Turing Symposium
Turing Symposium
Reading, October 12th 2008
Selmer Bringsjord "God Soul and Turing"
...
Michael Wheleer
the Turing test is not enough to affirm a machine has consciousness
alternative:
doubt and uncertainty
behaveioural diversity --> a machine can variate like a man
Turing said: a machine can variate like a man
Descartes "Discourse on the method" (1637)
what is a machine?
cartesian: --- laws of blind physical causation --- norms of correct or incorrect functioning --- special purpose on an integration of them
Disposition of organs --> producing appropriate actions with some ... task domain
non mechanistic --> adaptive plasticity
MECHANICS IS NOT MAGIC
COMPUTER --> change the notion of machine it is different from Descartes' machine which was a set of purposes. --> general purpose: reasoning machine
frame problem: Heuristics = representation of context
CRC ==> Continuous Reciprocal Causation
Gas Net --> electrical (non synaptic)
modeling
turing test: you pass it only if "distinctive context sensitive adaptive plasticity"
Andrew Hodges: Alan Turing's Imitation Game
theory of twistors --> new approach to fundamental physics by Roger Penrose
building a brain = a new type of machine
1945 --> programs imitating human thinking
march 1946 --> artificial conversational entity
goedel question: relevant or not? Turing did not believe so.
II War --> top/down | bottom/up how can algorithms come to life
programs evolving into another program through experience
the action of the brain could be simulated --> that's what AI thought
turing's conceptual intelligence is somehow dissociated by actual reality
physical body is not important but symbolic --> neurons are having very specific positions
1950--> Turing: dynamical systems are chaotic
Goedel --> mathematical objections: why should you believe that the computer program can simulate what goes on in the brain
nature of dynamic systems
---> uncertainty / indeterminacy
Luciano Floridi
luciano.floridi@philosophy.ox.ac.uc
Turing (1950) --> the imitation game guess the nature of the player
Turing: 1_ minimalist: better the imitation game than the "can machines think?" 2_ levels of abstraction
--> Epistemic = something is known only by an observer at a certain level of abstraction |LOA|
---> could be an interface es of the underground 1908
---> a certain type of observability
level of abstraction: must have a purpose
non-hierarchical = levels are parallel or disjoint
level of abstraction: interface finite, non empty set of observables that can be moderated by transition rules
OBSERVABLES = typed variable modeling in science
computer science --> formal methods
discrete mathematics specifies and analyses the behavior of information system
LOA --> analyses a system and generates a model of the system
epistemic as interface - networks of observables - observables are related by behaviors that moderate the LOA and can be expressed in terms of transition rules - conceptually placed between date and the information spaces of the agent - the threshold where independent systems can act
TURING'S LESSON intuition around intelligence: "do not try to define what intelligence is"
-maker's knowledge perspective what you really know is what you make
hypothesis ---> devise a system to evaluate your hypothesis
- CONSTRUCTABILITY (simulation) - CONTROLLABILITY (the system is fully controllable) - CONFIRMATION (if a machine M passes the test M can/cannot be considered intelligent as the human at that LOA) - MINIMALISM (Turing uses the minimal amount of resources he can)
The Turing test is based on a LOA of questions and answers.
limit: agents are not required to ask questions
manifestation of intelligence: ability to wonder and then the capacity to build semantic artifacts
competition: 25% simulation (5 minutes conversation, a bot fooled 3 humans out of 12)
Maggie Boden "The Turing test and artistic creativity"
We don't have direct human interaction to determine what it is going to happen You can get fundamental mutations in the shape of the images program= is it coming up with artistic style? Despite the stage differences you can get into computer art - passing the Turing test does not imply exceptions (where are humans interpreting computer art as human?)
- the concept of art and creativity --> range of creativity and art --> very human the notion of computer art and of computers passing the test is highly controversial
Owen Holland - "The argument from (machine) consciousness"
The ratio club: Cybernetic Dining Club (1949 - 1958) 21 members
funded by John Bates (neurologist)
Computation of the faculty of Ragione. Wiener --> book (Turing was in the book)
July 26th 1951 ---> telepathy Bates
"The mechanical mind in history"
machines: consciousness or mimicking consciousness?
next year: "International Journal of Machinic Consciousness"
model of the self ==> consciousness (self-representation)
bots:
CRONOS SIMNOS
consciousness test = more an internal behavior than external
what about distributed consciousness?
