Open Problems: Robustness, Alignment, Long-Horizon Memory
Still unsolved: agents that don't break on edge cases, that stay aligned, and that remember well over months.
The Frontier Is Still Frontier
Despite huge progress, fundamental agent problems remain unsolved. Awareness of these helps you scope what is realistic to build today.
Open Problem 1: Robustness
Agents that work 99% of the time still fail catastrophically on edge cases:
- Prompt injection attacks succeed too often
- Tools fail in novel ways
- Long-horizon tasks compound small errors
- Distribution shift breaks production agents weekly
All lessons in this course
- Agentic Reasoning (o1, o3, Reasoning Models)
- Hybrid Symbolic + Neural Agents
- Multimodal Agents (Vision + Voice + Action)
- Open Problems: Robustness, Alignment, Long-Horizon Memory