Professionals: Designing Challenging AI Prompts

Terac is conducting a paid AI research study to identify professionals and experienced users who are particularly skilled at creating challenging tasks that reveal the limitations of artificial intelligence models.

The study focuses on transforming real-world professional or personal workflows into complex AI prompts that require reasoning, research, and practical problem-solving. This initial paid trial will also help Terac identify strong candidates for potential long-term AI prompt engineering and evaluation opportunities.

How the Study Works

Participants will spend approximately one hour turning a complex workflow from their professional or personal experience into a demanding prompt for an AI model.

You will test the prompt using ChatGPT and analyze the response to identify mistakes, weaknesses, or areas where the model struggles. If the AI completes the task successfully, you will refine the prompt and increase its difficulty to uncover meaningful limitations.

After testing the prompt, you will create a detailed grading rubric that another person could use to objectively evaluate an AI model’s response to the same task.

The entire exercise will be screen-recorded because Terac is interested not only in the final submission but also in understanding how participants reason, evaluate results, and refine their tasks.

What You’ll Do

• Select a professional or personal workflow you understand extremely well.
• Transform the workflow into a challenging AI prompt requiring reasoning and real-world information.
• Test the prompt with ChatGPT and identify where the model makes mistakes or fails.
• Increase the difficulty of the prompt when necessary.
• Develop a comprehensive and easy-to-follow grading rubric.
• Share your screen, camera, and microphone while completing the study.
• Submit your final prompt, notes about the AI’s failures, grading rubric, and generated output file.

Who Should Apply

This opportunity is suitable for professionals, subject-matter experts, and experienced AI users who have strong knowledge of a particular professional or personal workflow.

Applicants should be able to quickly recognize whether an AI-generated response is accurate, incomplete, or incorrect.

You will also need:

• Deep familiarity with at least one specific workflow or area of expertise.
• Strong judgment when evaluating the accuracy and quality of AI-generated content.
• Access to a laptop or desktop computer.
• An active ChatGPT account.
• The ability and willingness to share your screen, camera, and microphone while completing the recorded exercise.

Participants who perform particularly well during the initial study may be considered for Terac’s longer-term pool of AI evaluators and experts.