Web Science Institute, University of Southampton
September 11, 2018
George A. Tilesch
This Photo by Unknown Author is licensed under CC BY-SA
My early influences
Cyberiad Paranoia RPG
“Artificial intelligence is the future (…)
for all humankind,”. “It comes with
colossal opportunities, but also threats
that are difficult to predict.
Whoever becomes the leader in this
sphere will become the ruler of the
world.”
This Photo by Unknown Author is licensed under CC BY-NC
Before the Storm
• Timeline: Incrementalism vs Sci-fi?
• Expected Impact: Disjointed
Narratives & Sense-making
• AI is here and now: Adoption Trends
& “Domestication”
• $60B Industry by 2025
• 1.7 million new AI robots in factories
by 2020
• 55% of households worldwide will
have an AI voice assistant by 2022
AI Trajectory: winner takes it all?
Weak/Narrow
AI (ANI = the
Present, the Tip
of the Iceberg)
Strong/
General
AI (AGI = Holy
Grail,
currently
“Enfants
Terribles”
Artificial
Super
Intelligence (
ASI =
Speculation,
Scifi)
Snapshot: The Global Race for AI
States
US: 66% of investments historically,
abundance of AI startups +
DoD/DARPA but no natl strategy
EU: Prio but behind; EUR 3 B
investment. Focus: ethics &
impacts, H2020 funding, boards
and committees
China: national prio, closing in
concentration of startups, funding
surpasses US (Cca. $5Bn)
Russia: state-driven “hustle”,
cyberwarfare/EWF edge
Corp
Investment, Acquisitions: Google,
Apple, FB
AI Platforms/ Projects: Watson,
TensorFlow, Azure ML, SF Einstein
etc
Recent trends
•Opensourcing boosts startups,
skills & adoption
•Academic brain drain
“Startups”
Funding up 150% YoY, $10 B in
2017 only
Top 100 has $12B in Funding
China’s Bytedance top funded $3.1,
but 76/100 US
“Instant AI Unicorns”
Stewards
Future of Life; Asilomar AI
Principles endorsed by Big Tech, CA
State
EU MS Forum: AI 4 People
Standards: IEEE Ethical Design
Open AI: nonprofit Safe AGI
research funded by Elon Musk etc.
AI Sense-making:
Does it make sense?
Main current trends in AI Technologies
Robotic Process
Automation (RPA)
Expert Systems
(Decisions support
continuum)
Computer Vision
(CV)
Natural Language
Processing (NLP)
Neural Networks
(NNs or ANNs)
Autonomous
Systems (AI +
Robotics)
Distributed
Artificial
Intelligence (DAI)
Affective
Computing:
(Emotions)
Evolutionary
Algorithms
(EA): Genetics
Inductive Logic
Programming (ILP)
Decision Networks
Probabilistic
Programming
Ambient
Intelligence (AmI):
Challenges
and fears
framework
AI-supported Bias
People-centered values
Consciousness vs. Hardwiring
Trust/identity
Cognitive: Dependence &
Resilience
Data: Ownership, Consent,
Privacy
Delegation: Human supervision
Responsibility: Distributed?
Automation: Job/Tax loss
Misinformation/Democratic
collapse
The AI Landscape
Notes from the field
The Google Assistant
Conundrum
• “You are already a Cyborg”
(Musk)
• Bookings: First world
problems
• Why not a scheduling system
instead in Gsuite?
• “Automatons” talking to AI
agents: Loss of job dignity &
meaning
This Photo by Unknown Author is licensed under CC BY
AI & jobs: Future of
Work
• Recent political turmoils: AI/RPA links? Skills
gap/mismatch?
• Automation: Wipe out/Replace or
Symbiotic/Assist?
• Cashierless stores with retail gong down
• Driverless truck: labour 45% costs, 5.5 M jobs impacted
• Workerless factory: AI-augmented productivity
but jobs “automated out”?
• Are “smart” jobs safe?
• Doctors. e.g 82% Cancer detection, learned in 2 hours
• OMG, not the lawyers!
