behavior of heuristics: where it
comes from, how it becomes so convincing,
how it changes
68
Connectionism
A neural network is a set of interconnected neurons
(simple processing units)
Each neuron receives signals from other neurons and
sends an output to other neurons
The signals are “amplified” by the “strength” of the
connection
Connectionism
The strength of the connection changes over time
according to a feedback mechanism (output desired
minus actual output)
The net can be “trained”
Output
Correction
algorithm
Connectionism
• Distributed memory
• Non-sequential programming
• Fault-tolerance
• Recognition
• Learning
Connectionism
Where are we?
Biggest neural computer:
– Stanford and NVIDIA (2013): 11.2 billion connections
(three servers accelerated using 16 GPUs)
– Google (2012): 1.7 billion connections (on a 1,000-
server network) learn to recognize cats in YouTube
videos
• Worm’s brain:
– 1,000 neurons
– But the worm’s brain still outperforms neural nets
Human brain:
– 100 billion neurons
– 200,000 billion connections
72
The Decline of Knowledge-based Systems
• 1. Google it… – Artificial Intelligence was trying to develop
“expert systems” capable of finding a solution to every problem in a given domain, just like a human expert in that domain
– Overt assumption: Domain knowledge is the key to finding solutions
– Hidden assumption: Logical inference is the key to finding the solution
73
The Decline of Knowledge-based Systems
• Google it… – Artificial Intelligence never delivered on the
promises of “expert systems”…
– …but search engines did: there is at least one webpage somewhere that has the solution to a given problem, and it’s just a matter of finding it
– Crwdsourcing did it
74
The Decline of Knowledge-based Systems
• Google it… – Logical inference (intelligence) is irrelevant.
– It’s the quantity of information (not the quality of inference) that matters
– All we needed is a (digital) library big enough and computers powerful enough to search it
– What those computers don’t need is: intelligence
75
The Decline of Knowledge-based Systems
• Google it… – A person can solve any problem as long as
she is capable of searching the Web for the solution
– No other skills required beyond reading skills
– No large, expensive supercomputer required: just a (relatively dumb) smartphone
– The Web plus the search engine does what AI wanted to do: it gives an answer to every possible question that a human can answer (in fact, many more than any one person can answer)
76
The Decline of Knowledge-based Systems
• 2. Big Data
– Very soon Homo Sapiens will be producing
more data every year than in the previous
200,000 years
– Big Data shift the disadvantage to the
knowledge-based approach: too much
knowledge makes it unfeasible
76
77
The Decline of Knowledge-based Systems
• 3. Cheap computing
– Computational power per $ increased
dramatically
– Neural nets and Bayesian nets became
feasible (“deep learning”)
– A.I. based on statistical analysis
– “Best Guess AI”
• Translation
• Search
• Voice recognition
78
2010s
• Machine learning
– Statistical method yields a plausible
result but it has not learned why
– The learned skills cannot be applied to
other fields
79
2010s
• Machine learning
– Jürgen Schmidhuber (1990-…): active, unsupervised, curious, creative systems that create their own self-generated tasks to improve their understanding of the world
• an adaptive predictor or compressor or model of the growing data history as the agent is interacting with its environment
• a reinforcement learner (motivated to invent skills leading to interesting or surprising novel patterns that the predictor/compressor can learn)
80
2010s
• The personal assistant
– Siri (2011)
– GoogleNow (2012)
– Tom Mitchell’s “learning personal
assistant” at CMU (1994-…)
Apple 2011
Stanley Kubrick (1968)
“2001: A Space Odyssey”
81
2010s
• Multi-billion dollar investments in artificial intelligence and robotics in the 2010s
– Amazon (Kiva, 2012),
– Google (Industrial Robotics, Meka, Holomni, Bot & Dolly, DNNresearch, Schaft, Bost, DeepMind, 2013-14),
– IBM (Watson project),
– Microsoft (Project Adam, 2014),
– Apple (Siri, 2011),
– Facebook (DeepFace, 2013),
– Yahoo (LookFlow, 2013),
– etc 81
82
2010s
• DeepMinds machine learning
• Facebook image recognition
• Wise.io
• Saffron
• Narrative Science
• …
• Ebola fighting robot
82
83
The Future of Mind
Computers
Humans
Decision
Making
Data
Processing
1960 2014 Algorithmic Programming Artificial Intelligence
Singularity?
