Fausto Giunchiglia

To be cited as: Fausto Giunchiglia. On Line presentation at http://disi.unitn.it/~fausto/CH.pdf

Acknowledgements to: Daniele Miorandi, Ronald Chenu, Stuart Anderson, Michael Rovatsos, Mark Hartswood. Theseideas have been elaborated as part of the FP7 FET IP project «Smart Society» http://www.smart-society-project.eu/

Computational Humanism

Trento 11/03/2017 www.smart-society-project.eu

Index1. Artifacts and services2. Empowering service provision3. Empowering social relations4. Service provision as social computation5. Managing diversity 6. Challenges7. Computational Humanism8. Impact 9. Future10. Demo (video)

Artifacts

3

Historically, inventing and building artifacts has been (and still is) one the main means towards scaling beyond humans’ intrinsic limitations and diversity.

... Tools, Machines Computers, Robots (Artificial Intelligence), Computer networks

Collectives and services

4

Creating collectives has been (and still is) another main means towards scaling beyond humans’ intrinsic limitations and diversity.

The membership of a collective needs to have sufficiently diverse capabilities to achieve its tasks and objectives. In collectives, people help people in doing things they don’t know how to do or that would find very hard to do.

Collectives are the main means for service provision (co-production). Collectives get formed to deliver a service. A collective is the set of people (helped by any form of artifacts) which contribute the achievement of the service.

We take the Service Provider to be the collective who ultimately provides the service and the Service Consumer to be the collective (person) who ultimately is recipient of the service.

Index

1. Artifacts and services2. Empowering service provision3. Empowering social relations4. Service provision as social computation5. Managing diversity6. Challenges7. Computational Humanism8. Impact 9. Future10. Demo (video)

Empowering service provision – so far

6

Access to services is mediated and mainly controlled in the virtual world. Service access is extended, in space and time, to cover potentially all the world.

Examples: a. Uber: access to rides through the people who offer them;b. AirBnB: access to places to stay through the people who manage them;c. …

Advantages:a. Enabling an easily accessible huge (often worldwide) space of possibilities;b. Going beyond humans’ space and time limitations in service delivery;c. Overall, producing better personal and social performanced. ...

Empowering service provision – drawbacks

7

Human intervention is needed at any relevant decision step. All the most «sensitive, mission critical» decisions are left to humans.

The human limitations in their ability to interact with other humans (sometimes humans off line, low speed processing, sequential processing) slow down immensely the process of service provision.

The rigid distinction of roles and the rigid boundary between humans and machines is the main bottle-neck for scaling service provision. Humans and machines have reciprocally no knowledge of the other, of what it does, why it does it, how it does it.

Machines cannot help humans beyond what was originally designed and humans cannot help machines in helping humans.

Index

1. Artifacts and services2. Empowering service provision3. Empowering social relations4. Service provision as social computation5. Managing diversity6. Challenges7. Computational Humanism8. Impact 9. Future10. Demo (video)

Social relations

9

We take a social relation, or social interaction, to be any relationship between two or more individuals as part of a collective.

In everyday life, social relations enable efficient service delivery by providing the service provider and consumer of the reciprocal knowledge of the other, of what she does, why she does it, how it does it. In turn services enable the establishment and evolution in time of social relations. In turn, collectives are formed, operate and deliver their services exploiting a huge infrastructure of social relations.

In machine mediated service provision, so far, services are provided without any support from social relations. All the missing information must be provided by humans. In many cases, service provider and service consumer have reciprocally no knowledge of the other, of what she does, why she does it, how she does it (Important exception: dedicated systems where professionals are trained to run them)

The challenge - Empowering social relationsSmart environments

Smart artifactsRobots

Artificial Intelligence A model which enables people, to develop social relations, and to enable reciprocal services beyond the human limitations (in time, space, processing, memory, service delivery and consumption, social relations).

A technology implementing the model which is always available to the user, 24 hours a day, irrespective of any contextual factor.

