Goals

We analyze the ‘big picture’ of how artificial intelligence can be utilized most effectively and safely to accelerate scientific, technological and societal progress.

Goals

Our major goals are to:

  • Achieve a novel and more complete conceptualization of this problem space
  • Identify gaps and bottlenecks and formulate mitigation strategies
  • E.g., missing knowledge, data, AI capabilities, benchmarks
    • Identify synergies
    • Synergies between groups and stakeholders (better connect domain needs to AI expertise)
  • Synergies between AI development tasks (transfer learning etc.)
  • Identify risks
    • Greater use of AI and greater AI capabilities come with various kinds of risks
    • Formulating mitigation strategies
  • Compare opportunities for radical change vs. incremental change
    • Power-law distribution of effects of innovations (only a few innovations responsible for most of progress)
    • Overall tractability and scalability substantially different between different routes of innovation
    • Example: AI for clinical decision support vs. drug development vs. basic biological research vs. biological research methodology development vs. multi-purpose tool for information retrieval and analysis
  • Build a community and inform policy decisions