Diese Präsentation wurde erfolgreich gemeldet.
Wir verwenden Ihre LinkedIn Profilangaben und Informationen zu Ihren Aktivitäten, um Anzeigen zu personalisieren und Ihnen relevantere Inhalte anzuzeigen. Sie können Ihre Anzeigeneinstellungen jederzeit ändern.
AI Standards
Issues on Artificial Intelligence
and Future
November 16, 2017
Dr. Seungyun Lee <syl@etri.re.kr>
人工知能と未来の課題
AI
9
T a b l e o f C o n t e n t s
Prologue
Epiloge
AtaGlance,AI
MajorIssuesonAI
StandardsPerspectives
AI
10
Prologue
11
<Movie Clip>
AI
12
History and Issues
At a Glance, Artificial Intelligence
I am not an expert on Artificial Intelligence …
13
AI
14
Artificial Intelligence
not a new term …
AI
15
Artificial Intelligence in 1950
AI
16
History of AI
Source> Wikipedia & IBM’s developerWorks, June 2017
Golden Age (1965~1974) Dark Age (74-80)
Boom up (8...
AI
17
Limitation of Search, Pattern Matching, …
The reason for failure:
Too Big positive expectation …
AI
18
Intelligence vs. Expression
How to realize AI ?
AI
19
RDF(a), GRDDL, POWDER,
Linked Data, …
Semantic Web
Technology
Since 2003
Feasible approaches
AI
20
Make it possible AI again !
Recent revolution & emerging technology :
Learning
Technology
AI
21
https://www.linkedin.com/pulse/artificial-intelligence-deep-learning-trends-insights-anuj-saxena
AI
22
https://blogs.nvidia.com/blog/2016/07/29/whats-difference-artificial-intelligence-machine-learning-deep-learning-ai/
AI
23
What's the Difference between Robotics and Artificial Intelligence?
source> https://blog.robotiq.com/whats-the-differe...
AI
24
Is AIBO Robot or AI?
AI
25
https://www.ibm.com/developerworks/library/cc-beginner-guide-
machine-learning-ai-cognitive/index.html
A beginner's ...
AI
26
Ball, a cognitive scientist, began exploring the gap between how our brains
interpret information and how computers ...
AI
27
Need to be considered seriously
Major Issues on Artificial Intelligence
AI
28
Spectators watch a broadcast of the final, decisive game in the rematch
between Garry Kasparov and the IBM computer D...
AI
29
Artificial Intelligence in 2016
AI
30
Google's Go-playing AI still undefeated
with victory over world number one…
http://www.theguardian.com/
AI
31
“Artificial intelligence would be the ultimate version of Google…It would
understand exactly what you wanted, and it...
AI
Computational

