Cognition for Intelligent Robotics Consciousness and Ethics Joanna - - PowerPoint PPT Presentation

Joanna J. Bryson

University of Bath, United Kingdom

Cognition for Intelligent Robotics

Consciousness and Ethics

Outline

• Conscious Robots
• Building Robots Ethically

Joanna J. Bryson

Second Annual Vienna Conference on Consciousness (2008) AISB Symposium on Computing & Philosophy (2009) Science properly referenced in paper, demonstrated further in Bryson (CogSci 2009).

Crude, Cheesy, Second-Rate Consciousness

“If the best the roboticists can hope for is the creation of some crude, cheesy, second-rate artificial consciousness, they still win.”

• D. C. Dennett (1994), “The Practical

Requirements for Making a Conscious Robot”, Philosophical Transactions: Physical Sciences and Engineering, 349 p. 137 (133-146).

it is nor hand, nor foot, nor arm, nor face, nor any other part belonging to a man.

What’s Consciousness?

Glenn Matsumura, Wired 2007

SG5-UT Robotic Arm

Tad McGeer's passive dynamic walker Chuck Rosenbergʼs IT, 1997

by any other name?

by any other name? what qualifies for consciousness?

Sept 1 1 9 9 3 Sept 1 1 9 9 4 Sept 1 1 9 9 5 Sept 1 1 9 9 6 Sept 1 1 9 9 7 Syst em soft ware (0t h) Syst em soft ware (commercial processor) Periperhal Mot ion Saccades VO R Smoot h pursuit Vergence based st ereo Ullman-esque visual rout ines Face pop-out s Face remembering Face recognit ion Gest ure recognit ion Facial gest ure recog. Body mot ion recog. Own hand t racking Physical schema based obj. recog. Bring hands midline Hand linking Grasping, & t ransf er Specific obj. recog. Generic object recog. Body-based met aphors DOF reduct ion (specific coords) DOF reduct ion (generic coords) Bat t ing st at ic

• bject s

Head/ eye coord Body st abilit y, leaning, rest ing Head/ body/ eye/ coord Body+arm reaching Body mimicry Manipulat ion t urn t aking Sound localizat ion Sound/ mot ion correl Human voice ext ract ion Sound-based manip. Voice/ face assoc Voice t urn t aking Prot o language Visual imagery Symbolizat ion Imaginat ion Ment al rehearsal Mult iple-draft s emergence Tone ident ificat ion

“Building Brains for Bodies”, Brooks & Stein (1993), MIT AI lab tech report 1439.

Sept 1 1 9 9 3 Sept 1 1 9 9 4 Sept 1 1 9 9 5 Sept 1 1 9 9 6 Sept 1 1 9 9 7 Syst em soft ware (0t h) Syst em soft ware (commercial processor) Periperhal Mot ion Saccades VO R Smoot h pursuit Vergence based st ereo Ullman-esque visual rout ines Face pop-out s Face remembering Face recognit ion Gest ure recognit ion Facial gest ure recog. Body mot ion recog. Own hand t racking Physical schema based obj. recog. Bring hands midline Hand linking Grasping, & t ransf er Specific obj. recog. Generic object recog. Body-based met aphors DOF reduct ion (specific coords) DOF reduct ion (generic coords) Bat t ing st at ic

• bject s

Head/ eye coord Body st abilit y, leaning, rest ing Head/ body/ eye/ coord Body+arm reaching Body mimicry Manipulat ion t urn t aking Sound localizat ion Sound/ mot ion correl Human voice ext ract ion Sound-based manip. Voice/ face assoc Voice t urn t aking Prot o language Visual imagery Symbolizat ion Imaginat ion Ment al rehearsal Mult iple-draft s emergence Tone ident ificat ion

“Building Brains for Bodies”, Brooks & Stein (1993), MIT AI lab tech report 1439.