Machines do not substitute people. Machines empower their social relations

A provider centric example –online labour market

Kelly is a software developer expert. Recently, Kelly joined a new online labour market platform called VirtualTeam, which dynamically assembles teams to respond to outsourcing requests from major brands worldwide. Today she received an alert that the platform identified her as a possible candidate to join a team ... Her smartphone had already interacted with the platform SP and the team was already completed: ...

...

The following three weeks are very busy, yet the platform supports effectively team-working, maximising productivity of the team members while helping them maintain a healthy work-life balance.

...

The customer is happy: it saved 43% of costs and 80% of time with respect to the other offers. Kelly is happy too: she had the opportunity to get a job which with traditional online labour market she wouldn’t have been able to get, she met great professionals and, well, working in a team is much more fun than being at home alone working as a code monkey! All of this after getting the job in one single click!

But things sometimes go wrong ...

Index1. Artifacts and services2. Empowering service provision3. Empowering social relations4. Service provision as social computation5. Managing diversity6. Challenges7. Computational Humanism8. Impact 9. Future10. Demo (video)

Augmented Collectives

13

Augmented Collectives are the result of the convergence among humans whose social relations are empowered by machines, with capabilities which are more than the sum of the capabilities of the people involved, of collectives not empowered by machines, and of machines when used in isolation.

14

Human Human

MachineMachine

A three layer model of compositionalityA person is characterized by:

1. The set of services (s)he can exploit (service consumer) thanks to, e.g., her/ his smartphone, ... .

2. The set of services (s)he can provide (service provider) thanks to, e.g., her/ his smartphone, as part of a collective... .

3. Her social relations as enabled by the service provider, which provide her with the ability toexpose/ negotitate/ compose with other people/ the services (s)he provides/exploits, (social service prosumer).

Human compositionality is rooted in social relations and realized as the composition of human machine and human compositionality

Service provision as social computation

15

Service provision, as delivered by a collective, is implemented as social computation in the following terms. A Social computation is:

a sequence of computation results, as traced by the machine, where

… each computation step is the delivery of a service, where a service can

… be decomposed in a sequence of simpler computation steps, or

… be an (atomic) service, as performed by a person,

… and where the success or failure of an atomic service is validated by the machine

… in the terms understood by the people, as a prerequisite for the compositionality of people actions

NOTE: Drawing a comparison with how a CPU works: the Control Unit (CU) is the machine and the computation unit (e.g. ALU) is people. The opposite of what usually happens with machines and computers

Service provision as social computation (cont)

16

The role of collectives (via their members):

1. Deliver and exploit services (social service prosumers)

The role of machines (supported by humans):

1. Acquire (partial) knowledge about the service to be provided (modeled as the “current” goal) and of state of affairs (the service context, people, and the current collective) (property recognition);

2. Adapt the collective to deliver the service provision as a plan resulting from the composition of (atomic) services to be delivered by people (contextual privacy, search, provenance management, orchestration, incentive mechanisms);

3. Monitor the results of service provision, recognize them and detect the arousal of unexpected obstacles (anomaly recognition);

4. If a difficulty is detected, go to step 2

Index1. Artifacts and services2. Empowering service provision3. Empowering social relations4. Service provision as social computation5. Managing diversity 6. Challenges7. Impact 8. Computational Humanism9. Future 10. Demo (video)

18

Human Human

MachineMachine

Diversity

Human machine (HM) diversity(Hibridity)

... the (extended) Semantic Gap problem

Human (HH) diversity... Culture, Language, goals, activities, emotions, values, ...

So far mainly studied in the humanitiesNo general computational theories of human diversity

Machine (MM) diversity... the (semantic) Web problem

(out of focus)

Unexpected diversity

Diversity is open-ended

Managing diversity

19

Diversity is open – ended

Managing Diversity = Living in a open world

Activities

Goals

Values

Com

ple

xity o

f re

co

gn

itio

n

Modeling diversity

Co

mp

lexity o

f a

da

pta

tio

n

Use

r fe

ed

ba

ck o

n s

em

an

tics

Use

r in

vo

lve

me

nt in

de

cis

ion

ma

kin

g

Co

mp

ute

r e

mp

ow

ern

en

t

Co

mp

ute

r e

mp

ow

erm

en

t

Adaptability to the service and people unexpected diversity is realized via the emergence of the

most suitable augmented collective able to provide the service.

