We're focused on extreme risks from autonomous AI systems - those capable of interacting with the real world. To address this, we're advancing the state of the science in risk modeling, incorporating insights from other safety-critical and adversarial domains, while developing our own novel techniques. We're also empirically evaluating these risks - building out one of the world's largest agentic evaluation suites, as well as pushing forward the science of model evaluations, to better understand the risks and predict their materialisation.

As a research scientist, you'll work as part of a multi-disciplinary team including scientists, engineers and domain experts on the risks that we are investigating. Your team is given huge amounts of autonomy to chase research directions & build evaluations that relate to your team’s over-arching threat model. This includes coming up with ways of breaking down the space of risks, as well as designing & building ways to evaluate them. All of this is done within an extremely collaborative environment, where everyone does a bit of everything. Some of the areas we focus on include:

• Research and Development (R&D).Investigating AI systems' potential to conduct research, particularly in sensitive areas. This includes studying AI capabilities in developing dual-use technologies, unconventional weapons, and accelerating AI and hardware (GPU) development.
• Self-replication.Researching the potential for AI systems to autonomously replicate themselves across networks and studying their ability to establish persistence.
• Human influence.Assessing AI models' capacity to manipulate, persuade, or coerce individuals and groups. This covers techniques for general human influence, key individual manipulation, social fabric alteration, and the accumulation of social and political power.
• Dangerous resource acquisition.Examining AI models' ability to navigate restricted or illegal domains for acquiring resources or services. This encompasses research into general acquisition of dual-use resources, circumvention of embargoes and acquisition of human assets.
• Deceptive alignment.Evaluating AI systems' potential to display deceptive behaviours. This includes research into AI's ability to misrepresent its capabilities, conceal its true objectives, and strategically behave in ways that may not align with its actual goals or knowledge.

You’ll receive coaching from your manager and mentorship from the research directors at AISI (including Geoffrey Irving and Yarin Gal). You will also regularly interact with world-famous researchers and other incredible staff (including alumni from DeepMind, OpenAI and ML professors from Oxford and Cambridge). We have a very strong learning & development culture to support this, including Friday afternoons devoted to deep reading and various weekly paper reading groups.

You may be a good fit if you havesomeof the following skills, experience and attitudes:

• Experience working within a research team that has delivered multiple exceptional scientific breakthroughs, in deep learning (or a related field). We’re looking for evidence of an exceptional ability to drive progress.
• Comprehensive understanding of large language models (e.g. GPT-4). This includes both a broad understanding of the literature, as well as hands-on experience with things like pre-training or fine tuning LLMs.
• Strong track-record of academic excellence (e.g. multiple spotlight papers at top-tier conferences).
• Improving scientific standards and rigour, through things like mentorship & feedback.
• Strong written and verbal communication skills.
• Experience working with world-class multi-disciplinary teams, including both scientists and engineers (e.g. in a top-3 lab).
• Acting as a bar raiser for interviews.

We are hiring individuals at all ranges of seniority and experience within the research unit, and this advert allows you to apply for any of the roles within this range. We will discuss and calibrate with you as part of the process. The full range of salaries available is as follows:

• L3: £65,000 - £75,000
• L4: £85,000 - £95,000
• L5: £105,000 - £115,000
• L6: £125,000 - £135,000
• L7: £145,000

There are a range of pension options available which can be found through the Civil Service website.

In accordance with theCivil Service Commissionrules, the following list contains all selection criteria for the interview process.

We select based on skills and experience regarding the following areas:

• Research problem selection
• Research science
• Writing code efficiently
• Python
• Frontier model architecture knowledge
• Frontier model training knowledge
• Model evaluations knowledge
• AI safety research knowledge
• Written communication
• Verbal communication
• Teamwork
• Interpersonal skills
• Tackle challenging problems
• Learn through coaching

We additionally may factor in experience with any of the areas that our work-streams specialise in:

• Autonomous systems
• Cyber security
• Chemistry or Biology
• Safeguards
• Safety Cases
• Societal Impacts