Power

32
Computational Power vs. Intelligence Power
What is the Key point in AI ?
Human Level
Intellige...
AI
Availability

33
Availability vs. Responsibility
What is the Key point in AI ?
Expectated Level
Responsibility

(Safety...
AI
34
“The development of full artificial intelligence
could spell the end of the human race…”
Stephen Hawking
AI
35
Artificial Intelligence is like Summoning the Demon
Elon Musk
AI
36
Does AI be Regulated?
AI
37
Autonomous Vehicle - AI application
AI
some of the biggest examples
38
Top 10 AI failures of 2016
source> https://www.techrepublic.com/article/top-10-ai-failu...
AI
39
Microsoft forced to apologize after epic chatbot fail
March 2016, https://blogs.microsoft.com/blog/2016/03/25/learni...
AI
40
As evidenced by these examples, AI systems are deeply prone to bias—and
it is critical that machine learning algorit...
AI
41
Why do we need AI standards?
Standards Perspectives
AI
42
JTC 1
Dejure De-FactoRegional
30 YEARS AGO
43
AI
44
ISO/IEC JTC 1 N 365
AI
45
ISO/IEC JTC 1 N 392
AI
46
ISO/IEC JTC 1 N 374
AIAI in ISO/IEC JTC 1
JTC 1 has a 20 years history with working on standards in AI
technology area.
• ISO/IEC 2382-28:1995...
AIDefinitions
Artificial intelligence (AI) is intelligence exhibited by machines. In
computer science, the field of AI res...
AI Standards
Why Standardization?
49
AIJustifications
50
Why are standards required ?
AIWhy Standardization?
While AI has been around for many years with active research and development
over the last 10 years...
AIWhy Standardization?
Standards provide requirements, specifications, guidelines, or
characteristics that can be used con...
AI Standards
The Area of Standardization
53
AIThe Area of Standardization
Domains
Software engineering
Performance
Metrics
Safety
Usability
Interoperability
Security
...
AIThe Area of Standardization
Domains
Software engineering
Performance
Metrics
Safety
Usability
Interoperability
Security
...
AI Standards
Related Activities
56
AIRelated Activities
57
AIRelated Activities – IEEE
• The IEEE Standards Association (IEEE-SA) launched the IEEE
Global Initiative for Ethical Con...
AIRelated Activities – ITU
• ITU believes that the Artificial Intelligence (AI) innovation will be central to the
achievem...
AIRelated Activities – ISO & IEC
• ISO and IEC have published the standards for vocabulary on Artificial Intelligence (ISO...
AI Standards
Analysis of Survey Results
61
AISurvey Results
ISO/IEC JTC 1 JAG conducted online survey on AI and AS to JTC 1 entities (for JTC
1/SC Chairs and JTC 1/W...
AISurvey Results
As a result of survey, there were 13 responds in total (response rate is 59%) from
following JTC 1 entiti...
AISurvey Results
64
RelevancetoAI&AS
40%
AISurvey Results
65
Numberofrelevant
activities
56
AISurvey Results
66
Planofdevelopment
onAIandAS
4 7 3
18% 32%
AISurvey Results
67
UrgentareaofStandardization
1.Interoperability
2.Domains
3.Security
4.Safety
5.Ethics
AIRelated Activities – Others
National Strategy on AI:
[China] China to launch national AI strategy
[France] France alread...
AI
69
The Japanese Society for Artificial Intelligence Ethical Guidelines
http://ai-elsi.org/wp-content/uploads/2017/05/JSA...
AI
70
AI
71
Proposed Principles in “AI R&D Guideline” in Japan
http://www.soumu.go.jp/joho_kokusai/g7ict/english/main_content/ai...
AI
72
https://www.ai-gakkai.or.jp/sigconf/
AI Standards
Recommendations to JTC 1
73
AIRecommendations to JTC 1
According to the investigation work on AI and AS, 

it is strongly recommended that 

“JTC 1 wo...
AI Standards
ISO/IEC JTC 1 established a new Subcommittee,
SC 42 for the development of Artificial Intelligence
(2017 JTC 1...
AI Standards
Artificial Intelligence,
there have been a lot of failures in the meantime,
and it will continue…
76
AI
77
We need the right compass
for journeying into future AI world !
AI Standards
AI
78
Epilogue
79
<Movie Clip>
Misuse of technology brings disaster
Standards Make Human being better !
AI
82
Thanks You for your attention
ご清聴ありがとうございました
Nächste SlideShare
Wird geladen in …5
×

Issues on Artificial Intelligence and Future (Standards Perspective)

879 Aufrufe

Veröffentlicht am

Talked in Keio Univ. in November 2017.

Veröffentlicht in: Technologie

Issues on Artificial Intelligence and Future (Standards Perspective)

