Notes by Retraice

(The below text version of the notes is for search purposes and convenience. See the PDF version for proper formatting such as bold, italics, etc., and graphics where applicable. Copyright: 2022 Retraice, Inc.)

AIMA4e Annotations

A companion to the great white brick.

As of November 24, 2022

(Start date: November 21, 2022.)

[1]retraice.com

Version notes: [2]Retraice ([3]2022/11/21) (Re57), first draft, covered Preface, Sections I, II; [4]Retraice ([5]2022/11/22) (Re58), no footnotes, covered Sections III, IV; [6]Retraice ([7]2022/11/22) (Re58) again, moved some notes from Re57 and Re58 notes to footnotes here; [8]Retraice ([9]2022/11/23) (Re59), covered Sections V, VI, VII.

PREFACE

* The phenomenon: intelligent agents[10]^1 * The discipline: artificial intelligence,[11]^2 "the study of agents that receive percepts from the environment and perform actions." (vii)

* Aspects of the phenomenon: + Agent function: "Each ...agent implements a function that maps percept sequences to actions" (vii) o Ways to represent agent functions include: "reactive agents, real-time planners, decision-theoretic systems, and deep learning systems." (vii) + Learning o "a construction method for competent systems" (viii) o "a way of extending the reach of the designer into unknown environments." (viii) + Goals o Robotics and vision: # "not ...independently defined problems" # "[things] in the service of achieving goals."

I INTELLIGENCE --"Artificial Intelligence"

1 Intro:

definitions, foundations, history, philosophy, state of the art, risks-benefits

2 Agents:

environments, `good' behavior, agent structure and types

II SOLVING--"Problem-solving"

3 Searching: Looking ahead to find a sequence.

Algorithms, strategies, informed/heuristic[12]^3 strategies.

4 Complex Environments: More realistic environments.

Local search, optimization, continuous spaces, nondeterministic actions, partially observable env.s, online search and unknown env.s.

5 Adversarial Games: Other agents competing against us.

Theory, optimal decisions, alpha-beta tree search, Monte Carlo tree search, stochastic g.s, partially observable g.s, limitations.

6 Constraint Satisfaction Problems: States as domains, solutions as allowable combinations of states.

Constraint propagation, inference, backtracking search, local search, structure of problems

III THINKING--"Knowledge, reasoning, and planning"

7 Logical Agents: Forming representations and reasoning before acting.

Knowledge-based agents; representing[13]^4 worlds; logic, world models and `possible worlds';[14]^5 logic without objects.

8 First-Order Logic: A formal language for objects and their relations.

`Ontological commitment' (what is assumed about reality); syntax, semantics; knowledge engineering (building formal representations of important[15]^6 objects and relations in a domain).

9 First-Order Inference: Reasoning about objects and their relations.

Algorithms to answer any 1st-order logic question.

10 Knowledge Representation: Representing the real world for problem solving.

What content to put into a knowledge base.

Knowledge representation languages and their uses (315): * First-order logic: reasoning about a world of objects and relations; * Hierarchical task networks: for reasoning about plans (chpt. 11); * Bayesian networks: for reasoning with uncertainty (chpt. 13); * Markov models: for reasoning over time (chpt. 17); * Deep neural networks: for reasoning about images, sounds, other data (chpt. 21).

11 Automated Planning: Hierarchical task networks.

Planning for spacecraft, factories, military campaigns; representing actions and states; efficient algorithms and heuristics.

IV UNCERTAINTY--"Uncertain knowledge and reasoning"

12 Quantifying Uncertainty: An answer to the laziness and ignorance that kill formal logic.

Causes of uncertainty are environment types (partially observable,[16]^7 nondeterministic, adversarial[17]^8 ); belief state grows big and unlikely fast (384); agents still need a way to act; absolute certainty is impossible;[18]^9 it comes down to importance, likelihood and degree of success (385-386).

Logic fails because laziness and ignorance; probability theory solves the qualification problem by summarizing the uncertainty.[19]^10 * Laziness: too much work to list everything, or use such a list; * Ignorance: (theoretical) there are no complete theories; (practical) we can never run all the tests.

13 Probabilistic Reasoning [big]: Bayesian networks.

For reasoning with uncertainty by representing causal independence (398) and conditional independence (401) relationships to simplify probabilistic representations of the world.

14 Probabilistic Reasoning Over Time: Comprehending the uncertain past, present and future.

[20]^11

Belief state plus transition model yields prediction (chpt 4, 7, 11); percepts and sensor model yield updated belief state; add probability theory to switch from possible states to probable states.[21]^12

15 Probabilistic Programming: Universal formal languages to represent any computable probability model, and they come with algorithms.

Using formal logic and traditional programming languages to represent probabilistic information.

16 Making Simple Decisions: Agents getting what they want in an uncertain world--as much as possible, on average.

Beliefs, desires; utility theory; utility functions; decision networks; the value of information (547);[22]^13 this chapter is concerned with one-shot or episodic decisions problems (as opposed to sequential) (cf. 562, below).

17 Making Complex Decisions: What to do today given decisions to be made tomorrow.

Sequential decision problems (as opposed to one-shot episodic, cf. above): the agent's utility depends on a sequence of decisions in stochastic (explicitly probabilistic (45)) and partially observable environments. Markov models (563; cf. 463) for reasoning over time (chpt. 17).

18 Multiagent Decision Making [big]: When there's more than one agent in the environment.

The nature of such environments and the strategies for problem-solving depend on the relationships between agents: non-cooperative and cooperative game theory; collective decision-making.

V LEARNING--"Machine learning"

19 Learning From Examples [big]: Improving behavior by observing the present (past?) and predicting the future.

Learning is improving performance (behavior) after making observations.[23]^14

If the agent is a computer: Machine learning: "a computer observes some data, builds a model based on the data, and uses the model as both a hypothesis about the world and a piece of software that can solve problems." (651)

Subsections: * supervised learning; * learning decision trees; * model selection and optimization; * theory of learning; * linear regression (finding the best-fit line, i.e. predicting `future' [dependent] values based on plotting `past' [independent] values), classification;[24]^15 * nonparametric models (which retain all the examples, aka `instance-based' or `memory-based' learning, which is more true to large datasets [scalable?] than parametric, which summarize, and then discard, training data in fixed numbers of parameters), * ensemble learning (using multiple hypotheses instead of one, and averaging or voting--`base' models are combined into an `ensemble' model); * ML system development, the practice (software engineering and design patters in ML ops).

20 Learning Probabilistic Models: View `learning' as "uncertain reasoning from observations" and model the world accordingly.

Agents can't use probability and decision theories until they learn them from experience: treat learning itself as an inference process in a probabilistic world. Use Bayesian networks. Key concepts: data and hypotheses. "Here, the data are evidence ...instantiations of some or all of the random variables describing the domain."[25]^16

21 Deep Learning: represent hypotheses as "complex algebraic circuits with tunable connection strengths."

The circuits are orginzed into layers, a multi-step computation path. Ideal for recognizing, translating and generating images (including objects in images) and speech; `neural networks'.