Sept 1 1 9 9 3 Sept 1 1 9 9 4 Sept 1 1 9 9 5 Sept 1 1 9 9 6 Sept 1 1 9 9 7 Syst em soft ware (0t h) Syst em soft ware (commercial processor) Periperhal Mot ion Saccades VO R Smoot h pursuit Vergence based st ereo Ullman-esque visual rout ines Face pop-out s Face remembering Face recognit ion Gest ure recognit ion Facial gest ure recog. Body mot ion recog. Own hand t racking Physical schema based obj. recog. Bring hands midline Hand linking Grasping, & t ransf er Specific obj. recog. Generic object recog. Body-based met aphors DOF reduct ion (specific coords) DOF reduct ion (generic coords) Bat t ing st at ic

• bject s

Head/ eye coord Body st abilit y, leaning, rest ing Head/ body/ eye/ coord Body+arm reaching Body mimicry Manipulat ion t urn t aking Sound localizat ion Sound/ mot ion correl Human voice ext ract ion Sound-based manip. Voice/ face assoc Voice t urn t aking Prot o language Visual imagery Symbolizat ion Imaginat ion Ment al rehearsal Mult iple-draft s emergence Tone ident ificat ion

“Building Brains for Bodies”, Brooks & Stein (1993), MIT AI lab tech report 1439.

Outline

• The very idea of robot brains and

consciousness.

• Characteristics of consciousness.
• A functional theory & animal models.
• Consciousness in other intelligent systems.

Dennett (VCC08)

“Contents arise, get revised, contribute to... the modulation of behavior, and in the process leave their traces in memory...” “Only [commonality is] the historical property of having won a temporally local competition with sufficient decisiveness... to enable recollection...”

• Selection from concurrent options.
• Indicated by episodic memory.

Outline

• The very idea of robot brains and

consciousness.

• Characteristics of consciousness.
• A functional theory & animal models.
• Consciousness in other intelligent systems.

Function Theory

• Consciousness is holding one stimulus in

mind while searching options primed by it for a better response.

• Only triggered when next action isn’t
• bvious (reflexive or trained).
• Side effect 1: special types of learning.
• Side effect 2: long reaction times. Focus

attention longer when less certain.

Ex 1: find the green T

Ex 1: find the green T

T T T T T T T T T T T T T T T T T T T T T T T T T T T

Ex 1: find the green T

Ex 1: find the green T

L L L T L T L T T T L T L L T L T T L T T L T T T L T L

Ex 2: Hippocampal & Episodic Memory

• People who lose their

hippocampus say they can’t remember anything, but can still be conditioned (and learn semantic knowledge.)

• Rats that lose their hippocampus

can be conditioned, but act like they can’t remember where they were just earlier.

Med Associates Mazes

Ex 3: Learning New Rewards (or not)

• Monkeys that learn chained pairs of

values (A>B; B>C; C>D; D>E; E>F) normally are faster at assessing stimuli the further they are on the chain (B>E faster than B>D).

• Elderly monkeys are always fast.
• Elderly monkeys also don’t learn

when you change the reward scheme -- not aware?

Herb Terrace, Columbia, NY

Outline

• The very idea of robot brains and

consciousness.

• Characteristics of consciousness.
• A functional theory & animal models.
• Consciousness in other intelligent systems.

Intelligent Systems

Andrea Thomaz, MIT

Intelligent Systems

Andrea Thomaz, MIT Charlie Kemp, GA Tech

Intelligent Systems

by any other name?

Andrea Thomaz, MIT Charlie Kemp, GA Tech

Intelligent Systems

by any other name?

Andrea Thomaz, MIT Charlie Kemp, GA Tech

How about “spreading- activation implementation

• f bounded depth-first

search”?

Conscious Systems

Davide Vecchi, KLI

Conscious Systems

• 1. Hold one stimulus in mind

(from concurrent options).

Davide Vecchi, KLI

Conscious Systems

• 1. Hold one stimulus in mind

(from concurrent options).

• 2. Search primed responses.

Davide Vecchi, KLI

Conscious Systems

• 1. Hold one stimulus in mind

(from concurrent options).

• 2. Search primed responses.
• 3. Take time proportional to

uncertainty.

Davide Vecchi, KLI

Conscious Systems

• 1. Hold one stimulus in mind

(from concurrent options).

• 2. Search primed responses.
• 3. Take time proportional to

uncertainty.

• Concurrent cognitive systems
• You, your browser & Google

all remember the episode.

Davide Vecchi, KLI

Dennett (reminder)

“Contents arise, get revised, contribute to... the modulation of behavior, and in the process leave their traces in memory...” “Only [commonality is] the historical property of having won a temporally local competition with sufficient decisiveness... to enable recollection...”

• Selection from concurrent options.
• Indicated by episodic memory.

Conclusions

• A few robots could already be called

“conscious” (in a crude, cheesy, second- rate kind of way.)