  1. 1. AI Standards Issues on Artificial Intelligence and Future November 16, 2017 Dr. Seungyun Lee <syl@etri.re.kr> 人工知能と未来の課題
  2. 2. AI 9 T a b l e o f C o n t e n t s Prologue Epiloge AtaGlance,AI MajorIssuesonAI StandardsPerspectives
  3. 3. AI 10 Prologue
  4. 4. 11 <Movie Clip>
  5. 5. AI 12 History and Issues At a Glance, Artificial Intelligence
  6. 6. I am not an expert on Artificial Intelligence … 13
  7. 7. AI 14 Artificial Intelligence not a new term …
  8. 8. AI 15 Artificial Intelligence in 1950
  9. 9. AI 16 History of AI Source> Wikipedia & IBM’s developerWorks, June 2017 Golden Age (1965~1974) Dark Age (74-80) Boom up (80-87) Dark Age (87-93) Challenging (94-10) New Golden Age? AI WinterExpert System Neural Network ? Stop Investment The 4th Industrial Revolution Watson, AlphaGo Searching to Learning
  10. 10. AI 17 Limitation of Search, Pattern Matching, … The reason for failure: Too Big positive expectation …
  11. 11. AI 18 Intelligence vs. Expression How to realize AI ?
  12. 12. AI 19 RDF(a), GRDDL, POWDER, Linked Data, … Semantic Web Technology Since 2003 Feasible approaches
  13. 13. AI 20 Make it possible AI again ! Recent revolution & emerging technology : Learning Technology
  14. 14. AI 21 https://www.linkedin.com/pulse/artificial-intelligence-deep-learning-trends-insights-anuj-saxena
  15. 15. AI 22 https://blogs.nvidia.com/blog/2016/07/29/whats-difference-artificial-intelligence-machine-learning-deep-learning-ai/
  16. 16. AI 23 What's the Difference between Robotics and Artificial Intelligence? source> https://blog.robotiq.com/whats-the-difference-between-robotics-and-artificial-intelligence
  17. 17. AI 24 Is AIBO Robot or AI?
  18. 18. AI 25 https://www.ibm.com/developerworks/library/cc-beginner-guide- machine-learning-ai-cognitive/index.html A beginner's guide to artificial intelligence, machine learning, and cognitive computing M. Tim Jones Published on June 01, 2017 Recommend to read:
  19. 19. AI 26 Ball, a cognitive scientist, began exploring the gap between how our brains interpret information and how computers work in 1983. Recommend to read:
  20. 20. AI 27 Need to be considered seriously Major Issues on Artificial Intelligence
  21. 21. AI 28 Spectators watch a broadcast of the final, decisive game in the rematch between Garry Kasparov and the IBM computer Deep Blue. (1977)
  22. 22. AI 29 Artificial Intelligence in 2016
  23. 23. AI 30 Google's Go-playing AI still undefeated with victory over world number one… http://www.theguardian.com/
  24. 24. AI 31 “Artificial intelligence would be the ultimate version of Google…It would understand exactly what you wanted, and it would give you the right thing.” Larry Page
  25. 25. AI Computational Power 32 Computational Power vs. Intelligence Power What is the Key point in AI ? Human Level Intelligence/Perception Power
  26. 26. AI Availability 33 Availability vs. Responsibility What is the Key point in AI ? Expectated Level Responsibility (Safety, Security…)
  27. 27. AI 34 “The development of full artificial intelligence could spell the end of the human race…” Stephen Hawking
  28. 28. AI 35 Artificial Intelligence is like Summoning the Demon Elon Musk
  29. 29. AI 36 Does AI be Regulated?
  30. 30. AI 37 Autonomous Vehicle - AI application
  31. 31. AI some of the biggest examples 38 Top 10 AI failures of 2016 source> https://www.techrepublic.com/article/top-10-ai-failures-of-2016/ 1. AI built to predict future crime was racist 2. Non-player characters in a video game crafted weapons beyond creator's plans 3. Robot injured a child 4. Fatality in Tesla Autopilot mode 5. Microsoft's chatbot Tay utters racist, sexist, homophobic slurs 6. AI-judged beauty contest is racist 7. Pokémon Go keeps game-players in white neighborhoods 8. Google's AI, AlphaGo, loses game 4 of Go to Lee Sedol 9. Chinese facial recognition study predicts convicts but shows bias 10.Insurance company uses Facebook data to issue rates, shows bias *Racist = 人種差別主義者
  32. 32. AI 39 Microsoft forced to apologize after epic chatbot fail March 2016, https://blogs.microsoft.com/blog/2016/03/25/learning-tays-introduction/ … Although we had prepared for many types of abuses of the system, we had made a critical oversight for this specific attack. As a result, Tay tweeted wildly inappropriate and reprehensible words and images. …
  33. 33. AI 40 As evidenced by these examples, AI systems are deeply prone to bias—and it is critical that machine learning algorithms train on diverse sets of data in order to prevent it. As AI increases its capabilities, ensuring proper checks, diverse data, and ethical standards for research is of utmost importance. これらの例からもわかるように、AIシステムはバイアスが深くなりがちです。 機械学習アルゴリズムは、データを保護するためにさまざまなデータセット を学習することが重要です。 AIの能力が向上するにつれて、適切なチェック、 多様なデータ、研究の倫理基準を保証することが最も重要です。 Top 10 AI failures of 2016 some of the biggest examples source> https://www.techrepublic.com/article/top-10-ai-failures-of-2016/