From chpt. 10 on knowledge rep. languages, above notes: "deep neural networks: for reasoning about images, sounds, other data."

Think: gradient descent, back-propagation, convolutional neural networks.

22 Reinforcement Learning: Learning from experiences of reward and punishment instead of correct examples from a supervisor

Passive and active RL., Q-learning, apprenticeships and inverse RL.

Cf. Reward is Enough, May 2021: [26]https://www.deepmind.com/publications/reward-is-enough

VI INTERACTING--"Communicating, perceiving, and acting"

23 Natural Language Processing: Communicating with humans and learning from what they've written.

Language model: "a probability distribution describing the likelihood of any string." (824)

N-grams, grammar, syntax, semantics, parsing, vagueness, ambiguity, quantification.

24 Deep Learning for Natural Language Processing: Using neural nets on natural language to effectively handle the complexity.

"[R]epresenting words as points in a high-dimensional space." RNNs for "long-distance context."

Cf. Attention Is All You Need, 2017: [27]https://arxiv.org/abs/1706.03762 and AIMA4e p. 868, transformer architecture, self-attention.

25 Computer Vision: Connecting AI to cameras.

Photons provide a lot of valuable information to agents--too much information.

Surveillance cameras--good and bad; cars. Lots of machines do better if they can see.

From the Preface: Robotics and vision: "not ...independently defined problems"..."[things] in the service of achieving goals."

26 Robotics: Connecting AI to sensors, effectors and actuators

To enable movement in-and of--the physical world. Cars, spacecraft, surgeons, submarines, delivery bots.

From the Preface: Robotics and vision: "not ...independently defined problems"..."[things] in the service of achieving goals."

VII CONCLUSIONS--"Conclusions"

27 Philosophy, Ethics, and Safety of AI: What is AI? What should we do with it? What might it do with us?

Trust--of systems, humans, ourselves, each other.

The human use of human beings. Usefulness of human beings at all?

Medicine. War.

28 The Future of AI: Our tools will improve dramatically; our ends might remain the same.

Our preferences, our tools, our architectures. They're ours, for now.

Minimize the negative impacts, don't maximize the positive?

A: MATH--"Appendix A: Mathematical Background"

B: CODE--"Appendix B: Notes on Languages and Algorithms"

__

References

Aleksandrov, A. D., Kolmogorov, A. N., & Lavrent'ev, M. A. (1969). Mathematics: Its Content, Methods and Meaning (3 Volumes in One). Dover, 1999 reprint ed. ISBN: 0486409163. Searches: [28]https://www.amazon.com/s?k=0486409163 [29]https://www.google.com/search?q=isbn+0486409163 [30]https://lccn.loc.gov/99033023

Bulmer, M. G. (1967). Principles of Statistics. Dover, 1979 reprint ed. ISBN: 0486637603. Searches: [31]https://www.amazon.com/s?k=0486637603 [32]https://www.google.com/search?q=isbn+0486637603 [33]https://lccn.loc.gov/78072991

Crawford, K. (2021). Atlas of AI: Power, Politics, and the Planetary Costs of Artificial Intelligence. Yale University Press. ISBN: 978-0300209570. Searches: [34]https://www.amazon.com/s?k=9780300209570 [35]https://www.google.com/search?q=isbn+9780300209570 [36]https://lccn.loc.gov/2020947842

Frankfurt, H. G. (1988). The Importance of What We Care About. Cambridge. ISBN: 978-0521336116. Searches: [37]https://www.amazon.com/s?k=978-0521336116 [38]https://www.google.com/search?q=isbn+978-0521336116 [39]https://lccn.loc.gov/87026941

Gerrish, S. (2018). How Smart Machines Think. The MIT Press. ISBN: 978-0262038409. Searches: [40]https://www.amazon.com/s?k=9780262038409 [41]https://www.google.com/search?q=isbn+9780262038409 [42]https://lccn.loc.gov/2017059862

Kahneman, D. (2011). Thinking, Fast and Slow. Farrar, Straus and Giroux. ISBN: 978-0374533557. Searches: [43]https://www.amazon.com/s?k=978-0374533557 [44]https://www.google.com/search?q=isbn+978-0374533557 [45]https://lccn.loc.gov/2012533187

Kramer, E. E. (1970). The Nature and Growth of Modern Mathematics. Hawthorn Books. No ISBN. Searches: [46]https://www.amazon.com/s?k=kramer+nature+and+growth+of+modern+mathematics [47]https://www.google.com/search?q=kramer+nature+and+growth+of+modern+mathematics [48]https://lccn.loc.gov/73091317

Macphail, E. M. (1982). Brain and Intelligence in Vertebrates. Oxford. ISBN 0198545517. Searches: [49]https://www.amazon.com/s?k=0198545517 [50]https://www.google.com/search?q=isbn+0198545517 [51]https://lccn.loc.gov/82166301

Retraice (2020/09/07). Re1: Three Kinds of Intelligence. retraice.com. [52]https://www.retraice.com/segments/re1 Retrieved 22nd Sep. 2020.

Retraice (2020/11/02). Re10: Living to Guess Another Day. retraice.com. [53]https://www.retraice.com/segments/re10 Retrieved 2nd Nov. 2020.

Retraice (2020/11/10). Re13: The Care Factor. retraice.com. [54]https://www.retraice.com/segments/re13 Retrieved 10th Nov. 2020.

Retraice (2020/11/25). Re15: Trust and Sources. retraice.com. [55]https://www.retraice.com/segments/re15 Retrieved 28th Feb. 2022.

Retraice (2022/10/19). Re23: You Need a World Model. retraice.com. [56]https://www.retraice.com/segments/re23 Retrieved 20th Oct. 2022.

Retraice (2022/10/23). Re27: Now That's a World Model - WM4. retraice.com. [57]https://www.retraice.com/segments/re27 Retrieved 24th Oct. 2022.

Retraice (2022/10/31). Re36: Notes on Conspiracy. retraice.com. [58]https://www.retraice.com/segments/re36 Retrieved 4th Nov. 2022.

Retraice (2022/11/12). Re48: From Drugs to Mao to Money. retraice.com. [59]https://www.retraice.com/segments/re48 Retrieved 14th Nov. 2022.

Retraice (2022/11/16). Re52: Big Questions About AI. retraice.com. [60]https://www.retraice.com/segments/re52 Retrieved 17th Nov. 2022.

Retraice (2022/11/21). Re57: AI, Agents, Problem-solving, Searching, Environments, Games (AIMA4e chpts. 1-6). retraice.com. [61]https://www.retraice.com/segments/re57 Retrieved 22nd Nov. 2022.

Retraice (2022/11/22). Re58: Thinking and Uncertainty (AIMA4e chpts. 7-18). retraice.com. [62]https://www.retraice.com/segments/re58 Retrieved 23rd Nov. 2022.

Retraice (2022/11/23). Re59: Learning, Interacting, Conclusions (AIMA4e chpts. 19-28). retraice.com. [63]https://www.retraice.com/segments/re59 Retrieved 24th Nov. 2022.

