The Artificial General Intelligence (AGI) Safety and Alignment Team (ASAT) aims to reduce existential and catastrophic risk from AGI and eventually Artificial Superintelligence (ASI). We research novel techniques and work with the rest of GDM and Google to apply them.
ASAT has sub-teams specialising in making future Geminis more thoroughly aligned by finding and fixing sources of misalignment and exploring alignment techniques with better generalization. Preparing for future AGI risks by simulating them today and using interpretability techniques to understand AI and solve practical problems like model forensics or eval awareness. Building control for GDM’s agents as defense-in-depth against potential misaligned internal deployments. Researching training techniques, like debate, for aligning superhuman AI and ways to retain, improve, and measure monitorability. Researching and implementing ways to assess the ways in which a given model might be imperfectly aligned and developing and implementing tools and AI assistance that accelerates safety research.
Artificial intelligence will be one of humanity’s most transformative inventions. At DeepMind, we are a pioneering AI lab with exceptional interdisciplinary teams focused on advancing AI development to solve complex global challenges and accelerate high-quality product innovation for billions of users. We use our technologies for widespread public benefit and scientific discovery, ensuring safety and ethics are always our highest priority.
We are pushing the boundaries across multiple domains. Our global teams offer learning opportunities and varied career pathways for those driven to achieve exceptional results through collective effort.
Individual pay is determined by factors including job-related skills, experience, and relevant education or training.
US: $174000 - $253000 (USD) + 15% bonus target + equity + benefits
Learn more about benefits at Google.
Responsibilities
- Research new alignment methods, studying alignment failures, and applying AGI-scalable alignment techniques to frontier models.
- Develop adversarially robust AGI control systems and implement them in production.
- Research interpretability techniques to understand what AI systems are thinking.
- Work with product teams to ensure that our research is correctly adopted.
Qualifications
Minimum qualifications:
- Bachelor's degree in Computer Science, a related Software Engineering field, or equivalent practical experience.
- 3 years of experience in software development, ML engineering, or ML research.
- Experience working with research teams.
Preferred qualifications:
- Experience conducting or contributing to applied research to improve the safety and alignment of frontier AI systems.
- Experience with training large models (e.g., supervised fine-tuning, RLHF).