The Rise of Crowd Computing (December 2015)
2 likes1,273 views
Crowd computing is rising with two waves - the first using crowds to label large amounts of data for artificial intelligence applications. The second wave delivers applications that go beyond AI abilities by incorporating human computation. Open problems remain around ensuring high quality outputs, task design, understanding the worker context and experience, and addressing ethics concerns around opaque platforms and working conditions. The future holds potential for empowering crowd work but also risks like digital sweatshops if worker freedoms and conditions are not considered.
![The Early Days
• S
• Artificial Intelligence, With Help From the Humans.
– J. Pontin. NY Times, March 25, 2007
• Is Amazon's Mechanical Turk a Failure? April 9, 2007
– “As of this writing, there are [only] 128 Human Intelligence
Tasks available via the Mechanical Turk task page.”
• Su et al., WWW 2007: “a web-based human data
collection system that we [call] ‘System M’ ”
11](https://image.slidesharecdn.com/qatar-dec2015-151223131011/85/The-Rise-of-Crowd-Computing-December-2015-11-320.jpg)
![Why Eytan Adar hates MTurk Research
(CHI 2011 CHC Workshop)
• Overly-narrow research focus on MTurk
– Distinguish general vs. platform-specific problems
– Distinguish research vs. industry concerns
• Should researchers really focus on…
– “...writing the user’s manual for MTurk ...”?
– “…struggl[ing] against the limits of the platform...”?
“…by rewarding quick demonstrations of the tool’s
use, we fail to attain a deeper understanding of the
problems to which it is applied…”](https://image.slidesharecdn.com/qatar-dec2015-151223131011/85/The-Rise-of-Crowd-Computing-December-2015-20-320.jpg)
The Rise of Crowd Computing (December 2015)
- 1. The Rise of Crowd Computing Matt Lease School of Information @mattlease University of Texas at Austin [email protected] Slides: slideshare.net/mattlease
- 2. “The place where people & technology meet” ~ Wobbrock et al., 2009 “iSchools” now exist at 65 universities around the world www.ischools.org What’s an Information School? 2
- 3. Roadmap • Motivation from Artificial Intelligence (AI) – Need for Plentiful Labeled Data – Need for Capabilities Beyond What AI Can Deliver • The Rise of Crowd Computing – 1st Wave: Crowd-based data labeling • Mechanical Turk & Beyond – 2nd Wave: Crowd-based Human Computation • Delivering beyond state-of-the-art AI applications today • Open Problems 3
- 4. Motivation 1: AI effectiveness is often limited by training data size Problem: creating labeled data is expensive! Banko and Brill (2001)
- 5. Motivation 2: What do we do when state-of-the-art AI isn’t good enough?
- 7. Crowdsourcing • Jeff Howe. Wired, June 2006. • Take a job traditionally performed by a known agent (often an employee) • Outsource it to an undefined, generally large group of people via an open call 7
- 9. • Marketplace for paid crowd work (“micro-tasks”) – Created in 2005 (remains in “beta” today) • On-demand, scalable, 24/7 global workforce • API lets human labor be integrated into software – “You’ve heard of software-as-a-service. Now this is human-as-a-service.” Amazon Mechanical Turk (MTurk)
- 10. Find Jim Gray (February 2007)
- 11. The Early Days • S • Artificial Intelligence, With Help From the Humans. – J. Pontin. NY Times, March 25, 2007 • Is Amazon's Mechanical Turk a Failure? April 9, 2007 – “As of this writing, there are [only] 128 Human Intelligence Tasks available via the Mechanical Turk task page.” • Su et al., WWW 2007: “a web-based human data collection system that we [call] ‘System M’ ” 11
- 12. The 1st Wave of Crowd Computing: Data Collection via Crowdsourcing @mattlease
- 13. 2008 MTurk “Discovery” sparks rush for “gold” labels across areas • Alonso et al., SIGIR Forum (Information Retrieval) • Kittur et al., CHI (Human-Computer Interaction) • Sorokin and Forsythe, CVPR (Computer Vision) Snow et al, EMNLP (NLP) • Annotating human language • 22,000 labels for only US $26 • Crowd’s consensus labels can replace traditional expert labels
- 14. NLP Example – Dialect Identification 14 See work by Chris Callison-Burch. Interface: http://ugrad.cs.jhu.edu/~sjeblee/arabic-classification-plus.shtml
- 15. August 12, 2012 15
- 16. Social & Behavioral Sciences • A Guide to Behavioral Experiments on Mechanical Turk – W. Mason and S. Suri (2010). SSRN online. • Crowdsourcing for Human Subjects Research – L. Schmidt (CrowdConf 2010) • Crowdsourcing Content Analysis for Behavioral Research: Insights from Mechanical Turk – Conley & Tosti-Kharas (2010). Academy of Management • Amazon's Mechanical Turk : A New Source of Inexpensive, Yet High-Quality, Data? – M. Buhrmester et al. (2011). Perspectives… 6(1):3-5. – see also: Amazon Mechanical Turk Guide for Social Scientists 16
- 18. Crowdsourcing ≠ Mechanical Turk!