Russell, B. (1948). Human Knowledge: Its Scope and Limits. Routledge. First published in 1948. This edition 1992. ISBN: 0415083028. Searches: [64]https://archive.org/search.php?query=Human%20Knowledge%3A%20Its%20Scope%20and%20Limits [65]https://www.amazon.com/s?k=0415083028 [66]https://www.google.com/search?q=isbn+0415083028 [67]https://lccn.loc.gov/94209784

Vallee, J. (1979). Messengers of Deception: UFO Contacts and Cults. And/Or Press. ISBN: 0915904381. Different edition and searches: [68]https://archive.org/details/MessengersOfDeceptionUFOContactsAndCultsJacquesValle1979/mode/2up [69]https://www.amazon.com/s?k=0915904381 [70]https://www.google.com/search?q=isbn+0915904381 [71]https://catalog.loc.gov/vwebv/search?searchArg=0915904381

References

1. https://retraice.com/ 2. XReSeg57 3. XReSeg57 4. XReSeg58 5. XReSeg58 6. XReSeg58 7. XReSeg58 8. XReSeg59 9. XReSeg59 10. fn1x0 11. fn2x0 12. fn3x0 13. fn4x0 14. fn5x0 15. fn6x0 16. fn7x0 17. fn8x0 18. fn9x0 19. fn10x0 20. fn11x0 21. fn12x0 22. fn13x0 23. fn14x0 24. fn15x0 25. fn16x0 26. https://www.deepmind.com/publications/reward-is-enough 27. https://arxiv.org/abs/1706.03762 28. https://www.amazon.com/s?k=0486409163 29. https://www.google.com/search?q=isbn+0486409163 30. https://lccn.loc.gov/99033023 31. https://www.amazon.com/s?k=0486637603 32. https://www.google.com/search?q=isbn+0486637603 33. https://lccn.loc.gov/78072991 34. https://www.amazon.com/s?k=9780300209570 35. https://www.google.com/search?q=isbn+9780300209570 36. https://lccn.loc.gov/2020947842 37. https://www.amazon.com/s?k=978-0521336116 38. https://www.google.com/search?q=isbn+978-0521336116 39. https://lccn.loc.gov/87026941 40. https://www.amazon.com/s?k=9780262038409 41. https://www.google.com/search?q=isbn+9780262038409 42. https://lccn.loc.gov/2017059862 43. https://www.amazon.com/s?k=978-0374533557 44. https://www.google.com/search?q=isbn+978-0374533557 45. https://lccn.loc.gov/2012533187 46. https://www.amazon.com/s?k=kramer+nature+and+growth+of+modern+mathematics 47. https://www.google.com/search?q=kramer+nature+and+growth+of+modern+mathematics 48. https://lccn.loc.gov/73091317 49. https://www.amazon.com/s?k=0198545517 50. https://www.google.com/search?q=isbn+0198545517 51. https://lccn.loc.gov/82166301 52. https://www.retraice.com/segments/re1 53. https://www.retraice.com/segments/re10 54. https://www.retraice.com/segments/re13 55. https://www.retraice.com/segments/re15 56. https://www.retraice.com/segments/re23 57. https://www.retraice.com/segments/re27 58. https://www.retraice.com/segments/re36 59. https://www.retraice.com/segments/re48 60. https://www.retraice.com/segments/re52 61. https://www.retraice.com/segments/re57 62. https://www.retraice.com/segments/re58 63. https://www.retraice.com/segments/re59 64. https://archive.org/search.php?query=Human%20Knowledge%3A%20Its%20Scope%20and%20Limits 65. https://www.amazon.com/s?k=0415083028 66. https://www.google.com/search?q=isbn+0415083028 67. https://lccn.loc.gov/94209784 68. https://archive.org/details/MessengersOfDeceptionUFOContactsAndCultsJacquesValle1979/mode/2up 69. https://www.amazon.com/s?k=0915904381 70. https://www.google.com/search?q=isbn+0915904381 71. https://catalog.loc.gov/vwebv/search?searchArg=0915904381

(The below text version of the notes is for search purposes and convenience. See the PDF version for proper formatting such as bold, italics, etc., and graphics where applicable. Copyright: 2022 Retraice, Inc.)

Re59: Learning, Interacting, Conclusions (AIMA4e chpts. 19-28)

retraice.com

Artificial Intelligence: A Modern Approach, 4th ed., pages 651-1022.

Air date: Wednesday, 23rd Nov. 2022, 11:00 PM Eastern/US.

Day 3

The latest version of our notes on AIMA4e (Russell & Norvig (2020)) will be at: https://aima4e.retraice.com.

Older versions will be preserved in the Retraice Notes feed, https://notes.retraice.com, or at the page for this segment, https://www.retraice.com/segments/re59.

Section V -- Learning

Chapters 19-22.

Section VI -- Interacting

Chapters 23-26.

On "whether submarines can swim": Dijkstra (1984).

Section VII -- Conclusions

Chapters 27-28.

_

References

Dijkstra, E. W. (1984). The threats to computing science. Delivered at the ACM 1984 South Central Regional Conference, November 16-18, Austin, Texas. https://www.cs.utexas.edu/~EWD/transcriptions/EWD08xx/EWD898.html Retrieved 24th Nov. 2022.

Russell, S., & Norvig, P. (2020). Artificial Intelligence: A Modern Approach. Pearson, 4th ed. ISBN: 978-0134610993. Searches: https://www.amazon.com/s?k=978-0134610993 https://www.google.com/search?q=isbn+978-0134610993 https://lccn.loc.gov/2019047498

(The below text version of the notes is for search purposes and convenience. See the PDF version for proper formatting such as bold, italics, etc., and graphics where applicable. Copyright: 2022 Retraice, Inc.)

AIMA4e Annotations

A companion to the great white brick.

As of November 23, 2022

(Start date: November 21, 2022.)

[1]retraice.com

Version notes: [2]Retraice ([3]2022/11/21) (Re57), first draft; [4]Retraice ([5]2022/11/22) (Re58), no footnotes; [6]Retraice ([7]2022/11/22) (Re58) again, moved some notes from Re57 and Re58 notes to footnotes here.

PREFACE

* The phenomenon: intelligent agents[8]^1 * The discipline: artificial intelligence,[9]^2 "the study of agents that receive percepts from the environment and perform actions." (vii)

* Aspects of the phenomenon: + Agent function: "Each ...agent implements a function that maps percept sequences to actions" (vii) o Ways to represent agent functions include: "reactive agents, real-time planners, decision-theoretic systems, and deep learning systems." (vii) + Learning o "a construction method for competent systems" (viii) o "a way of extending the reach of the designer into unknown environments." (viii) + Goals o Robotics and vision: # "not ...independently defined problems" # "[things] in the service of achieving goals."

I INTELLIGENCE --"Artificial Intelligence"

1 Intro:

definitions, foundations, history, philosophy, state of the art, risks-benefits

2 Agents:

environments, `good' behavior, agent structure and types

II SOLVING--"Problem-solving"

3 Searching: Looking ahead to find a sequence.

Algorithms, strategies, informed/heuristic[10]^3 strategies.

