Social Connections 14 - You Get What You Give
Download as PPTX, PDF•1 like•611 views
The document discusses the themes and insights shared during the Social Connections 14 conference held in Berlin on October 16-17, 2018, focusing on the challenges of employee engagement, data accuracy, and the role of AI in the workplace. It highlights the limitations of AI and advocates for augmenting human capabilities rather than replacing them, emphasizing the need for human understanding alongside technological advancements. Additionally, it presents case studies and historical context regarding AI developments and their implications in various fields, including healthcare.
![Social Connections 14 Berlin, October 16-17 2018
Watson for Oncology
• Watson reportedly suggested the man be administered
both chemotherapy and the drug “Bevacizumab.” But the
drug can lead to “severe or fatal hemorrhage,”
• One doctor called it “a piece of shit”
• “The report puts the blame on the IBM engineers and the
Memorial Sloan Kettering (MSK) Cancer Center doctors
who helped train the AI”
• “[recommendation] was just a part of system testing”](https://image.slidesharecdn.com/you-get-what-you-givev2-181025154920/85/Social-Connections-14-You-Get-What-You-Give-34-320.jpg)
![Social Connections 14 Berlin, October 16-17 2018
Watson for Oncology
• “the hospital still uses Watson's recommendations…[sees]
them as an extra opinion when they can't agree on a
treatment”
• “we have learned and improved Watson Health based on
continuous feedback from clients, new scientific evidence, and
new cancers and treatment alternatives”
• Watson for Oncology is trained to help oncologists treat 13
cancers and is being used by 230 hospitals around the world,
and has “supported care for more than 84,000 patients.”
• “No technology can replace a doctor and his or her knowledge
about their individual patient.”](https://image.slidesharecdn.com/you-get-what-you-givev2-181025154920/85/Social-Connections-14-You-Get-What-You-Give-35-320.jpg)
Social Connections 14 - You Get What You Give
- 1. Berlin, October 16-17 2018 You Get What You GivePaul Withers, Intec Systems Ltd @paulswithers Luis Suarez, panagenda @elsua
- 2. PLATINUM SPONSORS GOLD SPONSORS BRONZE SPONSORS SILVER SPONSORS SPEEDSPONSORING BEER SPONSOR
- 3. Social Connections 14 Berlin, October 16-17 2018 Paul Withers • ICS Developer, Intec Systems • IBM Lifetime Champion • OpenNTF Board Member
- 4. Social Connections 14 Berlin, October 16-17 2018 Luis Suarez • Digital Transformation & Data Analytics Adviser - panagenda • IBM ICS Champion • http://elsua.net • http://twitter.com/elsua • http://www.linkedin.com/in/elsua
- 5. Social Connections 14 Berlin, October 16-17 2018 Why are we here today? Source - https://unsplash.com/photos/vLXdetEcB40
- 6. Social Connections 14 Berlin, October 16-17 2018 Is it because … • If 15% show high engagement, what about the “other” 85%? • 5% fluctuation * 1,000 employees = 50 * 43,000 = 2,150,000 EUR – Excluding side effects (i.e. think of losing sellers) •To reach optimal productivity, a new hire usually needs 3.5 years Source - http://bit.ly/GallupUKDisengaged
- 7. Social Connections 14 Berlin, October 16-17 2018 Or is it because? “For 70% of Executives, the accuracy of their data is what keeps them up at night.” (Forbes Insight) “Only one third of our decisions are correct.” (Martina Koederitz, IBM GM DACH) “We ignore about 80% of our data.” (Christoph Keese, Axel Springer AG) 80% Shadow IT Source: Frost & Sullivan 2013, GigaOM 2014 $330b Cost of Burnout per Year Source: WHO
