Scalable, game-based citizen science: Developing hybrid … · 2019-12-20 · Scalable, game-based citizen science: Developing hybrid intelligence and fostering a population thinking

Scalable, game-based citizen science: Developing hybrid intelligence and fostering a population thinking like scientists

Jacob Sherson, Founder and director of ScienceAtHome and the Center for Hybrid Intelligence

Digitrascope, 30/10 2019

https://hybridintelligence.eu/ sponsored by Carlsberg Foundation

Part 1: the long term perspective of AIPart 2: behind the AI-hypePart 3: Hybrid intelligence: Engaging citizens in the mapping of the human-AI boundaryPart 4: Research-Enabling Game-Based Education: fostering a generation Thinking Like a ScientistPart 5: Quantifying the unquantifiable: assessing and fostering 21st century skills

Are we approaching a singularity?

Ray KurzweilPredicts we will reachsingularity in 2045.

1900 1920 1940 1960 1980 2000 2020 2045

Analyticalengine

Colossus

Univac I

Apple II

IBM PC

PowerMac 64

IBM1130

Surpasses brainpower ofmouse in 2015

Surpasses brainpower of human in 2023

Surpasses brainpower equivalent to that of allhuman brains combined

2023

2045

2015

0.00001

100 000

1

1020

1015

1026

10 000 000 000

Mac Pro

Life 3.0

Max TegmarkLife 3.0: Being Human in the Age of Artificial Intelligence

Are we the generation that will experiencethe singularity?

source: https://www.nature.com/articles/nature14236

Human levelor beyond


Are we really approaching a singularity?Ray KurzweilPredicts we will reachsingularity in 2045.

Need to understand human ability to compute based on little amount of data (intuition or common sense)

Gary Marcus; Deep Learning: A Critical Appraisal https://arxiv.org/abs/1801.00631

VSJust because something appears to be intelligent for a moment or two doesn't means that it really is…

In the world of hybrid intelligence, we need to understand ourselves much better

Daniel Kahneman,

author

• Automation of System 2 tasks (chess)

• Our every day decisions (intuition, creativity) still pose challenges

Fast

Unconscious

Automatic

EverydayDecisions

Error prone

Slow

Conscious

Effortful

ComplexDecisions

Reliable

System 2System 1

Democratic challenge: tech-companies should not monopolize big-data insights

source: https://www.esquire.com/news-politics/a15895746/bust-big-tech-silicon-valley/


https://hybridintelligence.dk/

In the world of hybrid intelligence, don’t just buildinterfaces/products based on a technical solution.

Incoporate (or expand upon) newest human cognitive insightsto create true synergetic hybrid solutions

Specific challenge: understanding and supportingcomplex problem solving

…

Our approach: involving +300k citizens in the quest for hybrid intelligence

+250k citizen helping construct Our Quantum Computer

JJ Sørensen et al. Nature 2016

The coldestcrystal in the world (~4nK)

Presenting complex natural science problems as games

Computer science:NP Hard Problems Millenium Math Challenge:

Turbulence

Statistical thinking

Quantum Physics Experimental control

Presenting complex natural science problems as games

Experimental control

Top player statement (retiredmicrowave systems engineer): ”To me, playing this game was just likemy old work. I never understood

all details of the microwavesystems but decades of

experience must have given mean intuitive understanding of how

to change the black box”

Citizen science games can:• Generate new domain specific knowledge• Offer generic insights into the process of complex problem solving• Facilitate meta‐cognitive reflection in the participants (Thinking like a

scientist)

The ScienceAtHome social science supercollider

Research:1. Realistic (NOT WEIRD*) settings for collective behavior… 2. with precise instrumentation and measurement…3. studied longitudinally over periods of time…4. with a high degree of experimental control (A/B testing).* Social science so far has mainly been confined to lab studies involving students or graduates who are Western, Educated, Industrialized, Rich and Democratic

Virtual world simulator

What are the effects of different policies on economic inequality?

What are the causal effects of American vs. Danish systems of government on a given population?

Do the macroeconomic policies enacted by central banks (e.g. ECB, Federal Reserve) actually work as intended?

The first super collider games

Cognitive profiling, phase1: Systematic validation of basic cognitive skills

Starting position

Memorize Recall

Corsiblocks

Groton mazelearning

Tower of London tasks

Goal position

Skill Lab, the missing link between microscopic and macroscopic data

We believe that equal access to data willlead to equal access of knowledge.

Our cognitive science exploration game has applications including team formation personalised education, diagnostics, and policy advice towards equal opportunity societies.

The citizen science of ###? (### insert your field)

Hyper-personalization through making a pact with customers/citizens

Big take away: Google believes in AI first, we should think HI first!