4 Complex Environments: More realistic environments.

Local search, optimization, continuous spaces, nondeterministic actions, partially observable env.s, online search and unknown env.s.

5 Adversarial Games: Other agents competing against us.

Theory, optimal decisions, alpha-beta tree search, Monte Carlo tree search, stochastic g.s, partially observable g.s, limitations.

6 Constraint Satisfaction Problems: States as domains, solutions as allowable combinations of states.

Constraint propagation, inference, backtracking search, local search, structure of problems

III THINKING--"Knowledge, reasoning, and planning"

7 Logical Agents: Forming representations and reasoning before acting.

Knowledge-based agents; representing[11]^4 worlds; logic, world models and `possible worlds';[12]^5 logic without objects.

8 First-Order Logic: A formal language for objects and their relations.

`Ontological commitment' (what is assumed about reality); syntax, semantics; knowledge engineering (building formal representations of important[13]^6 objects and relations in a domain).

9 First-Order Inference: Reasoning about objects and their relations.

Algorithms to answer any 1st-order logic question.

10 Knowledge Representation: Representing the real world for problem solving.

What content to put into a knowledge base.

Knowledge representation languages and their uses (315): * First-order logic: reasoning about a world of objects and relations; * Hierarchical task networks: for reasoning about plans (chpt. 11); * Bayesian networks: for reasoning with uncertainty (chpt. 13); * Markov models: for reasoning over time (chpt. 17); * Deep neural networks: for reasoning about images, sounds, other data (chpt. 21).

11 Automated Planning: Hierarchical task networks.

Planning for spacecraft, factories, military campaigns; representing actions and states; efficient algorithms and heuristics.

IV UNCERTAINTY--"Uncertain knowledge and reasoning"

12 Quantifying Uncertainty: An answer to the laziness and ignorance that kill formal logic.

Causes of uncertainty are environment types (partially observable,[14]^7 nondeterministic, adversarial[15]^8 ); belief state grows big and unlikely fast (384); agents still need a way to act; absolute certainty is impossible;[16]^9 it comes down to importance, likelihood and degree of success (385-386).

Logic fails because laziness and ignorance; probability theory solves the qualification problem by summarizing the uncertainty.[17]^10 * Laziness: too much work to list everything, or use such a list; * Ignorance: (theoretical) there are no complete theories; (practical) we can never run all the tests.

13 Probabilistic Reasoning [big]: Bayesian networks.

For reasoning with uncertainty by representing causal independence (398) and conditional independence (401) relationships to simplify probabilistic representations of the world.

14 Probabilistic Reasoning Over Time: Comprehending the uncertain past, present and future.

[18]^11

Belief state plus transition model yields prediction (chpt 4, 7, 11); percepts and sensor model yield updated belief state; add probability theory to switch from possible states to probable states.[19]^12

15 Probabilistic Programming: Universal formal languages to represent any computable probability model, and they come with algorithms.

Using formal logic and traditional programming languages to represent probabilistic information.

16 Making Simple Decisions: Agents getting what they want in an uncertain world--as much as possible, on average.

Beliefs, desires; utility theory; utility functions; decision networks; the value of information (547);[20]^13 this chapter is concerned with one-shot or episodic decisions problems (as opposed to sequential) (cf. 562, below).

17 Making Complex Decisions: What to do today given decisions to be made tomorrow.

Sequential decision problems (as opposed to one-shot episodic, cf. above): the agent's utility depends on a sequence of decisions in stochastic (explicitly probabilistic (45)) and partially observable environments. Markov models (563; cf. 463) for reasoning over time (chpt. 17).

18 Multiagent Decision Making [big]: When there's more than one agent in the environment.

The nature of such environments and the strategies for problem-solving depend on the relationships between agents: non-cooperative and cooperative game theory; collective decision-making.

V LEARNING--"Machine learning"

19: 20: 21: deep neural networks: for reasoning about images, sounds, other data (chpt. 21). 22:

VI INTERACTING--"Communicating, perceiving, and acting"

23: 24: 25: 26:

VII CONCLUSIONS--"Conclusions"

27: 28:

__

References

Crawford, K. (2021). Atlas of AI: Power, Politics, and the Planetary Costs of Artificial Intelligence. Yale University Press. ISBN: 978-0300209570. Searches: [21]https://www.amazon.com/s?k=9780300209570 [22]https://www.google.com/search?q=isbn+9780300209570 [23]https://lccn.loc.gov/2020947842

Frankfurt, H. G. (1988). The Importance of What We Care About. Cambridge. ISBN: 978-0521336116. Searches: [24]https://www.amazon.com/s?k=978-0521336116 [25]https://www.google.com/search?q=isbn+978-0521336116 [26]https://lccn.loc.gov/87026941

Kahneman, D. (2011). Thinking, Fast and Slow. Farrar, Straus and Giroux. ISBN: 978-0374533557. Searches: [27]https://www.amazon.com/s?k=978-0374533557 [28]https://www.google.com/search?q=isbn+978-0374533557 [29]https://lccn.loc.gov/2012533187

Retraice (2020/09/07). Re1: Three Kinds of Intelligence. retraice.com. [30]https://www.retraice.com/segments/re1 Retrieved 22nd Sep. 2020.

Retraice (2020/11/10). Re13: The Care Factor. retraice.com. [31]https://www.retraice.com/segments/re13 Retrieved 10th Nov. 2020.

Retraice (2020/11/25). Re15: Trust and Sources. retraice.com. [32]https://www.retraice.com/segments/re15 Retrieved 28th Feb. 2022.

Retraice (2022/10/19). Re23: You Need a World Model. retraice.com. [33]https://www.retraice.com/segments/re23 Retrieved 20th Oct. 2022.

Retraice (2022/10/23). Re27: Now That's a World Model - WM4. retraice.com. [34]https://www.retraice.com/segments/re27 Retrieved 24th Oct. 2022.

Retraice (2022/10/31). Re36: Notes on Conspiracy. retraice.com. [35]https://www.retraice.com/segments/re36 Retrieved 4th Nov. 2022.

Retraice (2022/11/12). Re48: From Drugs to Mao to Money. retraice.com. [36]https://www.retraice.com/segments/re48 Retrieved 14th Nov. 2022.

Retraice (2022/11/16). Re52: Big Questions About AI. retraice.com. [37]https://www.retraice.com/segments/re52 Retrieved 17th Nov. 2022.

Retraice (2022/11/21). Re57: AI, Agents, Problem-solving, Searching, Environments, Games (AIMA4e chpts. 1-6). retraice.com. [38]https://www.retraice.com/segments/re57 Retrieved 22nd Nov. 2022.

Retraice (2022/11/22). Re58: Thinking and Uncertainty (AIMA4e chpts. 7-18). retraice.com. [39]https://www.retraice.com/segments/re58 Retrieved 23rd Nov. 2022.