• Many robots and other technology are a

part of conscious systems, using humans to focus the action & learning.

Outline

• Conscious Robots
• Building Robots Ethically

Joanna J. Bryson

Building Robots Ethically

References

• Joanna Bryson and Phil Kime, “Just Another Artifact: Ethics and the

Empirical Experience of AI”, Fifteenth International Congress on Cybernetics, pp. 385–390, Namur, 1998.

• Joanna Bryson, “A Proposal for the Humanoid Agent-builders League

(HAL)”, The AISB Symposium on Artificial Intelligence, Ethics and (Quasi-)Human Rights, ed. J. Barnden, pp. 1-6; Birmingham UK, 2000.

• Joanna J. Bryson, “Robots Should Be Slaves”, in Artificial Companions in

Society: Scientific, Economic, Psychological and Philosophical Perspectives, Yorick Wilks (ed.), John Benjamins, 2009 in press.

• Joanna J. Bryson “Building Persons is a Choice”, an invited

commentary on Anne Forest, “Robots and Theology”, Erwägen Wissen Ethik, November 2009 in press.

Should Robots Be Ethical?

Is it Ethical to Unplug a Robot?

Is it Ethical to Unplug a Robot?

Will Robots Rule?

Calculators Are Better at Math than We Are

• Gorillas are stronger than we are.
• Terrorists have big guns.
• Many countries have nuclear weapons.
• Taking over the world (or indeed any

action) requires motivation.

Where Do Motivations Come From?

• In humans?
• In other animals?
• In robots?

My (Informed) Opinion

My (Informed) Opinion

• Robots and other AI are definitionally

human artifacts.

My (Informed) Opinion

• Robots and other AI are definitionally

human artifacts.

• We are ultimately responsible for any

motivation they have, and anything they do.

My (Informed) Opinion

• Robots and other AI are definitionally

human artifacts.

• We are ultimately responsible for any

motivation they have, and anything they do.

• Ethics are defined socially & legally.

My (Informed) Opinion

• Robots and other AI are definitionally

human artifacts.

• We are ultimately responsible for any

motivation they have, and anything they do.

• Ethics are defined socially & legally.
• It would be an abrogation of our own

responsibilities if we assigned robots ethical responsibility.

Ordinary Law

• If a machine does damage and it was

functioning correctly, it is the responsibility

• f the operator.

Ordinary Law

• If a machine does damage and it was

functioning correctly, it is the responsibility

• f the operator.
• If it does damage the operator could not

have anticipated, it is the responsibility of the manufacturer.

Military Law (UK)

• If an autonomous weapon system kills the

wrong people, it is the fault of the programmer (the military person that set its target, not the person who designed it.)

Military Law (UK)

• If an autonomous weapon system kills the

wrong people, it is the fault of the programmer (the military person that set its target, not the person who designed it.)

• This implies the UK military does not use

such weapons unless they have certified the manufacture to be perfect.

Should Robots Be Ethical?

Should Robots Be Ethical?

The US military pays for this research.

Implications

• If it would be an abrogation of our own

responsibilities if we assigned robots ethical responsibility...

Implications

• If it would be an abrogation of our own

responsibilities if we assigned robots ethical responsibility...

• Then perhaps if people want to assign

robots ethical responsibility, those people want to evade responsibility.

Professionalism (Think About What You Do)

• There are many other reasons ordinary

people want robots to be human-like.

• It would be exciting, you could buy &

boss around a superman, you could create something like life (be god-like)...

• IMHO: Professionals have an obligation not

to exploit these desires if it would damage society or our customers.

Conclusion

• Building robots ethically involves:
• making them reliable,
• being honest about their capabilities,
• not calling them “ethical”.

Outline

• Conscious Robots
• Building Robots Ethically

If you are more interested in AI & action selection than robots...

Bath’s Animation Doctoral Training Centre

• The EPSRC through Bath invests £180,000

per student “high-flier, industry leader”.

• UK Company interviews student, helps pick

project and 1 year of courses (Bath & Bournemouth), 3 years on company site.

• 50 studentships over 5(ish) years.

“Framestore, HP, Microsoft, Bizarre Games and others” already signed up.

Thanks!

Philipp Rohlfshagen Cyril Brom (et al) Jan Drugowitsch Sam Partington Tristan Caulfield

Joanna J. Bryson

University of Bath, United Kingdom

Cognition for Intelligent Robotics

Consciousness and Ethics