  34. 34. AI 41 Why do we need AI standards? Standards Perspectives
  35. 35. AI 42 JTC 1 Dejure De-FactoRegional
  36. 36. 30 YEARS AGO 43
  37. 37. AI 44 ISO/IEC JTC 1 N 365
  38. 38. AI 45 ISO/IEC JTC 1 N 392
  39. 39. AI 46 ISO/IEC JTC 1 N 374
  40. 40. AIAI in ISO/IEC JTC 1 JTC 1 has a 20 years history with working on standards in AI technology area. • ISO/IEC 2382-28:1995 Information technology – Vocabulary – Part 28: Artificial intelligence – Basic concepts and expert systems • ISO/IEC 2382-31: Artificial intelligence – Machine learning • ISO/IEC 2382-34: Artificial intelligence – Neural networks All those specifications have been consolidated and it have been revised • ISO/IEC 2382:2015: Information technology -- Vocabulary 47
  41. 41. AIDefinitions Artificial intelligence (AI) is intelligence exhibited by machines. In computer science, the field of AI research defines itself as the study of "intelligent agents” Artificial intelligence is technology that appears to emulate human performance typically by learning, coming to its own conclusions, appearing to understand complex content, engaging in natural dialogs with people. The capability of a functional unit to perform functions that are generally associated with human intelligence such as reasoning and learning. (ISO/ IEC 2382-28:1995) 48 Needs to be re-defined again ?
  42. 42. AI Standards Why Standardization? 49
  43. 43. AIJustifications 50 Why are standards required ?
  44. 44. AIWhy Standardization? While AI has been around for many years with active research and development over the last 10 years, we have seen an increase in the number of products and services implementing AI effectively, which has led to an increase of interest by new and/or existing organizations to look into standardization of AI. Countries and regulators have begun taking an interest in many aspects of AI with an increase in the application of AI into a wider set of domains and the possibility of new regulation. There is a need for standards to facilitate communication about Artificial Intelligence, to enable acceptance of approaches and concepts that support future development and to address the interests and concerns of governments and society. 51
  45. 45. AIWhy Standardization? Standards provide requirements, specifications, guidelines, or characteristics that can be used consistently to ensure that AI technologies meet critical objectives for functionality and interoperability, and that they perform reliably and safely. The development of standards must be hastened to keep pace with the rapidly evolving capabilities and expanding domains of AI applications. Furthermore, how timely development of global standards will be important to facilitate wider global adoption of AI technologies and in order to truly tap into AI’s potential. Adoption of standards brings credibility to technology advancements and facilitates an expanded interoperable marketplace. 52
  46. 46. AI Standards The Area of Standardization 53
  47. 47. AIThe Area of Standardization Domains Software engineering Performance Metrics Safety Usability Interoperability Security Privacy Traceability Risk Analysis Ethics 54
  48. 48. AIThe Area of Standardization Domains Software engineering Performance Metrics Safety Usability Interoperability Security Privacy Traceability Risk Analysis Ethics 55 Source of Regulations
  49. 49. AI Standards Related Activities 56
  50. 50. AIRelated Activities 57
  51. 51. AIRelated Activities – IEEE • The IEEE Standards Association (IEEE-SA) launched the IEEE Global Initiative for Ethical Considerations in Artificial Intelligence and Autonomous Systems (AI/AS). (April 2016) ★ Two primary deliverables - Ethically Aligned Design (EAD), Versions 1 & 2 • The IEEE also launched the IEEE Symbiotic Autonomous Systems (SAS) Initiative aims to following goals (July 2017): 1. Take the lead in developing the new field of Symbiotic Systems Science 2. Foster interdisciplinary technology deployments that take into account Ethical, Legal, and Societal considerations 3. Promote human-centric economic growth 58
  52. 52. AIRelated Activities – ITU • ITU believes that the Artificial Intelligence (AI) innovation will be central to the achievement of the United Nations' Sustainable Development Goals (SDGs) and will help in solving humanity's grand challenges. • As part of emerging ICT trends programme, ITU made its organize briefings on the topic of “Artificial Intelligence (AI) for Good”: • ITU Briefing on "Artificial Intelligence (AI)