Russell, B. (1948). Human Knowledge: Its Scope and Limits. Routledge. First published in 1948. This edition 1992. ISBN: 0415083028. Searches: [40]https://archive.org/search.php?query=Human%20Knowledge%3A%20Its%20Scope%20and%20Limits [41]https://www.amazon.com/s?k=0415083028 [42]https://www.google.com/search?q=isbn+0415083028 [43]https://lccn.loc.gov/94209784

Vallee, J. (1979). Messengers of Deception: UFO Contacts and Cults. And/Or Press. ISBN: 0915904381. Different edition and searches: [44]https://archive.org/details/MessengersOfDeceptionUFOContactsAndCultsJacquesValle1979/mode/2up [45]https://www.amazon.com/s?k=0915904381 [46]https://www.google.com/search?q=isbn+0915904381 [47]https://catalog.loc.gov/vwebv/search?searchArg=0915904381

References

1. https://retraice.com/ 2. XReSeg57 3. XReSeg57 4. XReSeg58 5. XReSeg58 6. XReSeg58 7. XReSeg58 8. fn1x0 9. fn2x0 10. fn3x0 11. fn4x0 12. fn5x0 13. fn6x0 14. fn7x0 15. fn8x0 16. fn9x0 17. fn10x0 18. fn11x0 19. fn12x0 20. fn13x0 21. https://www.amazon.com/s?k=9780300209570 22. https://www.google.com/search?q=isbn+9780300209570 23. https://lccn.loc.gov/2020947842 24. https://www.amazon.com/s?k=978-0521336116 25. https://www.google.com/search?q=isbn+978-0521336116 26. https://lccn.loc.gov/87026941 27. https://www.amazon.com/s?k=978-0374533557 28. https://www.google.com/search?q=isbn+978-0374533557 29. https://lccn.loc.gov/2012533187 30. https://www.retraice.com/segments/re1 31. https://www.retraice.com/segments/re13 32. https://www.retraice.com/segments/re15 33. https://www.retraice.com/segments/re23 34. https://www.retraice.com/segments/re27 35. https://www.retraice.com/segments/re36 36. https://www.retraice.com/segments/re48 37. https://www.retraice.com/segments/re52 38. https://www.retraice.com/segments/re57 39. https://www.retraice.com/segments/re58 40. https://archive.org/search.php?query=Human%20Knowledge%3A%20Its%20Scope%20and%20Limits 41. https://www.amazon.com/s?k=0415083028 42. https://www.google.com/search?q=isbn+0415083028 43. https://lccn.loc.gov/94209784 44. https://archive.org/details/MessengersOfDeceptionUFOContactsAndCultsJacquesValle1979/mode/2up 45. https://www.amazon.com/s?k=0915904381 46. https://www.google.com/search?q=isbn+0915904381 47. https://catalog.loc.gov/vwebv/search?searchArg=0915904381

(The below text version of the notes is for search purposes and convenience. See the PDF version for proper formatting such as bold, italics, etc., and graphics where applicable. Copyright: 2022 Retraice, Inc.)

Re58: Thinking and Uncertainty (AIMA4e chpts. 7-18)

retraice.com

AIMA4e pages 208-650.

Air date: Tuesday, 22nd Nov. 2022, 11:00 PM Eastern/US.

Day 2

The latest version of our notes on AIMA4e (Russell & Norvig (2020)) will be at: https://aima4e.retraice.com.

Older versions will be preserved in the Retraice Notes feed, https://notes.retraice.com, or at the page for this segment, https://www.retraice.com/segments/re58.

Also mentioned: Gawande (2009).

Section III -- Thinking

Chapters 7-11.

Section IV -- Uncertainty

Chapters 12-18. This is the largest section of the book.

Chapters 13 (Probabilistic Reasoning) and 18 (Multiagent Decision Making) are the largest chapters within section IV.

_

References

Gawande, A. (2009). The Checklist Manifesto: How to Get Things Right. Henry Holt and Co., Kindle ed. ISBN: 978-0805091748. Searches: https://www.amazon.com/s?k=978-0805091748 https://www.google.com/search?q=isbn+978-0805091748 https://lccn.loc.gov/2009046888

Russell, S., & Norvig, P. (2020). Artificial Intelligence: A Modern Approach. Pearson, 4th ed. ISBN: 978-0134610993. Searches: https://www.amazon.com/s?k=978-0134610993 https://www.google.com/search?q=isbn+978-0134610993 https://lccn.loc.gov/2019047498

(The below text version of the notes is for search purposes and convenience. See the PDF version for proper formatting such as bold, italics, etc., and graphics where applicable. Copyright: 2022 Retraice, Inc.)

AIMA4e Annotations

A companion to the great white brick.

As of November 23, 2022

(Start date: November 21, 2022.)

[1]retraice.com

PREFACE

* The phenomenon: intelligent agents * The discipline: artificial intelligence "the study of agents that receive percepts from the environment and perform actions." (vii)

* Aspects of the phenomenon: + Agent function: "Each ...agent implements a function that maps percept sequences to actions" (vii) o Ways to represent agent functions include: "reactive agents, real-time planners, decision-theoretic systems, and deep learning systems." (vii) + Learning o "a construction method for competent systems" (viii) o "a way of extending the reach of the designer into unknown environments." (viii) + Goals o Robotics and vision: # "not ...independently defined problems" # "[things] in the service of achieving goals."

I INTELLIGENCE --"Artificial Intelligence"

1 Intro:

definitions, foundations, history, philosophy, state of the art, risks-benefits

2 Agents:

environments, `good' behavior, agent structure and types

II SOLVING--"Problem-solving"

3 Searching: Looking ahead to find a sequence.

Algorithms, strategies, informed/heuristic strategies.

4 Complex Environments: More realistic environments.

Local search, optimization, continuous spaces, nondeterministic actions, partially observable env.s, online search and unknown env.s.

5 Adversarial Games: Other agents competing against us.

Theory, optimal decisions, alpha-beta tree search, Monte Carlo tree search, stochastic g.s, partially observable g.s, limitations.

6 Constraint Satisfaction Problems: States as domains, solutions as allowable combinations of states.

Constraint propagation, inference, backtracking search, local search, structure of problems

III THINKING--"Knowledge, reasoning, and planning"

7 Logical Agents: Forming representations and reasoning before acting.

Knowledge-based agents; representing worlds; logic, world models and `possible worlds'; logic without objects.

8 First-Order Logic: A formal language for objects and their relations.

`Ontological commitment' (what is assumed about reality); syntax, semantics; knowledge engineering (building formal representations of important objects and relations in a domain).

9 First-Order Inference: Reasoning about objects and their relations.

Algorithms to answer any 1st-order logic question.

10 Knowledge Representation: Representing the real world for problem solving.

What content to put into a knowledge base.

Knowledge representation languages and their uses (315): * First-order logic: reasoning about a world of objects and relations; * Hierarchical task networks: for reasoning about plans (chpt. 11); * Bayesian networks: for reasoning with uncertainty (chpt. 13); * Markov models: for reasoning over time (chpt. 17); * Deep neural networks: for reasoning about images, sounds, other data (chpt. 21).

11 Automated Planning: Hierarchical task networks.

Planning for spacecraft, factories, military campaigns; representing actions and states; efficient algorithms and heuristics.

