At Rise of AI 2019 in Berlin, Marek Rosa, CEO and CTO of GoodAI, addresses how progress toward artificial general intelligence can be measured. According to the talk's description, he asks how success can be defined when working towards general artificial intelligence, what key milestones need to be reached along the way, and how those milestones can be reached in practice. The talk's chapters indicate it covers what intelligence is as a starting point for any benchmarking effort, followed by a discussion of AI training tasks as a way to test and track progress. No reliable transcript is available for this recording, so the detailed arguments, examples and specific benchmarks or tasks Rosa presents cannot be reconstructed here with confidence. The talk sits within GoodAI's broader public work on defining and measuring general intelligence in AI systems, a theme reflected in the talk's title and its stated focus on benchmarking rather than on a single model or application.
Key takeaways
- Marek Rosa's talk frames benchmarking as central to tracking progress toward artificial general intelligence.
- The talk poses the question of how to define success when working towards general AI.
- The talk addresses what key milestones need to be reached on the way to AGI.
- The talk covers what intelligence is, as a basis for setting benchmarks.
- The talk discusses AI training tasks as a component of benchmarking progress.
Questions answered in this talk
What is Marek Rosa's talk at Rise of AI 2019 about?
According to the talk description, Marek Rosa, CEO and CTO of GoodAI, discusses how to measure success when working towards general artificial intelligence and what key milestones need to be reached.
Who is Marek Rosa and what company does he represent at this talk?
Marek Rosa is CEO and CTO of GoodAI, and he gave this talk on benchmarking progress towards AGI at the Rise of AI conference 2019 in Berlin.