ai

History §

origins: cybernetics, information science

• birth: 1940s
• first AI winter: 1974
• resurgence: 1980s
• second AI winter: 1987
• big data allows DL takeoff: 2005

Subproblems §

• reasoning and problem solving
• knowledge representation
• planning and decision making
• learning
• NLP
• perception (computer vision)
• social intelligence (affective computing)
• general intelligence (AGI)

Techniques §

• search and optimization
  • state space search
  • local search
• logic
• probabilistic methods
• classifiers and statistical learning
• artificial neural networks

Subfields §

• machine learning
  • deep learning
• AI safety

Camps of people §

roughly:

• E/acc vs doomerism
• “AGI by 2028, no doubt about it” vs “AGI? Hard to say.”
• “scale with some architecture tweaks will get us to AGI” vs “we need neuro-symbolic approaches”

notable people:

• Geoffrey Hinton: “AI risks are to be taken seriously”
• Gary Marcus: “deep learning is hitting a wall”