IV UNCERTAINTY--"Uncertain knowledge and reasoning"

12 Quantifying Uncertainty: An answer to the laziness and ignorance that kill formal logic.

Causes of uncertainty are environment types (partially observable, nondeterministic, adversarial); belief state grows big and unlikely fast (384); agents still need a way to act; absolute certainty is impossible; it comes down to importance, likelihood and degree of success (385-386).

Logic fails because laziness and ignorance; probability theory solves the qualification problem by summarizing the uncertainty. * Laziness: too much work to list everything, or use such a list; * Ignorance: (theoretical) there are no complete theories; (practical) we can never run all the tests.

13 Probabilistic Reasoning [big]: Bayesian networks.

For reasoning with uncertainty by representing causal independence (398) and conditional independence (401) relationships to simplify probabilistic representations of the world.

14 Probabilistic Reasoning Over Time: Comprehending the uncertain past, present and future.

Belief state plus transition model yields prediction (chpt 4, 7, 11); percepts and sensor model yield updated belief state; add probability theory to switch from possible states to probable states.

15 Probabilistic Programming: Universal formal languages to represent any computable probability model, and they come with algorithms.

Using formal logic and traditional programming languages to represent probabilistic information.

16 Making Simple Decisions: Agents getting what they want in an uncertain world--as much as possible, on average.

Beliefs, desires; utility theory; utility functions; decision networks; the value of information (547); this chapter is concerned with one-shot or episodic decsions problems (as opposed to sequential) (cf. 562).

17 Making Complex Decisions: What to do today given decisions to be made tomorrow.

Sequential decision problems (as opposed to one-shot episodic): the agent's utility depends on a sequence of decisions in stochastic (explicitly probabilistic (45)) and partially observable environments. Markov models (563; cf. 463) for reasoning over time (chpt. 17).

18 Multiagent Decision Making [big]: When there's more than one agent in the environment.

The nature of such environments and the strategies for problem-solving depend on the relationships between agents: non-cooperative and cooperative game theory; collective decision-making.

V LEARNING--"Machine learning"

19: 20: 21: deep neural networks: for reasoning about images, sounds, other data (chpt. 21). 22:

VI INTERACTING--"Communicating, perceiving, and acting"

23: 24: 25: 26:

VII CONCLUSIONS--"Conclusions"

27: 28:

__

References

1. https://retraice.com/

(The below text version of the notes is for search purposes and convenience. See the PDF version for proper formatting such as bold, italics, etc., and graphics where applicable. Copyright: 2022 Retraice, Inc.)

Re57: AI, Agents, Problem-solving, Searching, Environments, Games (AIMA4e chpts. 1-6)

retraice.com

AIMA4e pages 1-207.

Air date: Monday, 21st Nov. 2022, 11:00 PM Eastern/US.

Day 1

The latest version of our notes on AIMA4e (Russell & Norvig (2020)) will be at: https://aima4e.retraice.com.

Older versions will be preserved in the Retraice Notes feed, https://notes.retraice.com, or at the page for this segment, https://www.retraice.com/segments/re57.

Our nouns for the seven sections of AIMA4e

AIMA4e's noun phrases are in quotes. * I INTELLIGENCE--"Artificial Intelligence" * II SOLVING--"Problem-solving" * III THINKING--"Knowledge, reasoning, and planning" * IV UNCERTAINTY--"Uncertain knowledge and reasoning" * V LEARNING--"Machine learning" * VI INTERACTING--"Communicating, perceiving, and acting" * VII CONCLUSIONS--"Conclusions"

The Preface: phenomenon, science, definitions, hacks

The phenomenon about which AI scientists and engineers are concerned, per AIMA4e, is `intelligent agents'. It is easy to forget that this phenomenon goes far beyond code and computing.

The science of this phenomenon is called `artificial intelligence'.

Remember, definitions do work, often without our noticing, and sometimes to the detriment of our understanding.^1 Hacks will forget this easily and collapse `AI' into typa typa, looka looka. Most of us are hacks.

Chapters 1-6

See https://www.retraice.com/segments/re57. On heuristics and biases, see Kahneman (2011); cf. Russell & Norvig (2020) chapter 3.

_

References

Crawford, K. (2021). Atlas of AI: Power, Politics, and the Planetary Costs of Artificial Intelligence. Yale University Press. ISBN: 978-0300209570. Searches: https://www.amazon.com/s?k=9780300209570 https://www.google.com/search?q=isbn+9780300209570 https://lccn.loc.gov/2020947842

Kahneman, D. (2011). Thinking, Fast and Slow. Farrar, Straus and Giroux. ISBN: 978-0374533557. Searches: https://www.amazon.com/s?k=978-0374533557 https://www.google.com/search?q=isbn+978-0374533557 https://lccn.loc.gov/2012533187

Russell, S., & Norvig, P. (2020). Artificial Intelligence: A Modern Approach. Pearson, 4th ed. ISBN: 978-0134610993. Searches: https://www.amazon.com/s?k=978-0134610993 https://www.google.com/search?q=isbn+978-0134610993 https://lccn.loc.gov/2019047498

Footnotes

^1 Crawford (2021) p. 7 ff.

(The below text version of the notes is for search purposes and convenience. See the PDF version for proper formatting such as bold, italics, etc., and graphics where applicable. Copyright: 2022 Retraice, Inc.)

AIMA4e Annotations

A companion to the great white brick.

As of November 22, 2022

(Start date: November 21, 2022.)

[1]retraice.com

PREFACE

* The phenomenon: intelligent agents * The discipline: artificial intelligence "the study of agents that receive percepts from the environment and perform actions." (vii)

* Aspects of the phenomenon: + Agent function: "Each ...agent implements a function that maps percept sequences to actions" (vii) o Ways to represent agent functions include: "reactive agents, real-time planners, decision-theoretic systems, and deep learning systems." (vii) + Learning o "a construction method for competent systems" (viii) o "a way of extending the reach of the designer into unknown environments." (viii) + Goals o Robotics and vision: # "not ...independently defined problems" # "[things] in the service of achieving goals."

I INTELLIGENCE --"Artificial Intelligence"

1 Intro:

definitions, foundations, history, philosophy, state of the art, risks-benefits

2 Agents:

environments, `good' behavior, agent structure and types

II SOLVING--"Problem-solving"

3 Searching: looking ahead to find a sequence:

algorithms, strategies, informed/heuristic strategies

4 Complex environments: more realistic environments:

local search, optimization, continuous spaces, nondeterministic actions, partially observable env.s, online search and unknown env.s

5 Adversarial games: other agents competing against us:

theory, optimal decisions, alpha-beta tree search, Monte Carlo tree search, stochastic g.s, partially observable g.s, limitations

6 Constraint satisfaction problems: states as domains, solutions as allowable combinations of states

constraint propagation, inference, backtracking search, local search, structure of problems

III THINKING--"Knowledge, reasoning, and planning"

7 algorithms to answer any 1st-order logic question

IV UNCERTAINTY--"Uncertain knowledge and reasoning"

V LEARNING--"Machine learning"

VI INTERACTING--"Communicating, perceiving, and acting"

VII CONCLUSIONS--"Conclusions"

__

References

1. https://retraice.com/

(The below text version of the notes is for search purposes and convenience. See the PDF version for proper formatting such as bold, italics, etc., and graphics where applicable. Copyright: 2022 Retraice, Inc.)