84
The Future of Body
84
85
Robots
86
Robots
• Stats
87
Robots
1962: Joseph Engelberger deploys the
industrial robot Unimate at General
Motors
1969: Stanford Research Institute's
Shakey the Robot
88
Robots
• Valentino Breitenberg’s “vehicles” (1984)
– Vehicle 1: a motor and a sensor
– Vehicle 2: two motors and two sensors
– Increase little by little the circuitry, and
these vehicles seem to acquire not only
new skills, but also a personality.
89
Robots 2000: Cynthia Breazeal's emotional robot, "Kismet"
2003: Hiroshi Ishiguro's Actroid, a young woman
Which is the robot?
90
Robots 2004: Mark Tilden's biomorphic robot Robosapien
2005: Honda's humanoid robot "Asimo"
Asimo over the years
Robots
Special purpose robots:
2001: NEC PaPeRo (a social robot targeting children)
2005: Toyota's Partner (designed for assistance and elderly
care applications)
2007: RobotCub Consortium aggreement, the iCub (for
research in embodied cognition)
2008: Aldebaran Robotics' Nao (for research and education)
2010: NASA's Robonaut-2 (for exploration)
92
Robots 2005: Boston Dynamics' quadruped robot "BigDog“
2008: Nexi (MIT Media Lab), a mobile-dexterous-social robot
2010: Lola Canamero's Nao, a robot that can show its
emotions
2011: Osamu Hasegawa's SOINN-based robot that learns
functions it was not programmed to do
2012: Rodney Brooks' hand programmable robot "Baxter"
93
Robots
• Stats
94
Case study: Japan
Takayama Festival of Mechanical Puppets
95
Case study: Japan
• Joruri/ puppet theater (~1650)
• “Automated mechanisms, or karakuri, were originally
separate from the puppets, used only in stage machinery or
in robot dolls that performed between acts. But the
machinery eventually found its way into the bodies of the
puppets” (Chris Bolton)
96
Case study: Japan
• Oriza Hirata’s robot theater
“I, Worker” (2008)
“Sayonara” (2010)
97
Case study: Japan
• The uncanny valley – Ernst Jentsch: “On the Psychology of the Uncanny” (1906)
– Masahiro Mori: “The Uncanny Valley” (1970)
98
Case study: Japan
• The uncanny valley
– Japanese robots tend to be female because they
look less threatening
99
Artificial Intelligence
• McKensey on A.I.
99
100
A Brief History of Bionics
Jose Delgado
101
A Brief History of Bionic Beings
1957: The first electrical implant in an ear (André Djourno and Charles Eyriès)
1961: William House invents the "cochlear implant", an electronic implant that sends signals from the ear directly to the auditory nerve (as opposed to hearing aids that simply amplify the sound in the ear)
1952: Jose Delgado publishes the first paper on implanting electrodes into human brains: "Permanent Implantation of Multi-lead Electrodes in the Brain"
1965 : Jose Delgado controls a bull via a remote device, injecting fear at will into the beast's brain
1969: Jose Delgado’s book "Physical Control of the Mind - Toward a Psychocivilized Society"
1969: Jose Delgado implants devices in the brain of a monkey and then sends signals in response to the brain's activity, thus creating the first bidirectional brain-machine-brain interface.
102
A Brief History of Bionics
1997: Remotely controlled
cockroaches at Univ of Tokyo
1998: Philip Kennedy develops a brain
implant that can capture the "will"
of a paralyzed man to move an
arm (output neuroprosthetics:
getting data out of the brain into a
machine)
103
A Brief History of Bionics
2000: William Dobelle develops an implanted vision system that allows blind people to see outlines of the scene. His patients Jens Naumann and Cheri Robertson become "bionic" celebrities.
