Big Picture
The idea of “thinking machines” is older than computers. But modern AI evolved through well-known milestones: early theory, symbolic AI, the first “AI winters,” and then the rise of machine learning and deep learning. This lesson gives you a clean mental timeline.
Timeline Highlights
Turing proposes a test for machine intelligence (“Can machines think?”), inspiring AI as a field.
Often called AI’s birth: researchers coin the term “Artificial Intelligence” and set early goals.
Rules and logic dominate: “expert systems” try to encode human knowledge directly as rules.
Funding and excitement drop when expectations exceed results. Important lesson: avoid over-promising.
IBM’s chess system shows narrow AI can rival world champions in specific tasks.
Focus shifts to learning from data (SVMs, ensembles). The web creates huge datasets.
ImageNet results (AlexNet) spark the deep learning wave; GPUs make training feasible.
AlphaGo defeats Lee Sedol; speech, vision, and NLP reach human-level performance in many tasks.
Translation, image generation, chat assistants, and recommendation systems are widely used in daily life.
Key Ideas to Remember
Symbolic vs. Learning
Symbolic AI uses rules; machine learning learns patterns from data. Modern AI blends both ideas.
AI Winters
When hype exceeds results, funding falls. Progress continues, but slower — cycles repeat.
Compute & Data
GPUs and large datasets enabled deep learning breakthroughs from 2012 onward.