This work feeds into a larger effort to evaluate and improve how artificial intelligence handles complex, multi-step requests. High performers in this trial assessment will be considered for ongoing paid research projects.

You will spend about an hour creating a demanding prompt based on a real workflow you know deeply. During a screen-recorded session, you will test your prompt in ChatGPT, identify where the model fails to deliver a complete file, and refine the request to make it harder. Finally, you will write a comprehensive grading rubric that someone else could use to evaluate any AI's attempt at your task. You will hand back your prompt, failure notes, rubric, and the generated file for review.

• Translate a familiar, complex workflow into a demanding prompt for an AI model.
• Test your prompt using ChatGPT to identify where the model fails to produce a complete file.
• Record your screen, camera, and microphone while thinking through the task.
• Write a clear, structured rubric that allows a stranger to grade the AI's response.

We welcome professionals, domain experts, and knowledge workers from any field who deeply understand complex, specialized workflows. You must be able to recognize a high-quality output in your area of expertise within seconds. We are looking for individuals who can creatively translate their daily tasks into challenging scenarios that require real reasoning and calculation.

• Deep knowledge of a specialized workflow from your job or personal life.
• Access to a laptop or desktop computer.
• Active ChatGPT account (the free tier is acceptable).
• Comfortable sharing a screen recording with audio and video during the task.

Work arrangement: Fully remote

• Your participation will not involve access to confidential or proprietary information from any employer, client, or institution.
• We are unable to support H1-B or STEM OPT candidates at this time.

• Payment is issued as soon as your work is approved.