2002: John Chapin debuts the "roborats", rats whose brains are fed electrical signals via a remote computer to guide their movements
104
A Brief History of Bionics
2002: Miguel Nicolelis makes a monkey's brain control a robot's arm via an implanted microchip
2005: Cathy Hutchinson, a paralyzed woman, receives a brain implant from John Donoghue's team that allows her to operate a robotic arm (output neuroprosthetics)
2004: Theodore Berger demonstrates a hippocampal prosthesis that can provide the long-term-memory function lost by a damaged hippocampus
105
A Brief History of Bionics
The age of two-way neural transmission…
2006: The Defense Advanced Research Projects Agency (Darpa) asks scientists to submit "innovative proposals to develop technology to create insect-cyborgs
2013: Miguel Nicolelis makes two rats communicate by capturing the "thoughts" of one rat's brain and sending them to the other rat's brain over the Internet
106
A Brief History of Bionics The age of two-way neural
transmission…
2013: Rajesh Rao and Andrea Stocco devise a way to send a brain signal from Rao's brain to Stocco's hand over the Internet, i.e. Rao makes Stocco's hand move, the first time that a human controls the body part of another human
2014: An amputee, Dennis Aabo, receives an artificial hand from Silvestro Micera's team capable of sending electrical signals to the nervous system so as to create the touch sensation
107
A Brief History of Bionics
Neuro-engineering?
(http://targetedindividualscanada.com) (http://its-interesting.com)
108 108
Biotech
Genetics
1944: Oswald Avery discovers that genes are made of DNA
1953: Francis Crick and James Watson discover the double helix of the DNA
1961: Jacob and Monod discover gene regulation
1961: Jacob and Brenner discover messenger RNA
1961: Marshall Nirenberg cracks the genetic code (translation of four-letter genetic code into twenty-letter language of proteins)
109 109
Biotech
Genetics
1973: Stanley Cohen and Herbert Boyer create the first recombinant DNA organism
1976: Genentech, the first major biotech company
1977: Frederick Sanger publishes the first full DNA genome of a living being
1990: William French Anderson performs the first procedure of gene therapy
1992: Calgene creates the "Flavr Savr" tomato, the first genetically-engineered food to be sold in stores
110
Biotech
1997: Ian Wilmut clones the first mammal, the sheep Dolly
2003: The Human Genome Project is completed
2006: Personal genomics (23andMe, Syapse, Genophen)
2010: Craig Venter and Hamilton Smith reprogram a bacterium's DNA
2010: Cheap printers for living beings (OpenPCR, Cambrian Genomics)
2012: Markus Covert simulates an entire living organism in software (Mycoplasma Genitalium)
2012: Crisp/Cas9 technique for genome editing
Biotech
The price of DNA sequencing your genome has dropped 99% since 2003 ($3,000 in 2013)
111
Moore's law vs Cost per genome
Biotech
112
Biohacking
• “Biology is technology” (Rob Carlson)
• A community of worldwide hobbyists
• Public-domain databases of genetic parts
• MIT Registry of Standard Biological Parts (2003)
• iGEM Jamboree and BioBricks (2004)
• BioCurious (2010)
113
Biohacking
• Biocurious (Sunnyvale) - -20C Freezer
– PCR Machines
– qPCR
– Balance
– Autoclave
– Micropipettes, single and multi-channel
– Fluorescent Microscope
– Microcentrifuges
– Protein Purification System
– Vortexers
– Ultrasonic Bath
– CO2 Incubator
114
Biohacking
• iGEM = International Genetically Engineered Machine
• “Open source” biotech
• Global grassroots synthetic-biology revolution
• Student bioengineers from all over the world create new life forms and race them every year at the iGEM Jamboree in Boston (since 2004)
• 2014: 2,500 competitors from 32 countries
• Repository of 20,000 biological parts (biobricks)
• They create mostly microbes (e.g., organisms detecting and eliminating water pollutants)
115
Drew Endy
(Stanford), iGEM
and BioBricks
Foundation
Biohacking
• PCR printers (identify a piece of DNA and make copies of it)
• OpenPCR (cheap Polymerase Chain Reaction printer)
• Cambrian Genomics: a laser printer for living beings
116
Biohacking
• Autodesk’s Project Cyborg: design tools for biohackers (quote: “Project Cyborg is a cloud-based meta-platform of design tools for programming matter across domains and scales”)
117
118
The Future of Body
Machines
Humans
Mind Uploading
Flesh and Bones
2014
Singularity?
