In this conversation, Vivek Juneja engages with Daniel and Jan, two researchers deeply involved in the ARC Challenge and the field of deep learning. They discuss their backgrounds, motivations for participating in the ARC Challenge, and the implications of the challenge for future AI research. The conversation covers the limitations of current language models, the importance of energy efficiency, and the role of benchmarks in AI. They also explore the startup culture in Europe, the challenges posed by regulations, and the exciting breakthroughs in AI research, particularly in reasoning and program synthesis. The discussion concludes with advice for newcomers in the field and reflections on staying updated in a rapidly evolving landscape.Takeaways1. Daniel started coding at eight and switched to computer science and Jan began with mathematics before moving to computer science.2. The ARC Challenge represents cognitive puzzles for AI. LLMs showed promising results in the ARC Challenge.3. Energy efficiency is a significant concern for AI models.4. Consumer interaction with LLMs often leads to over-reliance.5. Benchmarks in AI may lose their relevance over time.6. Startup culture in Europe lacks the drive seen in Silicon Valley.7. Regulations can hinder innovation in AI development.8. Exciting breakthroughs in reasoning models are on the horizon9. Starting with AI is not that hard, and is highly approachable 1366a