 for Good", 59 • ITU has started the development of AI standards since 2016 ① ITU-T Y.AI4SC, Artificial Intelligence and IoT ② ITU-T Y.qos-ml, Requirements of machine
 learning based QoS assurance for IMT-2020
  53. 53. AIRelated Activities – ISO & IEC • ISO and IEC have published the standards for vocabulary on Artificial Intelligence (ISO/ IEC 2382:2015, formerly ISO/IEC 2382:1995) which will be basis of AI standards. • Recent Activities ★ New initiatives in JTC 1/SC 7 (SG Autonomous Systems and Ambient Intelligence Environment Engineering), SC 34 and SC 36 (under the AI investigation), SC 40 (Governance of AI/AS), and WG 9 (Data analytics). ★ IEC Market Strategy Board (MSB) started to develop a new white paper "Application of Artificial Intelligence (AI) across Vertical Industries” ★ In JTC 1, NWIP ballot on “AI Concepts and Terminology” (NP 22989) (July 2017) ★ USNB has proposed JTC 1 the establishment of a Systems Integration entity in the form of a new Subcommittee on the Artificial Intelligence. 60
  54. 54. AI Standards Analysis of Survey Results 61
  55. 55. AISurvey Results ISO/IEC JTC 1 JAG conducted online survey on AI and AS to JTC 1 entities (for JTC 1/SC Chairs and JTC 1/WG Conveners) during 29 July ~ 31 August 2017. Online Survey: https://www.surveymonkey.com/r/7GD52SP The purpose of survey is follows: 1. Identify the related activities on AI and AS (include ongoing projects) 2. Identify priority area on AI and AS standards 3. Identify the future work items on AI and AS 4. Other considerations 62
  56. 56. AISurvey Results As a result of survey, there were 13 responds in total (response rate is 59%) from following JTC 1 entities: SC 7 Software and systems engineering SC 17 Cards and personal identification SC 22 Programming languages, their environments and system software interfaces SC 24 Computer graphics, image processing and environmental data representation SC 25 Interconnection of information technology equipment SC 27 IT Security techniques SC 28 Office Equipment SC 29 Coding of audio, picture, multimedia and hypermedia information SC 36 Information technology for learning, education and training SC 37 Biometrics SC 40 IT Service Management and IT Governance SC 41 Internet of Things and related technologies WG 9 Big Data 63
  57. 57. AISurvey Results 64 RelevancetoAI&AS 40%
  58. 58. AISurvey Results 65 Numberofrelevant activities 56
  59. 59. AISurvey Results 66 Planofdevelopment onAIandAS 4 7 3 18% 32%
  60. 60. AISurvey Results 67 UrgentareaofStandardization 1.Interoperability 2.Domains 3.Security 4.Safety 5.Ethics
  61. 61. AIRelated Activities – Others National Strategy on AI: [China] China to launch national AI strategy [France] France already has one of the strongest AI [Japan] Japan’s Role in Establishing Standards for Artificial Intelligence Development [UK] £17 million boost for the UK's booming artificial intelligence sector [US] The National Artificial Intelligence Research and Development Strategy Industry Collaboration: Partnership on AI, MILA, etc. Open Source on AI: [TensorFlow] https://www.tensorflow.org/ [CaffeOnSpark] https://github.com/yahoo/CaffeOnSpark [Cortana] https://www.microsoft.com/en-us/windows/cortana [OpenAI] Open Source AI API: http://openai.sourceforge.net/docs.html [H2O.ai] bringing AI to enterprises: https://www.h2o.ai/gartner-magic-quadrant/ 68
  62. 62. AI 69 The Japanese Society for Artificial Intelligence Ethical Guidelines http://ai-elsi.org/wp-content/uploads/2017/05/JSAI-Ethical-Guidelines-1.pdf
  63. 63. AI 70
  64. 64. AI 71 Proposed Principles in “AI R&D Guideline” in Japan http://www.soumu.go.jp/joho_kokusai/g7ict/english/main_content/ai.pdf
  65. 65. AI 72 https://www.ai-gakkai.or.jp/sigconf/
  66. 66. AI Standards Recommendations to JTC 1 73
  67. 67. AIRecommendations to JTC 1 According to the investigation work on AI and AS, 
 it is strongly recommended that 
 “JTC 1 would urgently start the activity on the development of its standardization”. 74
  68. 68. AI Standards ISO/IEC JTC 1 established a new Subcommittee, SC 42 for the development of Artificial Intelligence (2017 JTC 1 Plenary in Vladivostok, Russia) 75 • Chair: USA (Vice-chair: China) • Secretariat - USA
  69. 69. AI Standards Artificial Intelligence, there have been a lot of failures in the meantime, and it will continue… 76
  70. 70. AI 77 We need the right compass for journeying into future AI world ! AI Standards
  71. 71. AI 78 Epilogue
  72. 72. 79 <Movie Clip>
  73. 73. Misuse of technology brings disaster
  74. 74. Standards Make Human being better !
  75. 75. AI 82 Thanks You for your attention ご清聴ありがとうございました

×