Bio re-engineering
Neural Implants
119
The Future of Body
• Meditation:
– The longest living bodies on the planet
have no brain: bacteria and trees.
120
The Singularity?
120
Ray Kurzweil at the
Singularity University
121
The Singularity?
Ray Kurzweil’s predictions
• Infinite life extension: “Medical technology will be more than a thousand times more advanced than it is today… every new year of research guaranteeing at least one more year of life expectancy”* (2022)
• Precise computer simulations of all regions of the human brain (2027)
• Small computers will have the same processing power as human brains (2029)
• 2030s: Mind uploading - humans become software-based
• 2045: The Singularity 121 * recently postponed to 2040
122
The Singularity?
2014: Deep Knowledge Ventures (Hong Kong)
appoints an algorithm to its board of directors
122
123
The Singularity?
• The Apocalypse has happened many times – Book of Revelation (1st c AD)
– …
– Year 1,000
– …
– Nostradamus (16th century)
– …
– Pierre Teilhard de Chardin’s Omega Point (1950)
– Dorothy Martin/Marion Keech’s planet Clarion (1954)
– Nuclear holocaust (1950s-80s)
– Year 2,000
– Harold Camping’s Biblical calculations (2011)
– End of the Mayan calendar (2012)
123
Albrecht Dürer: The
Four Horsemen of the
Apocalypse (1498)
124
The Singularity?
Five arguments against the Singularity
1. Reality Check
2. Accelerating Progress?
3. Non-human Intelligence
4. Human Intelligence
5. A Critique of the Turing Test
124
125
Reality Check
Why the Singularity is not Coming any
Time Soon & other Meditations on the
Post-Human Condition and the Future
of Intelligence
126
Reality Check
• The curse of Moore’s law
– The motivation to come up with creative ideas in A.I. was due to slow, big and expensive machines.
– Brute force (100s of supercomputers running in parallel) can find solutions using fairly dumb techniques
– Actually, you can find the answer to most questions by simply using a search engine: no need to think, no need for intelligence
127
Reality Check
• Recognizing a cat is something that any mouse
can do (it took 16,000 computers working in
parallel)
• Voice recognition and handwriting recognition
still fail most of the time, especially in
everyday interactions
128
Reality Check
• IBM's Watson does not understand the question
(it is fed in digital format)
• IBM’s "Deep Blue" beat a chess master but was
given unfair advantages
• “What Curiosity (robot) has done in 200 days a
human field researcher could do in an easy
afternoon" (NASA planetary scientist Chris
McKay, 2013)
129
Reality Check
• Machine translation in 2013 (random sentences
from my website translated by Google):
– "Graham Nash the content of which led nasal
harmony“
– "On that album historian who gave the blues
revival“
– "Started with a pompous hype on wave of
hippie phenomenon"
130
Reality Check
• A remote-controlled toy is NOT a step
toward superhuman intelligence
• Human-looking automata that mimic
human behavior have been built since
ancient times
• A human being is NOT a toy (yet)
131
Reality Check
• The brain of the roundworm (a few
hundred neurons connected by a few
thousand synapses) is still smarter than
the smartest neural network ever built.
132
Reality Check
• An easy science
– Artificial Intelligence is not subjected to the
same scrutiny as other sciences
– Its success stories are largely unproven
133
Reality Check
• 60 years later it is not machines that learned
to understand human language but humans
who got used to speak like machines in order
to be understood by automated customer
support (and mostly not even speak it but
simply press keys)
134
Reality Check
• What “automation” really means…
– The jobs that have been automated are
repetitive and trivial.
