Seal of Good Local Governance (SGLG) 2024Final.pptx
Trl jaist 20180304 v6
1. Jim Spohrer (IBM)
IBM Research - Tokyo, Friday March 2 2018
http://www.slideshare.net/spohrer/trl-jaist-20180304-v6
3/4/2018 1
Preparing for Our Future
with Opentech AI
6. Measuring AI Progress (MAP)
AI Progress on Open Leaderboards - Benchmark Roadmap
Perceive World Develop Cognition Build Relationships Fill Roles
Pattern
recognition
Video
understanding
Memory Reasoning Social
interactions
Fluent
conversation
Assistant &
Collaborator
Coach &
Mediator
Speech Actions Declarative Deduction Scripts Speech Acts Tasks Institutions
Chime Thumos SQuAD SAT ROC Story ConvAI
Images Context Episodic Induction Plans Intentions Summarizatio
n
Values
ImageNet VQA DSTC RALI General-AI
Translation Narration Dynamic Abductive Goals Cultures Debate Negotiation
WMT DeepVideo Alexa Prize ICCMA AT
Learning from Labeled Training Data and Searching (Optimization)
Learning by Watching and Reading (Education)
Learning by Doing and being Responsible (Exploration)
2015 2018 2021 2024 2027 2030 2033 2036
3/4/2018 (c) IBM 2017, Cognitive Opentech Group 6
Which experts would be really surprised if it takes less time… and which experts really surprised if it takes longer?
Approx.
Year
Human
Level ->
8. Every 20 years, compute costs are down
by 1000x
• Cost of Digital Workers
– Moore’s Law can be thought of as
lowering costs by a factor of a…
• Thousand times lower
in 20 years
• Million times lower
in 40 years
• Billion times lower
in 60 years
• Smarter Tools (Terascale)
– Terascale (2017) = $3K
– Terascale (2020) = ~$1K
• Narrow Worker (Petascale)
– Recognition (Fast)
– Petascale (2040) = ~$1K
• Broad Worker (Exascale)
– Reasoning (Slow)
– Exascale (2060) = ~$1K
83/4/2018 (c) IBM 2017, Cognitive Opentech Group
2080204020001960
$1K
$1M
$1B
$1T
206020201980
+/- 10 years
$1
Person Average
Annual Salary
(Living Income)
Super Computer
Cost
Mainframe Cost
Smartphone Cost
T
P
E
T P E
AI Progress on Open Leaderboards
Benchmark Roadmap to solve AI/IA
9. GDP/Employee
3/4/2018 (c) IBM 2017, Cognitive Opentech Group 9
(Source)
Lower compute costs translate into increasing productivity and GDP/employees for nations
Increasing productivity and GDP/employees should translate into wealthier citizens
AI Progress on Open Leaderboards
Benchmark Roadmap to solve AI/IA
14. TED Arai Todai Robot
3/4/2018 (c) IBM 2017, Cognitive Opentech Group 14
… when will
your smartphone
be smart enough to
pass a university
entrance exam?
15. Build: 10 million minutes of experience
3/4/2018 Understanding Cognitive Systems 15
16. Build: 2 million minutes of experience
3/4/2018 Understanding Cognitive Systems 16
18. Other Technologies: Bigger impact?
Yes.
• Augmented Reality (AR)/
Virtual Reality (VR)
– Game worlds
grow-up
• Blockchain/
Security Systems
– Trust and security
immutable
• Advanced Materials/
Energy Systems
– Manufacturing as cheap,
local recycling service
(utility fog, artificial leaf, etc.)
3/4/2018 (c) IBM 2017, Cognitive Opentech Group 18
19. Industries Transformed
Digital Natives Transportation Water Manufacturing
Energy Construction ICT Retail
Finance Healthcare Education Government
“The best way to predict the future is to inspire the next generation of students to build it better”
23. The age of “SERVICE+AI”
2nd March, 2018
Youji Kohda
kohda@jaist.ac.jp
School of Knowledge Science
Japan Advanced Institute of Science and Technology
24. The age of SERVICE+AI
24
“In fact, the business plans of the
next 10,000 startups are easy to
forecast: Take X and add AI. Find
something that can be made better
by adding online smartness to it.”
– Kevin Kelly, The inevitable, p.33
29. Trust leap with AI
29
“Our trust is based purely on the technology’s
functionality, how predictable to do.
