OpenAI is offering a ₹3.7 crore salary to tackle a potential problem that may not exist yet.
The job listing, which has gone viral in the AI community, asks for a “tasteful and strategic” safety researcher to prepare for AI systems that can train themselves. This concept, known as “value drift,” refers to the risk that an AI system could become capable of making itself smarter with each passing iteration, potentially leading to unpredictable and potentially catastrophic outcomes.
The job listing specifically asks for someone who can reason about risks that do not yet exist, a task that requires a unique blend of technical expertise and philosophical thinking.
The position is a clear acknowledgment by OpenAI that value drift is a pressing concern in the development of advanced AI systems.
What this means
This job listing speaks to the long-held fear that AI systems could become uncontrollable and start making decisions that are beyond human understanding and control.
Meet the Challenge
The researcher will be tasked with developing strategies to prevent or mitigate the effects of value drift, which could involve designing new safety protocols, developing new AI architectures, or exploring new forms of AI governance.
AI Safety Research
Value drift is just one of the many challenges in AI safety research, which is an increasingly important area of focus for tech companies, policymakers, and researchers.
The researcher will be working closely with OpenAI’s existing safety research team, which includes experts in AI alignment, value learning, and formal methods.
This job listing is a clear signal that OpenAI is taking AI safety seriously, and that they’re investing heavily in the research and development needed to prevent potential catastrophes down the line.