– And in most cases the automation of those
jobs has required the user/customer to accept
a lower (not higher) quality of service.
– The more automation around you, the more
you (you) are forced to behave like a machine
to interact with machines
135
Reality Check
• Intelligent Behavior from Structured
Environments
136
Reality Check
• Structuring the Environment
– We structure the chaos of nature because it
makes it easier to survive and thrive in it
– The more we structure the environment, the
easier for extremely dumb people and
machines to survive and thrive in it.
– It is easy to build a machine that has to operate
in a highly structured environment
– What really "does it" is not the machine: it's the
structured environment
137
Reality Check
• Semantics
– It is not intelligent to talk about intelligent
machines: whatever they do is not what we do,
and, therefore, is neither "intelligent" nor "stupid"
(attributes invented to define human behavior)
– We apply to machines many words invented for
humans simply because we don't have a
vocabulary for the states of machines
138
Reality Check
• Semantics
– Memory is reconstructive
– Data storage is not “memory”
– Exponentially increasing data storage does not
mean better memory
– What is “computer speed”?
– Who is faster at picking a cherry from a tree,
the fastest computer in the world or you?
139
Reality Check
• Where A.I. is truly successful…
– Most machine intelligence is being employed to couple real-time customization and machine learning in order to understand who you are and tailor situations in real time that will prompt you to buy some products (custom advertising)
– "The best minds of my generation are thinking about how to make people click ads" (former Facebook research scientist Jeff Hammerbacher in 2012)
– So far A.I. has not created better doctors or engineers, but better traveling salesmen
140
Artificial General Intelligence
• Task-specific vs General-purpose
Intelligence
• Originally, A.I. was looking for general-
purpose intelligence
• Today’s A.I. is looking for task-specific
intelligence (recognizing a cat, driving a car)
141
Artificial General Intelligence
• How to simulate an average human (not just one
human task) - the “logic theorist” solution (1960s):
create a system that can perform reasoning on
knowledge and infer the correct behavior for any
situation
• How to simulate an average human (not just one
human task) - the brute force solution (2000s):
create one specific program/robot for each of the
millions of possible situations, and then millions of
their variants
142
Artificial General Intelligence
• The Multiplication of Appliances and Artificial Intelligence by Enumeration – We have machines that dispense money
(ATMs), machines that wash clothes (washing machines), machines that control the temperature of a room (thermostats), and machines that control the speed of a car (cruise controls).
– We can build machines for all the other tasks and then collectively call them “equal” to humans
143
Artificial General Intelligence
• The enumeration problem: which human functions qualify as "intelligent"? – There are very human functions that people
don't normally associate with "intelligence". They just happen to be things that human bodies do.
– Do we really want machines that fall asleep or urinate?
– We swing arms when we walk, but we don't consider "swinging arms while walking" a necessary feature of intelligent beings.
144
Tips for better A.I.
1. IBM's Watson of 2013 consumes 85,000 Watts
compared with the human brain's 20 Watts.
2. The brain is an analog device, not digital
3. What we need: a machine that has only a limited
knowledge of all the chess games ever played and
is allowed to run only so many logical steps before
making a move and can still consistently beat the
world champion of chess.
4. Memory is not storage
145
Tips for better A.I.
• What conditions may foster a breakthrough:
it is not the abundance of a resource (such
as computing power or information) that
triggers a major paradigm shift but the
scarcity of a resource.
146 146
Accelerating progress?
• One century ago, within a relatively short period of time, the world adopted:
– the car,
– the airplane,
– the telephone,
– the radio
– the record
– cinema
• while at the same time the visual arts went through
– Impressionism,
– Cubism
– Expressionism
147 147
Accelerating progress?
• while at the same time science came up with
– Quantum Mechanics
– Relativity
• while at the same time the office was revolutionized by
– cash registers,
– adding machines,
– typewriters
• while at the same time the home was revolutionized by
– dishwasher,
– refrigerator,
– air conditioning
148 148
Accelerating progress?