The upcoming quantum education Flagship (100 M€)

Online May 2019

Existing models of CSBonney et al. 2009: 3 categories of Cit. Sci.: ~Brandt et al. (2010) - 3 models of CS

- Contributory- Collaborative- Co-created

Smithsonian Center for Science Education

Uni-directional

Transactional

Transformational

Price & Lee (2013): 2 models of contributory model: passive contribution and active contribution

English et al. Pyramid of participatory research

Why Research-Enabling Game-Based Education (REGAME)?

• Educational solutions for scalibility and motivation in 21st century education:• Game-based education

• has been attacked for being ”chocolate covered broccoli” (educational but not fun, or fun but not educational)

• Inquiry-based science education (IBSE)• Either addresses grand challenges at the risk of fluffyness or a well

defined role-playing game in which the teacher knows the solution• Challenging to couple to concrete analytic skills in higher grades

when STEM approaches final quantitative assessment

• REGAME increases student motivation • By providing ”epic meaning” through a direct link between the core curriculum

and front-line R&D challenges (i.a. global sustainability goals)• By demonstrating to the students that knowledge is not static but can be

expanded/explored even without a 10 year research education (open endedexploration)

• By turning normal teaching upside down: first explore the engaging/epicperspectives of the subject matter using with a powerful yet intuitive interface and then ”open the hood of the Ferrari” to understand the underlyingformalism. ”Challenged”

”Gifted”

Intuition Analytical

Under explored

How can we use classical physics to perform betterin Quantum Moves?

The sloshing quantum “liquid” is very similar to:• Sloshing water• A ball rolling in a bowl• A pendulum• A spring

Newton’s equations

ScienceAtHome youtube channel: ScienceAtHome - Potential Penguin

How should I move a bowl to make an oscillating ball stop (or push a swing)?

29

To remove energy you must shake the bowl in the same frequency as the rolling ball but with opposite phase. (Just like pushing a swing)

Towards transformational citizen science –an anthropological study of the scientific methods

Assignment – decompose elements of the our scientific approach

The Citizen Science Notebook –a digital tool for scaffolding the scientific method

https://www.scienceathome.org/education/citsci‐research‐tool/

The Citizen Science Notebook –a digital tool for scaffolding the scientific methodStudent publication framework


Use-case 1: Big ideas and socio-scientific issues (current test w/ 10th grade)

Why did the western culture become dominant? Natural scientific argumentation (logicalreasoning) and communication(referencing a critically evaluatinglitterature)The students will based on selected readingmaterial propose hypotheses such as becauseof:• Difference between races• Differences in natural environments• Better navigational techniquesIn the CSNB, the students will then addsupportive or contradicting arguments and thereby build a connected tree of knowledgeand arguments. In this way, the students willboth identify ultimate and proximatesubstantiated hypotheses and falsifiedhypotheses.

Why did mankind become ruler of the earth?


Usecase: • Define an open question• Curate the relevant litterature

Units of information:Just one fact/argument per publication

Final report

Possible division of roles in the class: • Compiler/editor• Fact and logical

argument finderFinal report will be evaluated based on the quality of the argumentation AND the correct and diverse identification of sources of knowledge (bonus points when the entire team contributes )

Final report

Falsified hypotheses

Post-reflektion: Observing the birth and death of hypothesesAfter the session, we willvisualize the generatedknowledge tree and therebyfoster meta-reflection over the used methodology.


Use-case 2: Citizen science

• The students generate new scientific results• Share them so others can extend upon them• Write “publications” with reflections on strategies

on the basis of quantum or classical physics phenomena such as movement of a pendulum or a ball in a bowl.


Use-case 3: computational thinking and simulations

Eksempel fra Silkeborg Gymnasium: Potentialer i Web 2.0

Arthur Hjort

Mathematics is (primarily) the language of the STEM topicsInformatics is (soon) the language of all topics

Mitchell Resnick (MIT): Coding is a new way for people to share ideas with the

world– an extended form of writing.


Cognitive profiling, phase 2: Measuring 21st century skillsOur philosophy: Start to assess what we value instead of valuing what we can assess!!!

We start with creativity

See a review of frameworks herehttp://exploresel.gse.harvard.edu/

Open Ended Problem Solving

CREATIVE FORAGING GAME

OPEN CREATIVITY

PIC BREEDER

CRYSTAL CROP FEVER

ALIEN GAME

QUANTUM MOVES

Players move one square at a time forming various ‘interesting’ Decominoes.

Players rotate two rectangles which form a shape where they overlap.

Players pick images which they want to evolve. Other players further evolve those images.

Players use various search strategies to maximize points in a 2D complex landscape.

Players decode an alien’s 8 tile long message by changing up to two binary tiles at a time.

Players transcend time and space in a game that exists in multiple planes of reality.

