AI Research Paradigm

A AI Research Paradigm is a research paradigm that focuses on the strategies and methodologies used in artificial intelligence (AI) research.

• Context:
  • It can (typically) involve a set of assumptions, values, and practices that dictate how AI research is conducted.
  • It can (often) include the debate between leveraging human knowledge and relying on increasing computation power.
  • ...
• Example(s):
  • A Computation-based methods AI Research Paradigm.
  • A Human-knowledge-based approaches in AI, as discussed in Richard S. Sutton's 2019 article, "The Bitter Lesson."
  • ...
• Counter-Example(s):
  • The Scientific Method as a more general research paradigm that applies across various fields of science, not specifically tailored to the nuances and challenges of AI research.
• See: AI Research History, Computation Power, Human Knowledge in AI, Moore's Law.

---

---

References

2019

• (Sutton, 2019) ⇒ Richard S. Sutton. (2019). “The Bitter Lesson.” Blog post
  • QUOTE: "The biggest lesson that can be read from 70 years of AI research is that general methods that leverage computation are ultimately the most effective, and by a large margin."
  • QUOTE: "Most AI research has been conducted as if the computation available to the agent were constant ... but, over a slightly longer time than a typical research project, massively more computation inevitably becomes available."
  • QUOTE: "The only thing that matters in the long run is the leveraging of computation."
  • QUOTE: "There were many examples of AI researchers' belated learning of this bitter lesson, and it is instructive to review some of the most prominent."

---