• while at the same time cities adopted high-rise
buildings
149 149
Accelerating progress?
• There were only 5 radio stations in 1921 but already 525 in 1923
• The USA produced 11,200 cars in 1903, but already 1.5 million in 1916
• By 1917 a whopping 40% of households had a telephone in the USA up from 5% in 1900.
• The Wright brothers flew the first plane in 1903: during World War I (1915-18) more than 200,000 planes were built
150 150
Accelerating progress?
• On the other hand today:
– 44 years after the Moon landing we still haven't
sent a human being to any planet
– The only supersonic plane (the Concorde) has
been retired
– We still drive cars, fly on planes, talk in
phones, use the same kitchen appliances
151 151
Accelerating progress?
• We chronically underestimate progress in
previous centuries because most of us are
ignorant about those eras.
152
A Comparative History of Accelerating
Progress
• On April 3, 1988 the Los Angeles Times
Magazine ran a piece titled "L.A. 2013“
– two robots per family (including cooking
and washing)
– Intelligent kitchen appliances widespread
– self-driving cars widespread
153
A Comparative History of Accelerating
Progress
• Today there is a lot of change
• But change is not necessarily progress
• It is mostly fashion created by marketing
and/or planned obsolescence (progress
for whom?)
154
What would Turing say today?
What took you
guys so long???
155
What would Turing say today?
• Why did it take you so long?
– The Hubble telescope transmits 0.1 terabytes of data
a week, about one million times more data than the
Palomar telescope of 1936
– In 1940 the highest point ever reached by an aviator
was 10 kms. In 1969 Neil Armstrong traveled 380
million kms up in the sky, i.e. 38 million times
higher.
– In 60 years the speed of computers has increased
“only” ten thousand times
156
What would Turing say today?
• Hardware: other than miniaturization, what
has really changed?
– It still runs on electricity
– It still uses binary logic
– It is still a Turing machine (e.g., wildly
different in nature and structure from a
human brain)
157
What would Turing say today?
• Software: other than having 12 million programmers work on thousands of programs (instead of the six who programmed the ENIAC), what has really changed?
– It is still written in an artificial language that is difficult to understand
– It is still full of bugs
– It still changes all the time
– It is still sequential processing (e.g., wildly different in nature and structure from a human brain)
158
What would Turing say today?
And I’m
supposed to
be impressed?
159
Non-human Intelligence
• Super-human intelligence has been around for a
long time: many animals have powers we don't
have
160
Non-human Intelligence
• Bats can avoid objects in absolute darkness at impressive speeds
• Migratory animals can navigate vast territories
• Birds are equipped with a sixth sense for the Earth's magnetic field
• Some animals have the ability to camouflage
• The best color vision is in birds, fish and insects
• Many animals have night vision
• Animals can see, sniff and hear things that we cannot
161
Non-human Intelligence
• And don't underestimate the brain of an insect
either: how many people can fly and land upside
down on a ceiling?
162
Non-human Intelligence
• We already built machines that can do things that
are impossible for humans:
– Telescopes and microscopes can see things that
humans cannot see
– We cannot do what light bulbs do
– We cannot touch the groove of a rotating vinyl
record and produce the sound of an entire
philharmonic orchestra
163
Super-human Machine Intelligence
• The medieval clock could already do
something that no human can
possibly do: keeping time
• That’s why we have to ask “What
time is it?”
164
Non-human Intelligence
• What is the difference between non-
human intelligence (which is already here
and has always existed) and super-human
intelligence?
165
Super-human intelligence
• Possible: Colin McGinn’s cognitive closure
(there are things we will never understand)
• Impossible: David Deutsch’s endless
explanation (we are as intelligent as it gets)
166
Dangers of machine intelligence
• Who's Responsible for a Machine's Action?
• We believe machines more than we believe
humans
• Should there be speed limits for machines?
• We are criminalizing Common Sense
• You Are a Budget
• The dangers of clouding - Wikipedia as a
force for evil
167
Dangers of machine intelligence
• The biggest danger of all: decelerating
human intelligence
168
The Turing Point
• The Turing Test was asking “when can machines be
said to be as intelligent as humans?”