- 8. Social Connections 14 Berlin, October 16-17 2018 Is this really us? Source - https://www.flickr.com/photos/49503168860@N01
- 9. Social Connections 14 Berlin, October 16-17 2018 What can we do? Engaged Not Engaged DIS-Engaged More information about Gallup and Q12® http://www.gallup.de/183104/engagement-index-deutschland.aspx
- 10. Social Connections 14 Berlin, October 16-17 2018 Can the machines (A.I.) save us? Source - http://bit.ly/2zYM6aA Source - https://twitter.com/timoreilly/status/983199489375653890
- 11. Social Connections 14 Berlin, October 16-17 2018 Maybe, maybe not! ‘I have had it with the stream of articles about what an “AI” can do. Yes, machine learning works. It is possible to analyze key words, correlate them with other key words, do a massive amount of statistics, and find out some stuff. People cannot do that and computers can. Is this AI? It sure isn't anything people can do, and it also doesn't correspond to anything I understand about what it means to be intelligent’. - Roger Schank (@rogerschank) Source - https://www.linkedin.com/pulse/ten-questions-ai-roger-schank To help understand “AI” it helps to understand “I” Source - https://www.linkedin.com/feed/update/urn:li:activity:6438037501369212928/
- 12. Social Connections 14 Berlin, October 16-17 2018 Don’t believe the hype around ‘A.I.’ Source - https://twitter.com/Cennydd/status/1019594285916794880
- 13. Social Connections 14 Berlin, October 16-17 2018 Instead, let’s focus on augmenting humans Source - https://www.gapingvoid.com/
- 14. Social Connections 14 Berlin, October 16-17 2018 And how do we do that…? Source - https://unsplash.com/photos/4-EeTnaC1S4
- 15. Social Connections 14 Berlin, October 16-17 2018 The Premise • AI uses insights from unstructured big data • But haven’t we heard this before? • Why will Watson (or A.N.Other) succeed where others failed? • If it does, are we an endangered species
- 16. Social Connections 14 Berlin, October 16-17 2018 Example 1: Autonomy
- 17. Social Connections 14 Berlin, October 16-17 2018 Autonomy • Autonomy Corp PLC HP Autonomy • Founded 1996, Cambridge (England) • KM tools analysing unstructured big data
- 18. Social Connections 14 Berlin, October 16-17 2018 Case Study • Brought in by IT • Categorisation of MS Office docs company-wide • Intended to: • Advise users of similar documents • Advise users of experts
- 19. Social Connections 14 Berlin, October 16-17 2018 However… • Sample content needed loading, categorising, tagging etc • Required business knowledge not IT knowledge • Required business time • Required business buy-in
- 20. Social Connections 14 Berlin, October 16-17 2018 Example 2:
- 21. Social Connections 14 Berlin, October 16-17 2018 Evolution of AI
- 22. Social Connections 14 Berlin, October 16-17 2018 Post-War • Turochamp chess program (1948) • Turing test (1950)
- 23. Social Connections 14 Berlin, October 16-17 2018 Herbert Simon, Alan Newell (1955) • The Logic Theorist • Proved theorems from Principia Mathematica, rise of heuristics
- 24. Social Connections 14 Berlin, October 16-17 2018 Arthur Samuel, IBM (1959) • Coined the term “machine learning”
- 25. Social Connections 14 Berlin, October 16-17 2018 Moravec’s Paradox “It is comparatively easy to make computers exhibit adult level performance on intelligence tests or playing checkers, and difficult or impossible to give them the skills of a one-year- old when it comes to perception and mobility” Moravec, Hans (1988), Mind Children, Harvard University Press
- 26. Social Connections 14 Berlin, October 16-17 2018 What Is This?