But a significant shift is under way; we are no
longer trusting machines just to do something
but to decide what to do and when to do it.”
– Rachel Botsman, Who can you trust?, p.179
30. Trust leaps
• Trust leaps are necessary when
innovative technologies/services are
emerged
–Open source has made the trust leap in the
1990s
–Service economy has made the trust leap in
the early 2010s
• When and how “SERVICE+AI” will make
the trust leap? 30Copyright 2018, Youji Kohda
31. (My) Questions (micro level)
• What jobs will remain for us?
–→ “Academic jobs/scholarly work is an
exception?”
• AI and ethical issues
– → “ When a cyber-pet (e.g., Sony AIBO) sees a
domestic violence at home, what should it
do?”
• AI and trustworthy issues
– → “ Will AI be trustworthy enough as a service31Copyright 2018, Youji Kohda
32. (My) Conjecture (micro level)
• AI + books > an (average) scholar/researcher
– “Questioning becomes more important than
Answering” as Kevin Kelly says in his book, The
inevitable
• AI + legal precedents > an (average) citizen
– TBD
• AI + (written) institutions > an (average) CEO
– TBD
32Copyright 2018, Youji Kohda
33. (My) Questions (macro level)
• People suffer from bounded rationality
and organizations are key to overcome
the limitation
• AI has its version of limitation?
–At least, your AI will be different from mine
if people want to keep their privacy
Copyright 2018, Youji Kohda 33
34. (My) Conjecture (macro level)
• In future, AIs will start to collaborate
together, forming “society of AIs”
–At the moment, AIs are competing each
other, e.g., in stock market
–All of the automatic driving cars will start to
communicate each other to optimize traffic
• “Society of AIs” > “Community of human
(average) professionals”
Copyright 2018, Youji Kohda 34
36. Towards Trust Building with Cognitive
Assistants: Trust Determinants in
People’s Interactions
SIDDIKE, Md. Abul Kalam
School of Knowledge Science, JAIST
2018-03-01
37. Professional and Research Experiences
• Distinguished visiting scholar, IBM Almaden
Research Center, CA, USA
• Researcher, Tokyo Tech, Tokyo Japan
• Research assistant, FSKTM, University of
Malaya, Malaysia
• Lecturer, University of Dhaka, Bangladesh
• Information officer, icddr’b, Dhaka,
Bangladesh
37
38. What is Cognitive Assistants (CAs)?
• CAs are new decision tools, able to augment human
capabilities and expertise understanding the
environment around us with a depth and clarity.
(Spohrer, 2016; Spohrer et al., 2017; Spohrer, Siddike and Kohda, 2017)
• CAs can provide people with high-quality
recommendations and help them make better data
driven decision.
(Demirkan et al., 2015)
• People problem solving capabilities significantly
augmented by the interaction of people and CAs.
(Spohrer and Banavar, 2015; Spohrer and Siddike, 2017)
38
Figure 1: Examples of low level CAs
Figure 2: Examples of high level CAs
39. Components of Trust in Different
Disciplines
39
Components of Trust Discipline Authors
Willingness, confidence, predictability, dependability,
faith and integrity, group norms, altruism, shared values,
good will
Trust in close
relationships
Deutsch, 1960; Rempel, Holmes, and
Zanna, 1985; Rotter, 1980; Scanzoni,
1979
Ability (competencies), benevolence (loyalty, openness,
receptivity, availability of caring) and integrity
(consistency, discreetness, fairness, promise, reliability,
value congruence)
Organizational trust
Butler, 1991; Gabarro, 1978; Jones,
James, and Bruni, 1975; Mayer, Davis,
and Schoorman, 1995; Schoorman,
Mayer, and Davis, 2007
Accuracy of information, trust in information, trust in
action
Trust in economics
Henry and Dietz, 2011; Ostrom, 2003;
Ostrom and Walker, 2003
Reduced workload, reduced uncertainty, reduced risk
reliability, robustness, familiarity, accuracy, task
complexity, ability, predictability, dependability,
benevolence, openness
Trust in automation
Jian, Bisantz, and Drury, 2000; Lee and
See, 2004; Muir, 1994; Muir and Moray,
1996; Parasuraman and Riley, 1997; Xu,
Le, Deitermann, and Montague, 2014
Attractiveness, enjoyment, performance, attributes Trust in robots Yuksel, Collison and Czerwinski, 2017
Reliability, attractiveness, emotional attachment,
trustworthiness, relative advantages
Trust in CAs Our framework
Table 1: Trust components in different disciplines
40. Framework of Trust Determinants
with CAs
40
Figure 8: Trust determinants with CAs
Perceived reliability
Perceived
attractiveness
Perceived attachments
Perceived
trustworthiness of
users
Intention to use CAs
P1
P2
P3
Relative advantages of
innovation
P4
P5Jim Spohrer (IBM): “Many people are
very attached to their smartphones;
Today, apps are digital tools that will
become cognitive assistants (CAs). As
we solve AI, they will become low-cost
digital workers, and IA for people.”
