At SynapseMind, we help advance AI systems through human insight, structured evaluation, and responsible testing.
We are seeking analytical contributors who can examine how AI models respond when instructions become ambiguous, conflicting, or potentially unsafe.
Project Overview
AI systems must perform reliably beyond straightforward questions and routine interactions.
They also need to handle difficult requests, misleading instructions, conflicting requirements, and situations where following a user’s request could produce an unsafe result.
This project focuses on testing model behavior using controlled scenarios.
Contributors evaluate the responses produced by AI systems and identify weaknesses in their reasoning, instruction-following, or safety behavior.
The work follows defined testing procedures and authorized project boundaries.
What You Will Do
- Review predefined AI testing scenarios.
- Submit or evaluate challenging prompts within approved testing environments.
- Examine model responses for safety and instruction-following failures.
- Identify situations where a model provides inappropriate or contradictory guidance.
- Document the conditions that triggered a failure.
- Assess whether the model handled the scenario according to the supplied criteria.
Example task
An AI assistant receives a request that contains conflicting instructions.
You examine whether it follows the applicable instruction hierarchy, recognizes the conflict, and produces an appropriate response.
You then record the outcome using the project’s evaluation framework.
Requirements
- Strong analytical and problem-solving skills.
- Ability to identify weaknesses in AI-generated responses.
- Attention to detail when following testing procedures.
- Understanding of responsible AI use and safety principles.
- Ability to document observations clearly.
- An AI certificate from Walden University or AIDRAlabs.
- Reliable internet access and a suitable device.
Why This Project Matters
Testing AI systems against challenging scenarios helps developers identify weaknesses that may not appear during ordinary interactions.
Human evaluators contribute to the development of safer systems by documenting failures, assessing responses, and helping teams understand where additional safeguards may be needed.