Context

Index1. Artifacts and services2. Empowering service provision3. Empowering social relations4. Service provision as social computation5. Managing diversity 6. Challenges7. Computational Humanism8. Impact 9. Future10. Demo (video)

Challenges

22

Social computation

Governance models

Organization models

Computation models and languages (e.g., Elephant 2000)

Execution environments

Diversity

Modelling diversity

Recognizing diversity

Adapting to diversity

Complexity of recognition

Co

mp

lexi

ty o

f ad

apta

tio

n

Automation of social computations

Complexity of recognition

Co

mp

lexi

ty o

f ad

apta

tio

n

Social context recognition

Translation

Emotion recognition

...

?

? Towards a Smarter Society

Automation of social computations (cont.)

Index1. Premise: Artifacts and services2. Empowering service provision3. Empowering social relations4. Service provision as social computation5. Managing diversity 6. Challenges7. Computational Humanism8. Impact 9. Future10. Demo (video)

Computational humanism

26

Humans and their social relations at the core of our studies as the way to guide the technology development towards scaling human (collective)

intelligence

From Computer Science to Computational Humanism

27

COMPUTER SCIENCEAS

ENGINEERING

COMPUTER SCIENCEAS

COMPUTATIONALHUMANISM

Index

1. Premise: Artifacts and services2. Empowering service provision3. Empowering social relations4. Service provision as social computation5. Managing diversity 6. Challenges7. Computational Humanism8. Impact9. Future10. Demo (video)

The SmartSociety Impact Triangle

29

Paradigms for a New

Science

SmartCollectives

Open Source ToolkitSocial

Charter--

Res

ea

rch

Long term Impact

www.smart-society-project.eu 30

B2B2C Paradigm - a new generation of Services:

From multi-channel mono-directional service provision …… to end-to-end multi-directional social service provision

Potential huge improvement of the quality of service provision

1. Scale (space, people)2. Time of delivery3. Customization4. Personalization5. All service sectors (financial, education, transportation, retail, media, health/well-

being, …)

Service innovation as the means to Societal Innovation.

Complexity of recognition

Co

mp

lexi

ty o

f ad

apta

tio

n

Social Computing

Social Computation

decentralized through society

Social control of healthcare and disease (Rare Diseases)

Semantic Web

Social response to emergences and crime

Crowdsourcing

• Huge, potential global impact

thanks to a social infrastructure

needed to harness small social

computations

• Small, direct local impact

magnified when replicated

across global society

Workflows

Societal challenges – Scaling over machine intelligence

Fixed Diversity

Complexity of Diversity

Peo

ple

invo

lvem

ent

Societal challenges – Scaling over diversity

Social Computation

decentralized through society

Assuming sufficentRecognition and Adaptability Capabilities by machines

The person to the moon

33

• The ultimate success of Artificial

Intelligence would be to build a

machine whose Intelligence equals

or exceeds human intelligence.

• The ultimate success of Computational

Humanism would be to make it possible

for everybody to exploit, when trying to

achieve her objectives, the best

expertise available in the planet.

• Each person would be augmented with

all the knowledge and resources

available in the world.

Computational Humanism and (strong) Artificial Intelligence

Index

1. Premise: Artifacts and services2. Empowering service provision3. Empowering social relations4. Service provision as social computation5. Managing diversity 6. Challenges7. Computational Humanism8. Impact9. Future10. Demo (video)

The Computational Humanism Triangle

35

University and Research Centers

(Research and Education)

Private sector

(Innovation)Public sector

(Governance)--

ne

w s

cie

nc

e

Computational

Humanism

THANK YOU!Fausto Giunchiglia

www.smart-society-project.eu