• This “Turing point” can be achieved by
1. Making machines smarter, or
2. Making humans dumber
HOMO MACHINE
IQ
HOMO MACHINE
IQ 1. 2.
169
What can machines do now that they
could not do 50 years ago?
• They are faster, cheaper, can store larger
amounts of information and can use
telecommunication lines
170
What can humans do now that they could
not do 50 years ago?
• Use the new machines
• On the other hand, they are not capable of doing a lot of things that they were capable of doing 50 years ago from arithmetic to finding a place not to mention attention span and social skills (and some of these skills may be vital for survival)
• Survival skills are higher in low-tech societies (this has been true for a while)
• General knowledge (history, geography, math) is higher in low-tech societies (coming soon)
171
The Post-Turing Thesis
• If machines are not getting
much smarter while humans
are getting dumber…
• … then eventually we will
have machines that are
smarter than humans
• The Turing Point (the
Singularity?) is coming HOMO MACHINE
IQ
172
A Tool is not a Skill
• In a sense, technology is about giving people
the tools to become dumber and still continue
to perform
• People make tools that make people
obsolete, redundant and dumb
173
Decelerating Human Intelligence
• The success of many high-tech projects
depends not on making smarter technology
but on making dumber users
• Users must change behavior in order to make
a new device or application appear more
useful than it is.
174
Turning People into Machines
• “They” increasingly expect us to behave like machines in order to interact efficiently with machines: we have to speak a “machine language” to phone customer support, automatic teller machines, gas pumps, etc.
• In most phone and web transactions the first question you are asked is a number (account #, frequent flyer#…) and you are talking to a machine
• Rules and regulations (driving a car, eating at restaurants, crossing a street) increasingly turn us into machines that must follow simple sequential steps in order to get what we need
175
Turning People into Machines
• Rules to hike in the *wilderness* (there is even a rule
for peeing)
176
Decelerating Human Intelligence
• Is it possible that humans have moved a
lot closer towards machines than
machines have moved towards humans?
177
The Silicon Valley Paradigm
• “They” increasingly expect us to study lengthy
manuals and to guess how a machine works
rather than design machines that do what we
want the way we like it
• A study by the Technical University of
Eindhoven found that half of the returned
electronic devices are not malfunctioning: the
consumer just couldn't figure out how to use
them
178
The Singularity
• The Turing Test may become a self-
fulfilling prophecy: as we (claim to) build
“smarter” machines, we may make dumber
people.
• Eventually there will be an army of greater-
than-human intelligence
179
The Future is not You
• The combination of smartphones and websites offers a glimpse of a day when one will not need to know anything because it will be possible to find everything in a second anywhere at any time by using just one omnipowerful tool.
• An individual will only need to be good at operating that one tool. That tool will be able to access an almost infinite library of knowledge and… intelligence.
180
The Difference: You vs It
• Human minds are better than machines at
– Improvisation
– Imagination
– (in a word: "creative improvisation")
• Human minds can manage dangerous and
unpredictable situations
• Human minds can be “irrational”
181
The Difference: You vs It
• Modern society organizes our lives to remove
danger and unpredictability.
• Modern society empowers us with tools that
eliminate the need for improvisation and
imagination
• Modern society dislikes (and sometimes outlaws)
irrationality
182
The Difference: You vs It
• We build
– Redundancy
– Backups
– Distributed systems
• to make sure that machines can do their job 24/7
in any conditions.
• We do not build anything to make sure that minds
can still do their job of creative improvisation
183
A Critique of the Turing Test
(while we’re still intelligent)
184
The Turing Test
The “Turing point”: a computer can be said to be intelligent if its
answers are indistinguishable from the answers of a human
being
? ?
