In this roundtable discussion led by Melissa Fisher, engineering professionals Gary Hawkes, Adam Davis, and Scout Burghardt explore the risks of relying on AI for technical tasks without maintaining human critical thinking.
They shared 'horror stories' regarding AI-generated acceptance criteria and user stories that lack context or sense, leading to unusable tickets and wasted refinement time.
The conversation also covered the distinction between using AI as a tool for efficiency versus using it as a 'thinking partner' to achieve meaningful outcomes.
Key themes included the importance of maintaining human intent, the challenges of 'agentic' testing, and the need for developers and testers to remain close to the process to ensure AI outputs are scalable, accurate, and useful for human collaborators.
Session Chapters:
- 00:03 - Roundtable introductions
- 01:43 - The risk of bypassing thinking
- 06:59 - One-shotting and agentic testing
- 11:30 - Identifying root causes of failure
- 14:35 - Managing excessive AI output
- 16:52 - AI training and usage principles
- 22:33 - Defining human intent and ownership
- 32:16 - Combating AI slop in teams
- 39:08 - Autonomous code review quality gates
- 46:05 - Positive AI use cases
- 49:52 - AI as a side project enabler
- 52:06 - Embracing AI through hands-on experimentation
- 53:41 - Personal creative uses for image generation
- 55:04 - Developing test policies for AI chatbots
- 57:26 - Final thoughts
Comments