- 27. Social Connections 14 Berlin, October 16-17 2018 • Deep Blue (1997) • Watson (2011) • AlphaGo (2016)
- 28. Social Connections 14 Berlin, October 16-17 2018 AI = Artificial Intelligence “That’s where all AI is today – capable of finding patterns in data with astonishing detail and sensitivity, but with no real understanding of what those patterns actually mean” The Joy of AI, BBC 4.1
- 29. Social Connections 14 Berlin, October 16-17 2018 AI Next “Somewhere between 18 months and 2 years old, children start doing something remarkable. Show them how to do things…even just once and they start practising it for themselves. This is called ‘one shot learning’. For computer scientists …this is like the holy grail.” The Joy of AI, BBC 4.1
- 30. Social Connections 14 Berlin, October 16-17 2018 Artificial Without Intelligence
- 31. Social Connections 14 Berlin, October 16-17 2018 Alexa… https://www.digitaltrends.com/home/funny-accidental-amazon-alexa-ordering-stories/
- 32. Social Connections 14 Berlin, October 16-17 2018 Watson Assistant Chatbot 01Code Intents (NLP), Entities (Structured Processing) and Dialog Flow 02“Try It”, review confidences and tweak it 03Deploy it 04Review conversations for weak understanding, incorrect understanding, incorrect flows and unanticipated flows
- 33. Social Connections 14 Berlin, October 16-17 2018 AI = Augmented Intelligence https://www.engadget.com/2018/07/27/ibm-watson-for- oncology-unsafe-treatment-plans-report/ https://gizmodo.com/ibm-watson-reportedly-recommended- cancer-treatments-tha-1827868882
- 34. Social Connections 14 Berlin, October 16-17 2018 Watson for Oncology • Watson reportedly suggested the man be administered both chemotherapy and the drug “Bevacizumab.” But the drug can lead to “severe or fatal hemorrhage,” • One doctor called it “a piece of shit” • “The report puts the blame on the IBM engineers and the Memorial Sloan Kettering (MSK) Cancer Center doctors who helped train the AI” • “[recommendation] was just a part of system testing”
- 35. Social Connections 14 Berlin, October 16-17 2018 Watson for Oncology • “the hospital still uses Watson's recommendations…[sees] them as an extra opinion when they can't agree on a treatment” • “we have learned and improved Watson Health based on continuous feedback from clients, new scientific evidence, and new cancers and treatment alternatives” • Watson for Oncology is trained to help oncologists treat 13 cancers and is being used by 230 hospitals around the world, and has “supported care for more than 84,000 patients.” • “No technology can replace a doctor and his or her knowledge about their individual patient.”
- 36. Social Connections 14 Berlin, October 16-17 2018 Or Can It? • Remember AlphaGo? • What if AI recommends the correct treatment for a reason a human can’t comprehend?
- 37. Social Connections 14 Berlin, October 16-17 2018 AI = Augmented Incompetence
- 38. Social Connections 14 Berlin, October 16-17 2018 And Are YOU Safe? https://www.perforce.com/blog/qac/will-ai- replace-programmers
- 40. PLATINUM SPONSORS GOLD SPONSORS BRONZE SPONSORS SILVER SPONSORS SPEEDSPONSORING BEER SPONSOR
Editor's Notes
- #16: Start with this from “The Imitation Game”? https://www.youtube.com/watch?v=lTMEUA6VnJE
- #23: https://en.wikipedia.org/wiki/Turing_test, https://en.wikipedia.org/wiki/Turochamp combinatorial explosion
- #24: https://en.wikipedia.org/wiki/Logic_Theorist - solved theorems from Principia Mathematica. Search tree, heuristics The proof of theorem 2.85 was actually more elegant than the proof produced laboriously by hand by Russell and Whitehead.
- #25: Machine learning uses pattern, from training data – decision tree learning, association rule learning, artificial neural networks, deep learning, Bayesian networks etc
- #26: https://en.wikipedia.org/wiki/Moravec%27s_paradox
- #27: AI understands patterns in data without understanding concepts, so what the patterns mean (like what a dog is)
- #28: AlphaGo made a move (move 37) in game 2 not in any books and one experts couldn’t understand (thought it was a mistake), Lee Sedol thought about it 20 minutes, but which was crucial to winning
- #30: “Our AI systems are nowhere near the capabilities even of a two-year-old” “[They] learn directly from data and experience rather like computers do with machine learning and artificial neural networks. But they also understand the world with abstract concepts. It’s the combination of the two, the way their learning seamlessly produces the concepts and the way the concepts then direct their learning that makes the like the most amazing computers you can imagine. It’s this combination that AI researchers are one day hoping to crack”