185
The Turing Test
The fundamental critique to the Turing Test
• The computer (a Turing machine) cannot (qualitatively)
do what the human brain does because the brain
– does parallel processing rather than sequential
processing
– uses pattern matching rather than binary logic
– is a connectionist network rather than a Turing
machine
186
The Turing Test
The Turing Test
• John Searle’s Chinese room (1980)
– Whatever a computer is computing, the computer
does not "know" that it is computing it
– A computer does not know what it is doing,
therefore “that” is not what it is doing
– Objection: The room + the machine “knows”
187
The Turing Test
The Turing Test
• Hubert Dreyfus (1972):
– Experience vs knowledge
– Meaning is contextual
– Novice to expert
– Minds do not use a theory about the everyday world
– Know-how vs know that
• Terry Winograd
– Intelligent systems act, don't think.
– People are “thrown” in the real world
188
The Turing Test
The Turing Test
• Rodney Brooks (1986)
– Situated reasoning
– Intelligence cannot be separated from the body.
– Intelligence is not only a process of the brain, it is
embodied in the physical world
– Cognition is grounded in the physical interactions
with the world
– There is no need for a central representation of the
world
– Objection: Brooks’ robots can’t do math
189
The Turing Test
The Turing Test
• John Randolph Lucas (1961) & Roger Penrose
– Goedel’s limit: Every formal system
(>Arithmetic) contains a statement that cannot
be proved
– Some logical operations are not computable,
nonetheless the human mind can treat them (at
least to prove that they are not computable)
– The human mind is superior to a computing
machine
190
The Turing Test
The Turing Test
• John Randolph Lucas (1961) & Roger Penrose
– Objection: a computer can observe the failure of
“another” computer’s formal system
– Goedel’s theorem is about the limitation of the
human mind: a machine that escapes Goedel’s
theorem can exist and can be discovered by
humans, but not built by humans
191
The Turing Test
• What is measured: intelligence, cognition, brain, mind, or consciousness?
• What is measured: one machine, ..., all machines?
• What is intelligence? What is a brain? What is a mind? What is life?
• Who is the observer? Who is the judge?
• What is the instrument (instrument = observer)?
• What if a human fails the Turing test?
The Turing Test
192
The Turing Test
• Someone has hidden a person in a room and a
computer in the other room.
• We are allowed to ask any questions.
• The person and the computer reply in their own way.
• If we cannot tell which one is the person and which
one is the computer, then the computer has become
intelligent.
193
Who is Testing
• Someone has to determine whether the answers to
her questions come from a human or a machine
• Who is the judge who decides if the Turing Test
succeeds? What instrument does this test use?
• A human? A machine?
• How “intelligent” is the judge?
194
Who is Testing
• Can a mentally retarded person judge the test?
• Can somebody under the influence of drugs
perform it?
• …a priest, an attorney, an Australian aborigine, a
farmer, a librarian, a physician, an economist...?
• …the most intelligent human?
• The result of the test can vary wildly depending
on who is the judge
195
Who are we Testing?
• If a machine fails the test (i.e. the judge thinks the
machine is a machine), then Turing concludes that
the machine is not intelligent
• What does Turing conclude if a human fails the test
(if the judge thinks that the human is a machine)?
That humans are not intelligent?
196
What are we Testing?
• The Turing Test is about behavior
• The Turing test measures how good a machine
is at answering questions, nothing more.
• “Can a machine be built that will fool a human
being into believing it is another human being?”
is not identical to “Can a machine think?”
• If we answer “yes” to the first question, we don’t
necessarily answer “yes” to the second.
197
The Turing Point
• The Turing Test asks when can we say that a
machine has become as intelligent as humans.
• The Turing Test is about humans as much as it is
about the machine because it can be equivalently
be formulated as: when can we say that humans
have become less intelligent than a machine?
• The Turing Test cannot be abstracted from a
sociological context. Whenever one separates
sociology and technology, one misses the point.
198
The ultimate Turing Test
• Build a machine that reproduces my brain,
neuron by neuron, synapses by synapses
• Will that machine behave exactly like me?
• If yes, is that machine “me”?
The Turing Test
199
The End (for now)
“A man provided with paper, pencil, and
rubber (and subject to strict discipline) is in
effect a universal machine”
(Alan Turing, 1948)
Recommended