Artificial tasks, Low Complexity

Real world, High Complexity

Comparing creativity to complex problem solving- can artist learn from scientists and AI and vice versa?

ScienceAtHome team

GAME DEVELOPMENT

DESIGN AND SOUND

EDCUATION

QUANTUM EXPERIMENT

QUANTUM THEORY

SCHOOL OF BUSINESS AND SOCIAL SCIENCES

COGNITIVE PSYCHOLOGY

INNOVATION AND ENTREPRENEURSHIP

ORGANIZATION

DATA OUTREACH

Anders Lund

Emil Stephansen

Emil Blæsbjerg Rahbek

Florian Korsakissok

Jacob Laurits BesenbacherKjeldsen

Kenneth Mathias Pedersen

Christian Poulsen

Asbjørn Gjøderum-Svenningsen

Nikolaj Lund Sloth

Jacob Sherson Ulrik K. RoosRobert Heck Jens Schultz

LaustsenOttóElíasson

Carrie Weidner

Jonathan Satchell

Peter Kjærgaard

Beata Biskupicova

Plamen Petkov Jeppe Grøn

Janet Rafner

ShaeemaZaman

Stefan Vidovic

Louise Kindt

Janni Ohrt Mads Kock

Miroslav Gajdacz

Christian Bach

Patricia Toth

Mariam TizarHansson

Jonas Korshøj

Katrine Nymann

Carsten Bergenholtz

Oana Vuculescu

Carlos Díaz Rajiv Basaiawmoit

Jens Jakob Sørensen

Jesper HasseriisMohr Jensen

Mogens Dalgaard

Jakub Czarkowski

CollaboratorsAarhus UniversityArtsCOMMUNICATION AND CULTURE

DANISH SCHOOL OF EDUCATION

BEHAVIORAL ECONOMICS

MANAGEMENTPOLITICAL SCIENCE

ECONOMICS AND BUSINESS

Business and Social Science

Science

PHYSICSDEAN’S OFFICE

ENGI-NEERING

BIO-INFORMATICS

MOLECULAR BIOLOGY AND GENETICS

PSYCHOLOGY AND PHILOSOPHY

Advanced Studies HealthCLINICAL MEDICINE

BIO-MEDICINE PSYCHOLOGY

PHYSICS, MATH, ENGINEERING

PSYCHOLOGY AND BEHAVIOURAL SCIENCES

CORPORATE INOVATION NEUROSCIENC

E

ExternalDIDACTICS; EDUCATION, LEARNING SCIENCES

SAMRT CITIESSCIENCE ENGAGEMENT

MEDICINE

AI AND MACHINE LEARNING

Morten Christiansen,computationalneuroscience

Peter Dalsgaard,Center for Digital Creativity

Michael Biskjær,Center for Digital Creativity

Clemens Klokmose,Centre for Digital Creativity

Martin Brynskov,Aarhus Smart Cities

Mikkel Wallentin,linguistics

Kristian Tylen,languagecomprehension

Andreas Liberoth,game psychology

Niels Bonderup Dohn,student motivation

Andrew Mao, Microsoft

Carsten Bergenholtz,individualand collectiveproblem solving

Oana Vuculescu,individual and collectiveproblem solving

Anders Villadsen, sector of employment and cognitiveabilities

Michael Bang PetersenPsychologyof Democracy

Alexander Koch,behaviouraleconomics

Klaus Mølmer,quantum physics

Rajiv Basaiawmiotentrepreneurship, team formation

Bo Gervang,fluid mechanics

Christian Storm Pedersen,machine learning

Christian Storm Pedersen,machine learning

Micah Allen, computationalneuroscience

Nikolai Ladegaard,clinicalpsychology

Kim Mathiasenclinicalpsychology

Anders Børglum,genetic and cognitive profiling

Jana Jarecki, Max Planck, Basel

Zoran Grujic,University of Virginia

Jakob Andreas Bærentzen, DTU

Ali Amidi,cognitive ageing

Karim Lakhani MarocIansti, HBS

Ralph Hertwig,Max Plank, Berlin

Sebastian Risi,ITU

Gabriel Guerra

Arthur Juliani,Oregon U

Rupert YoungPerceptualRobots

Lars Kai Hansen, DTU

Carol O’Donnell, SmithsonianScience Education Center

Rikke Schmidt Kjærgaard

Francois Tadalli, CRI

Ariel Linder, CRI

ManjulaDissanayake, Educate Lanka

AdriënneHeijnen, AUSC

Julien Bobroff, PhysicsReimagined

Martin Brynskov, AUSC

Jacob Thyssen,Copenhagen UniversityHospital

Funders

Documents

Scalable, game-based citizen science: Developing hybrid … · 2019-12-20 · Scalable, game-based citizen science: Developing hybrid intelligence and fostering a population thinking