The speaker introduces her work at Aiven and outlines her two talks on the human side of AI and using AI to debug complex systems like Kafka.
LLMs tend to provide a single, conclusive answer instead of collaborating and offering multiple options, which is problematic for complex debugging and expert decision-making.
Over-relying on AI for cognitive tasks can lead to an unintentional delegation of thinking, causing developers to lose their expertise without realizing it.
Low-quality, AI-generated pull requests create "PR fatigue" for open source maintainers, highlighting the need to verify a contributor's understanding of their own code.
Current LLMs are designed for individual users and lack the necessary mechanisms for team collaboration, communication, and shared context, which are crucial for software engineering.
The best source of inspiration and new ideas comes from talking with colleagues and the community, using AI as a tool to ask deeper questions and understand different roles.
In an age of AI-generated content and remote work, in-person conferences are essential for building genuine community and finding people who can solve your problems.
Jobs that call for the skills explored in this talk.