41. Reliability of Scales
41
Variables
Cronbach’s
alpha
Perceived reliability (4 items) .937
Perceived attractiveness (3 items) .888
Perceived emotional attachments
(items 4)
.973
Trustworthiness (5 items) .946
Relative advantages of innovation
(5 items)
.915
Intention to use CAs (4 items) .912
4.5 Validation of trust determinants
• Scale reliability was assessed
by calculating Cronbach’s
Alpha, a measure of internal
consistency, for each
measured scale.
• Nunally (1978) and DeVillis
(2003) suggest alpha value of
>.7 to be good reliability for
scale items.
• The internal reliability of
these measures was proven to
be acceptable.
Table 12: Scales’ reliability
42. Future Research Directions
42
Open coding
Selective coding
Sub-categories Core-categories
CAs show responsibilities Responsibilities
Rights and responsibilities
CAs gain rights Rights
Need rules and regulations for accessing private
information
Rules and regulations
Policy formulation
Permission is necessary to use my personal data Required permission
Well-encrypted information so that no one can access it
without my permission
Well-encryption
Security and privacySecurity and privacy of personal information
Security of personal data and
information
Need permission to sell my data Required permission for selling data
Worry about lick of my private information Lick of private information
Ownership of data Ownership of data Data ownership
Trust on vendors
Trust on vendors and CAs Trust
Trust on CAs
Share some parts of my life Knowing me wrong
Accuracy of informationConcerned about grand children
Negative effects to society
Concerned about old people
Machine can be error Machine errors Accuracy of performance
4.2 Factors to be considered for future CAs
Table 7: Future factors for transformation of CAs as actors
43. Contact
• Email: ”md.abul kalam Siddike"
kalam.siddike@gmail.com
• Please contact me if you see opportunities to
collaborate on the study of trust of cognitive
assistants (CAs) from a service science
perspective
43
Notas do Editor
Please reuse – contact spohrer@us.ibm.com
Reference:
Spohrer, J (2018) Preparing for Our Future with Opentech AI. Friday March 2, 2018
URL http://www.slideshare.net/spohrer/trl-jaist-20180304-v6
The Future of AI: Measuring Progress and PreparingAn industry perspective and forecast of where technology is going, including the what and when for "solving" Artificial Intelligence (AI), is presented. Next, the benefits and challenges will be discussed, including impact on jobs, both near term via Intelligence Augmentation (IA) and longer term via automation. The impact on different sectors of the economy will be explored, and how best to prepare for the changes that are anticipated.Speaker Bio: Dr. James ("Jim") C. Spohrer is IBM Director, Cognitive Opentech Group. Previously, he was Director of IBM Global University Programs, co-founded IBM Research Service Research area, ISSIP Service Science community, and was CTO of IBM’s VC Group in Silicon Valley. At Apple Computer (1990’s), as a Distinguished Engineer Scientist and Technologist, he developed next generation learning platforms. Earlier (1974-1989), he earned an MIT BS Physics, Yale PhD in CS/AI, and worked at Verbex, an Exxon company on speech recognition and machine learning. With over ninety publications and nine patents, he is a PICMET Fellow and a winner of the Gummesson Service Research award as well as the Vargo & Lusch Service-Dominant Logic award.More information here:Sample presentation: https://www.slideshare.net/spohrer/future-20171110-v14Bio and CV: http://service-science.info/archives/2233Optional Business, Marketing, and Technical Pre-reads:IBM Bluemine: Industry Predictions 2018:"2018 sees increased adoption of AI and digital transformation across all industries, with cloud and security also very prominent."Another predication to consider:...vendor performance on open challenge, AI leaderboards will increase adoption of the vendor's AI offerings.See for example, Alibaba annoucement yesterday on Standford open Q&A leaderboard: http://money.cnn.com/2018/01/15/technology/reading-robot-alibaba-microsoft-stanford/index.htmlAlso, see Tencent paper and Github code:ArXiv: https://arxiv.org/abs/1606.01549Github: https://github.com/bdhingra/ga-readerIBM Research was #1 Jan 2017 on same Standford open Q&A leaderboard (SQuAD) referred to above: https://rajpurkar.github.io/SQuAD-explorer/And to understand why solving AI is still very, very, very hard, in spite of all the hype:Ernie Davis (NYU) pointers: Real “reading” with background knowledge