- 19. Many Platforms for Paid Crowd Work And More! JobBoy, microWorkers, MiniFreelance, MiniJobz, MinuteWorkers, MyEasyTask, OpTask, ShortTask, SimpleWorkers
- 20. Why Eytan Adar hates MTurk Research (CHI 2011 CHC Workshop) • Overly-narrow research focus on MTurk – Distinguish general vs. platform-specific problems – Distinguish research vs. industry concerns • Should researchers really focus on… – “...writing the user’s manual for MTurk ...”? – “…struggl[ing] against the limits of the platform...”? “…by rewarding quick demonstrations of the tool’s use, we fail to attain a deeper understanding of the problems to which it is applied…”
- 21. Beyond Mechanical Turk: An Analysis of Paid Crowd Work Platforms Vakharia and Lease, iConference 2015 Qualitative assessment of 7 platforms for paid crowd work
- 22. 22 Crowdsourcing Transcription Beyond Mechanical Turk With Haofeng Zhou & Denys Baskov HCOMP 2013 Speech Workshop
- 23. A lot of volunteer, Citizen Science too!
- 24. Citizen science in the early internet (2000-2001) 24 www.nasaclickworkers.com
- 25. Zooniverse 25
- 26. Crowd4U (Another Open Platform) 26
- 27. 27
- 28. ESP Game (Gamification) L. Von Ahn and L. Dabbish (2004) 28
- 29. reCaptcha (Repurpose Existing Activity) von Ahn et al. (2008). In Science. 29
- 31. Tracking Sentiment (Access Resource) Brew et al., PAIS 2010 • Volunteer-crowd – Work in exchange for access to rich content • Never-ending Learning – Continual model updates as what is relevant vs. not changes over time 31
- 32. Beat the Machine (Earn Money) 32
- 33. The 2nd Wave of Crowd Computing: Human Computation @mattlease
- 34. What is a Computer? 34
- 35. Princeton University Press, 2005 • What was old is new • Crowdsourcing: A New Branch of Computer Science – D.A. Grier, IEEE President • Tabulating the heavens: computing the Nautical Almanac in 18th-century England – M. Croarken (2003) 35
- 36. The Mechanical Turk The original, constructed and unveiled in 1770 by Wolfgang von Kempelen (1734–1804) 36 J. Pontin. Artificial Intelligence, With Help From the Humans. New York Times (March 25, 2007)
- 37. The Human Processing Unit (HPU) Davis et al. (2010) HPU 37
- 38. ACM Queue, May 2006 38 “Software developers with innovative ideas for businesses and technologies are constrained by the limits of artificial intelligence… If software developers could programmatically access and incorporate human intelligence into their applications, a whole new class of innovative businesses and applications would be possible. This is the goal of Amazon Mechanical Turk… people are freer to innovate because they can now imbue software with real human intelligence.”
