Researcher, Robustness & Safety Training

Written by the T&S Network from OpenAI's posting.

Safety research on making OpenAI's deployed AI models follow nuanced safety policies without losing helpfulness, resist adversarial and malicious use, and handle privacy, security and safety-critical domains. It sits in the Safety Systems team, which works to get the company's most capable models deployed safely.

• AI misuse and jailbreaks

What you'd do

• Do frontier-level work on safety methods such as human-feedback tuning, adversarial training and robustness.
• Implement new safety methods in core model training and ship safety improvements into OpenAI's products.
• Set research directions and strategies to make AI systems safer, more aligned and more robust.
• Work with T&S, legal, policy and other research teams so products meet the company's safety standards.
• Evaluate the safety of models and systems, identify areas of risk and propose mitigation strategies.

What it asks for

• 4+ years of experience in AI safety, especially RLHF, adversarial training, robustness, or fairness and bias.
• Experience in safety work for AI model deployment.
• PhD or other degree in computer science, machine learning or a related field.
• Solid grounding in deep learning research, strong engineering, or a mix of both.

Worth knowing

• Background check administered under applicable law.

Terms in this role

RLHF.

Reinforcement learning from human feedback: training an AI model on people's ratings of its answers.

Our best summary of the posting, and it can be wrong. The employer's posting is what counts: read it in full on OpenAInorth_east before you apply. If we got something wrong, report this listing.