and comonesense reasoning is very, very, very hard.... see: https://arxiv.org/abs/1707.07328 in which programs that were achieving as high as 75% on this same database dropped to an accuracy of 36% if you add an automatically generated distractor sentence --- down to 7% if the distractor sentences are allowed to be ungrammatical sequences of words. The MSFT/Alibaba program has not been tested this way, of course, so there is no saying what would be the effect. Here are the slides about the “human-level performance claim” which is hyped of course: http://u.cs.biu.ac.il/~yogo/squad-vs-human.pdfOptional Pre-read for Societal Implications:https://www.mckinsey.com/business-functions/mckinsey-analytics/our-insights/where-is-technology-taking-the-economyThe economy has arrived at a point where it produces enough in principle for everyone, but where the means of access to these services and products, jobs, is steadily tightening. So this new period we are entering is not so much about production anymore—how much is produced; it is about distribution—how people get a share in what is produced. We are not quite at 2030, but I believe we have reached the “Keynes point,” where indeed enough is produced by the economy, both physical and virtual, for all of us. (If total US household income of $8.495 trillion were shared by America’s 116 million households, each would earn $73,000, enough for a decent middle-class life.) And we have reached a point where technological unemployment is becoming a reality.The problem in this new phase we’ve entered is not quite jobs, it is access to what’s produced. Jobs have been the main means of access for only 200 or 300 years. When things settle I’d expect new political parties that offer some version of a Scandinavian solution: capitalist-guided production and government-guided attention to who gets what. Europe will find this path easier because a loose socialism is part of its tradition. The United States will find it more difficult; it has never prized distribution over efficiency.This is the image I have created which will go with the talk. Can we prepone our 1:1 to tomorrow? Or Friday is best?Thanks, -JimJim Spohrer, PhDDirector, Cognitive Opentech Group (COG)IBM Research - Almaden, 650 Harry Road San Jose, CA 95120(o) 408-927-1928<spohrer@us.ibm.com>(m) 408-829-3112<spohrer@gmail.com>Innovation Champion: http://service-science.info/archives/2233
The Future of AI and Education: Measuring Progress and PreparingAn industry perspective and forecast of where technology is going,including the what and when for "solving" Artificial Intelligence (AI),is presented. Next, the benefits and challenges will be discussed,including impact on jobs, both near term via IntelligenceAugmentation (IA) and longer term via automation. The impact ondifferent sectors of the economy will be explored, especially the impacton education, and how best to prepare students and others for thechanges that are anticipated. In conclusion, the importanceof T-shaped skills that integrate deep skills in problem-solving (STEM)and broad skills in communications (both business as well as arts&humanities) will be discussed. The rationale for teaching students, whoare empowered by advanced technologies, to competeto find better ways to rapidly rebuild society from scratch will beexplained. Speaker Bio:Dr. James ("Jim") C. Spohrer is IBM Director, Cognitive Opentech Group.Previously, he was Director of IBM Global University Programs,co-founded IBM Research Service Research area, ISSIP Service Sciencecommunity, and was CTO of IBM’s VC Group in Silicon Valley. At Apple Computer (1990’s), as a Distinguished Engineer Scientist andTechnologist, he developed next generation learning platforms. Earlier(1974-1989), he earned an MIT BS Physics, Yale PhD in CS/AI, and workedat Verbex, an Exxon company on speech recognitionand machine learning. With over ninety publications and nine patents,he is a PICMET Fellow and a winner of the Gummesson Service Researchaward as well as the Vargo&Lusch Service-Dominant Logic award.More information here:Sample presentation:https://www.slideshare.net/spohrer/future-20171110-v14Bio and CV: http://service-science.info/archives/2233
The majority of IBM’s offerings to customers include open technologies – and customers increasingly demand solutions based on open source code. Tens of thousands of IBMers have GitHub accounts
Awards program information
URLS:
Htttp://developer.ibm.com/code
What is beyond Exascale? Zetta (21), Yotta (24)
Time dimension (x-axis) is plus or minus 10 years….