Autonomous
Machines of War
• AI Blitzkrieg (AWS, 0-
casualty war etc.)
• AI MAD vs. winner
takes it all
• Project Maven (Google
backlash & PPP
analysis)
• UN LAWs Consensus
seems far: blockers US,
Russia, Israel etc.
Nascent AI-triggered belief
systems
• Silicon Valley Techno-optimism  AI Church: e.g.
Way of the Future
• Transhumanism: Rise of the Centaur
• Neoluddism
• Consciousness Schools of Thought
AI & The Divine
THE A/IS Ethical Framework (draft)
Human Rights A/IS shall respect, promote, and protect internationally recognized human rights.
Prioritizing Well-Being A/IS shall adopt increased human well-being as a primary success criterion for
development.
Human Data Agency Developers and operators of AI/S shall develop effective models of consent for the
capturing, sharing and use of personal data.
Accountability It shall always be possible to find out who is responsible for an A/IS at any point in
time and in any stage of operation.
Transparency It shall always be possible to discover why an A/IS made a particular decision.
Measurability Always be possible to measure and provide evidence of the effectiveness (fitness
for purpose) of an A/IS.
Competence Designers shall specify the knowledge and skill required to apply the AI/S effectively
and safely, and operators must be trained accordingly.
A/IS Technology Misuse and Awareness Developers shall guard against all potential misuses and risks of AI/S.
The Light at the End?
• We need empowered multi-stakeholder structures to
guide an AI Future
• New models for innovation stewards: guardians &
curators & doers, all in one
• Conscience & Consciousness for machines and humans
alike
• A new core: Upgraded, flexible, strong belief and ethical
systems
• Mental Resilience, also offline/online, global/local
balance
• New collective creativity collaboration tools needed
• Cross-disciplinary will be the new normal/desired
This Photo by Unknown Author is licensed under CC BY-SA
The job (and the
responsibility) is yours
“Nobody phrases it this way, but I
think that artificial intelligence is
almost a humanities discipline.
It's really an attempt to
understand human intelligence
and human cognition.”
Sebastian Thrun

#AI: In Whose Image?

  • 1.
    Web Science Institute,University of Southampton September 11, 2018 George A. Tilesch This Photo by Unknown Author is licensed under CC BY-SA
  • 2.
  • 3.
    “Artificial intelligence isthe future (…) for all humankind,”. “It comes with colossal opportunities, but also threats that are difficult to predict. Whoever becomes the leader in this sphere will become the ruler of the world.” This Photo by Unknown Author is licensed under CC BY-NC
  • 4.
    Before the Storm •Timeline: Incrementalism vs Sci-fi? • Expected Impact: Disjointed Narratives & Sense-making • AI is here and now: Adoption Trends & “Domestication” • $60B Industry by 2025 • 1.7 million new AI robots in factories by 2020 • 55% of households worldwide will have an AI voice assistant by 2022
  • 5.
    AI Trajectory: winnertakes it all? Weak/Narrow AI (ANI = the Present, the Tip of the Iceberg) Strong/ General AI (AGI = Holy Grail, currently “Enfants Terribles” Artificial Super Intelligence ( ASI = Speculation, Scifi)
  • 6.
    Snapshot: The GlobalRace for AI States US: 66% of investments historically, abundance of AI startups + DoD/DARPA but no natl strategy EU: Prio but behind; EUR 3 B investment. Focus: ethics & impacts, H2020 funding, boards and committees China: national prio, closing in concentration of startups, funding surpasses US (Cca. $5Bn) Russia: state-driven “hustle”, cyberwarfare/EWF edge Corp Investment, Acquisitions: Google, Apple, FB AI Platforms/ Projects: Watson, TensorFlow, Azure ML, SF Einstein etc Recent trends •Opensourcing boosts startups, skills & adoption •Academic brain drain “Startups” Funding up 150% YoY, $10 B in 2017 only Top 100 has $12B in Funding China’s Bytedance top funded $3.1, but 76/100 US “Instant AI Unicorns” Stewards Future of Life; Asilomar AI Principles endorsed by Big Tech, CA State EU MS Forum: AI 4 People Standards: IEEE Ethical Design Open AI: nonprofit Safe AGI research funded by Elon Musk etc.