- 39. Creating A New Class of Intelligent Applications 39@mattlease
- 41. PlateMate (Noronha et al., UIST’10) 41
- 42. Ethics Checking: The Next Frontier? • Mark Johnson’s address at ACL 2003 – Transcript in Conduit 12(2) 2003 • Think how useful a little “ethics checker and corrector” program integrated into a word processor could be! 42
- 43. Soylent: A Word Processor with a Crowd Inside • Bernstein et al., UIST 2010 43
- 44. fold.it S. Cooper et al. (2010) Alice G. Walton. Online Gamers Help Solve Mystery of Critical AIDS Virus Enzyme. The Atlantic, October 8, 2011. 44
- 45. MonoTrans: Translation by monolingual speakers 45 • Bederson et al., 2010 • See also: Morita & Ishidi, ACM IUI 2009
- 46. VizWiz aaaaaaaa Bingham et al. (UIST 2010) 46Matt Lease - [email protected]
- 47. Zensors Laput et al., CSCW 2015 47
- 48. Flock: Hybrid Crowd-Machine Learning Classifiers (2015) 48
- 49. @mattlease
- 50. HCOMP 2013 Panel Anand Kulkarni: “How do we dramatically reduce the complexity of getting work done with the crowd?” Greg Little: How can we post a task and with 98% confidence know we’ll get a quality result? 50
- 51. How to ensure data quality? • Research on statistical quality control methods – Online vs. offline, feature-based vs. content-agnostic – Worker calibration, noise vs. bias, weighted voting • Human factors matter too! – Instructions, design, interface, interaction – Names, relationship, reputation Fair pay, hourly vs. per-task, recognition, advancement 51
- 52. 52 SQUARE: A Benchmark for Research on Computing Crowd Consensus @HCOMP’13 ir.ischool.utexas.edu/square (open source)
- 53. Is everyone just lazy, stupid, or deceitful?!? Many published papers seem to suggest this • “Cheaters” • “Fraudsters” • “Lazy Turkers” • “Scammers” • “Spammers” But why can’t the workers just get it right to begin with? 53
- 54. What is our responsibility? • Ill-defined/incomplete/ambiguous/subjective task? • Confusing, difficult, or unusable interface? • Incomplete or unclear instructions? • Insufficient or unhelpful examples given? • Gold standard with low or unknown inter-assessor agreement (i.e. measurement error in assessing response quality)? • Task design matters! (garbage in = garbage out) – Report it for review, completeness, & reproducibility 54
- 55. Task Decomposition & Workflow Design 55
- 56. What about context? “Best practices” for crowdsourcing design often minimizes context to maximize task efficiency – e.g. “Are these pictures of the same person?” 56
- 57. Importance of Informed Consent + Potential for Oppression, Crime, & War Jonathan Zittrain, Minds for Sale 57
- 58. Who are the workers? • A. Baio, November 2008. The Faces of Mechanical Turk. • P. Ipeirotis. March 2010. The New Demographics of Mechanical Turk • J. Ross, et al. Who are the Crowdworkers? CHI 2010. 58
- 59. What about ethics? • Silberman, Irani, and Ross (2010) – “How should we… conceptualize the role of these people who we ask to power our computing?” • Irani and Silberman (2013) – “…by hiding workers behind web forms and APIs… employers see themselves as builders of innovative technologies, rather than… unconcerned with working conditions… redirecting focus to the innovation of human computation as a field of technological achievement.” • Fort, Adda, and Cohen (2011) – “…opportunities for our community to deliberately value ethics above cost savings.” 59
- 60. Digital Dirty Jobs • The Googler who Looked at the Worst of the Internet • Policing the Web’s Lurid Precincts • Facebook content moderation • The dirty job of keeping Facebook clean • Even linguistic annotators report stress & nightmares from reading news articles! 60
- 61. What about freedom? • Crowdsourcing vision: empowering freedom – work whenever you want for whomever you want • Risk: people being compelled to perform work – Digital sweat shops? Digital slaves? – Chinese Prisoners used for online gold farming – We really don’t know (and need to learn more…) – Traction? Human Trafficking at MSR Summit’12 61
- 62. The Future of Crowd Work Paper @ CSCW 2013 by Kittur, Nickerson, Bernstein, Gerber, Shaw, Zimmerman, Lease, and Horton 62
- 63. Conclusion • Crowdsourcing is quickly transforming practice in industry and academia via greater efficiency • Human Computation enables a new design space for applications, augmenting state-of-the- art AI with human computation to offer new capabilities and user experiences • With people at the center of this new computing paradigm, important research questions span both technological and social/societal challenges
- 64. Matt Lease - [email protected] - @mattlease Thank You! ir.ischool.utexas.edu/crowd Slides: slideshare.net/mattlease