Re56: A Valuable Brick: Artificial Intelligence: A Modern Approach 4th ed.

Retraice^1

This is how we got from Re1 to AIMA4e.

Air date: Sunday, 20th Nov. 2022, 11:00 PM Eastern/US.

Findings

How we got to computer control (roughly the same as `AI') and players. o Retraice is about `What's going on out there'. o Our point of departure was the concept of intelligence--natural, artificial and strategic.^2 o We found that computer control is what's going on out there, the `current history' of our time.^3 o We found that computer control is strategic, like a game.^4 o We found that games are player-oriented.^5

Implications

Why we're changing modes of thinking.^6 * Mode 0 is `don't care', i.e. not knowing what happened. * Mode 1 is `don't know', i.e. wondering what happened (the previous Retraice mode). * Mode 2 is `don't doubt', i.e. making things happen.

In mode 2, it's `good' to do AI for RTFM reasons,^7 and for MICE reasons.^8

Details

What is AIMA4e? * The generalist compendium of AI knowledge is in this white brick.^9 + It might be the most valuable object for doing things in the 2020s. + It is our new point of departure. * But we will not be teaching the book! We are not qualified. Some learning might nevertheless happen.

Plan

The plan is to read the whole book in a week, as a first-pass. Thereafter, we'll layer more and more detail as we see fit.

_

References

Retraice (2020/09/07). Re1: Three Kinds of Intelligence. retraice.com. https://www.retraice.com/segments/re1 Retrieved 22nd Sep. 2020.

Retraice (2020/09/10). Re4: Trust No One. retraice.com. https://www.retraice.com/segments/re4 Retrieved 22nd Sep. 2020.

Retraice (2022/10/19). Re22: Computer Control. retraice.com. https://www.retraice.com/segments/re22 Retrieved 19th Oct. 2022.

Retraice (2022/10/24). Re28: What's Good? RTFM. retraice.com. https://www.retraice.com/segments/re28 Retrieved 25th Oct. 2022.

Retraice (2022/11/18). Re54: Implications and Endgames. retraice.com. https://www.retraice.com/segments/re54 Retrieved 19th Nov. 2022.

Retraice (2022/11/19). Re55: The Computer Control Game. retraice.com. https://www.retraice.com/segments/re55 Retrieved 20th Nov. 2022.

Russell, S., & Norvig, P. (2020). Artificial Intelligence: A Modern Approach. Pearson, 4th ed. ISBN: 978-0134610993. Searches: https://www.amazon.com/s?k=978-0134610993 https://www.google.com/search?q=isbn+978-0134610993 https://lccn.loc.gov/2019047498

Footnotes

^1 https://www.retraice.com/retraice

^2 Retraice (2020/09/07).

^3 Retraice (2022/10/19).

^4 Retraice (2022/11/19).

^5 Retraice (2022/11/18).

^6 Retraice (2022/11/19).

^7 `Right, true, fit and mixed.' Retraice (2022/10/24).

^8 `Money, ideology, compromise, ego.' Retraice (2020/09/10).

^9 Russell & Norvig (2020). Of course, it is not the only valuable AI book, and it certainly is not a specialist's book. It is a generalist book for someone who intends to specialize in AI.

(The below text version of the notes is for search purposes and convenience. See the PDF version for proper formatting such as bold, italics, etc., and graphics where applicable. Copyright: 2022 Retraice, Inc.)

Re55: The Computer Control Game

Retraice^1

It's like football--no, poker--no, chess--no, war.

Air date: Saturday, 19th Nov. 2022, 11:00 PM Eastern/US.

Is war a game?

The rise of computer control (AI, more or less^2) is happening at break-neck speed. If we agree that it, like other major events in history, is player-oriented,^3 it seems like a player-oriented game being played by players. But it also seems like a war.

What is victory? It's not going to be like sports, or chess or poker. It's going to be like war. And now, our time, feels like the lead-up to war. It seems like a game-theory sort of game, like the nuclear arms race was.^4

A singleton is bad, probably

Bostrom gives two definitions of `singleton':^5 o "[A] world order in which there is at the global level a single decision-making agency."^6 o "[A] sufficiently internally coordinated political structure with no external opponents."^7

The question will be whether a singleton is better or worse than the alternatives, not `bad' or `good'.^8

But a (new) singleton is game victory

A singleton might be what constitutes `victory'. And humanity is already a singleton over the Earth, "a single decision-making agency" "with no external opponents". So the players would be playing to create and/or join the new, computer-control-based singleton. Humans are the top of the food chain, literally and figuratively, in every way that we `eat' on this planet. There's nothing in the Amazon, or in the oceans, that can fight back against us.

But not all humans are part of the "decision-making agency" within humanity.

(Bostrom says we're not a singleton.^9)

If we are a singleton

We are, in a sense, multiple competing superintelligences, in the form of coalitions.^10 If we are a singleton, than we are, in a sense, a "multi-polar scenario": "[A] post-transition society with multiple competing superintelligent agencies."^11

I.e. we might already be effectively `in' one of the scenarios that Bostrom and others are anticipating. But they expect it in the future, as the result of future technology.^12

By a certain set of definitions, we're already in a singleton, and the players in the game are playing to win the next one.^13

Three modes of thinking

* Mode 0: Don't care (`don't know what happened'); * Mode 1: Don't know (`wonder what happened'); This is Retraice, so far. * Mode 2: Don't doubt (`make things happen'). This is `player' mode.

__

References

Bostrom, N. (2014). Superintelligence: Paths, Dangers, Strategies. Oxford. First published in 2014. Citations are from the pbk. edition, 2016. ISBN: 978-0198739838. Searches: https://www.amazon.com/s?k=978-0198739838 https://www.google.com/search?q=isbn+978-0198739838 https://lccn.loc.gov/2015956648

Brockman, J. (Ed.) (2019). Possible Minds: Twenty-Five Ways of Looking at AI. Penguin. ISBN: 978-0525557999. Searches: https://www.amazon.com/s?k=978-0525557999 https://www.google.com/search?q=isbn+978-0525557999 https://lccn.loc.gov/2018032888

Crawford, K. (2021). Atlas of AI: Power, Politics, and the Planetary Costs of Artificial Intelligence. Yale University Press. ISBN: 978-0300209570. Searches: https://www.amazon.com/s?k=9780300209570 https://www.google.com/search?q=isbn+9780300209570 https://lccn.loc.gov/2020947842

Lee, K.-F. (2018). AI Superpowers: China, Silicon Valley, and the New World Order. Houghton Mifflin Harcourt. ISBN: 978-1328546395. Searches: https://www.amazon.com/s?k=9781328546395 https://www.google.com/search?q=isbn+9781328546395 https://catalog.loc.gov/vwebv/search?searchArg=9781328546395

McLuhan, M. (1964). Understanding Media: The Extensions of Man. Gingko Press. ISBN: 1584230738. Originally published 1964. This ed. 2003. https://archive.org/details/understandingmed0000mclu_n3p7 Searches: https://www.amazon.com/s?k=1584230738 https://www.google.com/search?q=isbn+1584230738 https://lccn.loc.gov/2003012174

Retraice (2022/10/19). Re22: Computer Control. retraice.com. https://www.retraice.com/segments/re22 Retrieved 19th Oct. 2022.