  • 7.
  • 8.
    Main current trendsin AI Technologies Robotic Process Automation (RPA) Expert Systems (Decisions support continuum) Computer Vision (CV) Natural Language Processing (NLP) Neural Networks (NNs or ANNs) Autonomous Systems (AI + Robotics) Distributed Artificial Intelligence (DAI) Affective Computing: (Emotions) Evolutionary Algorithms (EA): Genetics Inductive Logic Programming (ILP) Decision Networks Probabilistic Programming Ambient Intelligence (AmI):
  • 9.
    Challenges and fears framework AI-supported Bias People-centeredvalues Consciousness vs. Hardwiring Trust/identity Cognitive: Dependence & Resilience Data: Ownership, Consent, Privacy Delegation: Human supervision Responsibility: Distributed? Automation: Job/Tax loss Misinformation/Democratic collapse
  • 10.
    The AI Landscape Notesfrom the field
  • 11.
    The Google Assistant Conundrum •“You are already a Cyborg” (Musk) • Bookings: First world problems • Why not a scheduling system instead in Gsuite? • “Automatons” talking to AI agents: Loss of job dignity & meaning This Photo by Unknown Author is licensed under CC BY
  • 12.
    AI & jobs:Future of Work • Recent political turmoils: AI/RPA links? Skills gap/mismatch? • Automation: Wipe out/Replace or Symbiotic/Assist? • Cashierless stores with retail gong down • Driverless truck: labour 45% costs, 5.5 M jobs impacted • Workerless factory: AI-augmented productivity but jobs “automated out”? • Are “smart” jobs safe? • Doctors. e.g 82% Cancer detection, learned in 2 hours • OMG, not the lawyers!
  • 13.
    Autonomous Machines of War •AI Blitzkrieg (AWS, 0- casualty war etc.) • AI MAD vs. winner takes it all • Project Maven (Google backlash & PPP analysis) • UN LAWs Consensus seems far: blockers US, Russia, Israel etc.
  • 14.
    Nascent AI-triggered belief systems •Silicon Valley Techno-optimism  AI Church: e.g. Way of the Future • Transhumanism: Rise of the Centaur • Neoluddism • Consciousness Schools of Thought AI & The Divine
  • 15.
    THE A/IS EthicalFramework (draft) Human Rights A/IS shall respect, promote, and protect internationally recognized human rights. Prioritizing Well-Being A/IS shall adopt increased human well-being as a primary success criterion for development. Human Data Agency Developers and operators of AI/S shall develop effective models of consent for the capturing, sharing and use of personal data. Accountability It shall always be possible to find out who is responsible for an A/IS at any point in time and in any stage of operation. Transparency It shall always be possible to discover why an A/IS made a particular decision. Measurability Always be possible to measure and provide evidence of the effectiveness (fitness for purpose) of an A/IS. Competence Designers shall specify the knowledge and skill required to apply the AI/S effectively and safely, and operators must be trained accordingly. A/IS Technology Misuse and Awareness Developers shall guard against all potential misuses and risks of AI/S.
  • 16.
    The Light atthe End? • We need empowered multi-stakeholder structures to guide an AI Future • New models for innovation stewards: guardians & curators & doers, all in one • Conscience & Consciousness for machines and humans alike • A new core: Upgraded, flexible, strong belief and ethical systems • Mental Resilience, also offline/online, global/local balance • New collective creativity collaboration tools needed • Cross-disciplinary will be the new normal/desired This Photo by Unknown Author is licensed under CC BY-SA
  • 17.
    The job (andthe responsibility) is yours “Nobody phrases it this way, but I think that artificial intelligence is almost a humanities discipline. It's really an attempt to understand human intelligence and human cognition.” Sebastian Thrun