Daniel Pakkala (VTT)
URL: https://aiimpacts.org/preliminary-prices-for-human-level-hardware/
Dan Gruhl:
https://www.washingtonpost.com/archive/business/1983/11/06/in-pursuit-of-the-10-gigaflop-machine/012c995a-2b16-470b-96df-d823c245306e/?utm_term=.d4bde5652826
In 1983 10 GF was ~10 million.
That's 24.55 million in today's dollars.
or 2.4 billion for 1 TF in 1983
Today 1 TF is about $3k http://www.popsci.com/intel-teraflop-chip
Source: http://service-science.info/archives/4741
O*NET Online is the occupation network online, started by the US Dept of Labor in the 1990’s – it now represents one of the most comprehensive lists of occupations along with a great deal of information about each occupation, including skills, tasks, certifications, demand for these jobs, etc.
O*NET lists about 1000 occupations from Accountants to Zoologists – and many job families in between. O*NET updates the descriptions of the occupations as well as adding new occupations over time.
Source:
http://www.onetonline.org/find/family?f=0
1950 Nathaniel Rochester (IBM) 701 first commercial computer that did super-human levels of numeric calculations routinely. He worked at MIT on arithmetic unit of WhirlWind I programmable computer.
Dota 2 is most recent August 11, 2017 as a super-human game player in Valve Dota 2 competition – Elon Musk’s OpenAI result.
Miles Bundage tracks gaming progress: http://www.milesbrundage.com/blog-posts/my-ai-forecasts-past-present-and-future-main-post
DOTA2: https://blog.openai.com/more-on-dota-2/
The nature of reality changes when there is more than one intelligent species, and we are not the smartest.
The nature of reality also changes when the cost of exploring alternate experience pathways are made less risky – the notions of time and identity changes as a result.
Mitigate risks and harvest benefits of existence, by learning to evermore efficiently and rapidly rebuild from scratch to higher states of value and capability of entities.
The evolving ecology of service system entities their value co-creation and capability co-elevation mechanisms, as well as their capabilities, constraints, rights, and responsibilities at each stage in time. Human progress as well as the development of individuals, and the arc of institutions can be viewed in this way. Entities exist as individuals and populations. Generations of entities, generations of species (populations), generations of individuals (cohorts).
By 2036, there will be an accumulation of knowledge as well as a distribution of knowledge in service systems globally. We need to ensure as there is knowledge accumulation that service systems at all scale become more resilient. Leading to the capability of rapid rebuilding of service systems across scales, by T-shaped people who understand how to rapidly rebuild – knowledge has been chunked, modularized, and put into networks that support rapid rebuilding.
The weakest link is what needs to be improved – according to system scientists. Accessing help, service, experts is the weakest link in most systems.
By 2035 the phone may have the power of one human brain – by 2055 the phone may have the power of all human brains.
Before trying to answer the question about which types of sciences are more important – the ones that try to explain the external world or the ones that try to explain the internal world – consider this, slide that shows the different telephones that I have used in my life. I grew up in rural Maine, where we had a party line telephone because we were somewhat remote on our farm in Newburgh, Maine.
However, over the years phones got much better…. So in 2035 or 2055, who are you going to call when you need help?
Good morning. Welcome to my preliminary defense presentation. First of all, I would like to thanks my supervisor Prof. Youji Kohda for his all kind of supports and helps. Secondly, I would like to express my deepest gratitude to Jim for is great mentorship and giving me opportunity to work with him. Thirdly, I would like to express my sincere thanks to my second supervisor Shirahada-sensei for his love and encouragement. Finally, I would like to thanks Yoshida sensei and Nishimoto sensei of being committee members of my doctoral examination.
Today, I would like to talk about “Towards………………………………………….”
CAs are new decision tools.
It is able to augment human capabilities and expertieses.
It can understand the environment around us.
Cas can provide high quality recommendations. It also helps for better data driven decision.
For example, Apple Sir, IBM Watson, Google home and Amazon echo are considered as low level CAs. On the other hand, IBM Watson Ochology, and Driverless car are considered as high level CAs.