Retraice (2022/11/07). Re43: The Midterms -- Part 1. retraice.com. https://www.retraice.com/segments/re43 Retrieved 9th Nov. 2022.

Retraice (2022/11/18). Re54: Implications and Endgames. retraice.com. https://www.retraice.com/segments/re54 Retrieved 19th Nov. 2022.

Weizenbaum, J. (1976). Computer Power and Human Reason: From Judgment to Calculation. W. H. Freeman and Company. ISBN: 0716704633. Also available at: https://archive.org/details/computerpowerhum0000weiz

Wiener, N. (1954). The Human Use Of Human Beings: Cybernetics and Society. Da Capo, 2nd ed. ISBN: 978-0306803208. This 1954 ed. missing `The Voices of Rigidity' chapter of the original 1950 ed. See 1st ed.: https://archive.org/details/humanuseofhumanb00wien/page/n11/mode/2up. See also Brockman (2019) p. xviii. Searches for the 2nd ed.: https://www.amazon.com/s?k=9780306803208 https://www.google.com/search?q=isbn+9780306803208 https://lccn.loc.gov/87037102

Yudkowsky, E. (2013). Intelligence explosion microeconomics. Machine Intelligence Research Institute. Technical report 2013-1. https://intelligence.org/files/IEM.pdf Retrieved ca. 9th Dec. 2018.

Footnotes

^1 https://www.retraice.com/retraice

^2 Retraice (2022/10/19). Cf. Crawford (2021) pp. 8-9 on the "discomfort in the computer science community" with the term `artificial intelligence', as opposed to, now, `machine learning'.

^3 Retraice (2022/11/18).

^4 See Lee (2018) on the competition in AI between the U.S. and China.

^5 He also gives a more broad definition, in a glossary (p. 409), that includes `world democracy' and `world dictator', which we didn't mention during the livestream.

^6 Bostrom (2014) p. 96.

^7 Bostrom (2014) p. 124.

^8 Bostrom (2014) p. 109.

^9 Bostrom (2014) p. 125.

^10 See Retraice (2022/11/07) on coalitions.

^11 Bostrom (2014) p. 194.

^12 Bostrom (2014); Yudkowsky (2013). Cf. Weizenbaum (1976) and Wiener (1954), and perhaps McLuhan (1964), who focus more on the human aspect of the rise of computer control.

^13 There is an apparent contradiction, if we define a singleton as permanent or unchanging. But why do that? Singletons should not be defined to be eternal and static.

(The below text version of the notes is for search purposes and convenience. See the PDF version for proper formatting such as bold, italics, etc., and graphics where applicable. Copyright: 2022 Retraice, Inc.)

Re54: Implications and Endgames

Retraice^1

WAAIT Part 5: There are constraints on what we can believe and what we should do.

Air date: Friday, 18th Nov. 2022, 11:00 PM Eastern/US.

The big questions we've asked

o What about AI though? o What is AI again? o How does AI mix with other things? o How much do I matter, individually? o As a species: What are we doing? Where are we headed? o As players:^2 How to win? o What can't we see? o What about cumulative progress? o What about superpowers and secrecy?

Some answers are more clear than others

* AI is 3D: space, time, value. * AI is hard to define. * AI is already mixed in, and will only become more so. * We could be headed for great or terrible things. * Ignorant, incapable individuals will not affect where we're going. Think yap, yap, yap vs. typa typa typa. * Players in the game will focus on strategies and tactics. * We're already living inside a world shaped by `strategic intelligence' (espionage). * Secrets have been kept. * Secret progress would be cumulative. * AI is a new superpower in the mix.

auto

Belief and action implications

Belief implications:^3 * records and claims are evidence, but... * hard evidence is better: things that are true at all times, places and values:^4 Think math, logic, technology, artifacts, and to some extent empirical science.

Action implications: * Players: Cause good, prevent bad.^5 * Individuals: If you have a choice, choose your work wisely. Outside forces in the form of coalitions will, to greater and lesser extents, affect the game. But there are many positions in the world that will not be part of any such force. * Humanity: Do or die,^6 and prevent a singleton:^7 We won't be humanity if we're cells in a singleton, beholden to a larger entity. Alternatively, if a singleton is necessary or unavoidable, aim for a happy healthy one?

Endgames: skeptical optimism; player orientation

* AI is trying to be great. The propaganda video series put out by DeepMind recently, called `AI by me', is, even through a true skeptic's eyes, inspiring and hopeful stuff. * We can blow it. * Players win games. * It's going to be hard.

_

References

Bostrom, N. (2014). Superintelligence: Paths, Dangers, Strategies. Oxford. First published in 2014. Citations are from the pbk. edition, 2016. ISBN: 978-0198739838. Searches: https://www.amazon.com/s?k=978-0198739838 https://www.google.com/search?q=isbn+978-0198739838 https://lccn.loc.gov/2015956648

Ord, T. (2020). The Precipice: Existential Risk and the Future of Humanity. Hachette. ISBN: 978-0316484916. Searches: https://www.amazon.com/s?k=978-0316484916 https://www.google.com/search?q=isbn+978-0316484916 https://lccn.loc.gov/2019956459

Rees, M. (2003). Our Final Hour: A Scientist's Warning. Basic Books. ISBN: 0465068634. Searches: https://www.amazon.com/s?k=0465068634 https://www.google.com/search?q=isbn+0465068634 https://lccn.loc.gov/2004556001

Russell, S. (2019). Human Compatible: Artificial Intelligence and the Problem of Control. Viking. ISBN: 978-0525558613. Searches: https://www.amazon.com/s?k=978-0525558613 https://www.google.com/search?q=isbn+978-0525558613 https://lccn.loc.gov/2019029688

Footnotes

^1 https://www.retraice.com/retraice

^2 Players: the people in the DeepMind and OpenAI buildings in the rooms with the whiteboards. They're the players.

^3 Belief is a matter of degree, and it's not obvious that we can choose our beliefs, if we have control over them at all.

^4 The problem of deciding what's `true' of human values is a huge topic. See: Russell (2019) chpt. 9; Bostrom (2014) chpt. 12. The main point is that the people who happen to be working at the frontiers of the most important technologies are competent at that work, but likely not at representing humanity collectively, especially morally and ethically.

^5 This is something like the Hippocratic Oath.

^6 See Ord (2020); Rees (2003). Side note: `Environmentalism' should account for the fact that computer control, AI, and also things like the solar system are part of the human environment, and therefore should be taken very seriously as threats to our future. But this is really a matter of definition.

^7 "[A] world order in which there is at the global level a single decision-making agency"^8