OpenAI unveils new taskforce to combat 'superintelligent' AI systems

American private artificial intelligence (AI) research corporation OpenAI has recently kickstarted its new taskforce Superalignment to curate solutions to steer and control “superintelligent” AI systems.

OpenAI took to breaking the news via a blog post published on its official website, where it wrote that the Superalignment team will be spearheaded by Ilya Sutskever and Jan Leike, cofounder and chief scientist and head of alignment at OpenAI respectively. The team will constitute of researchers and engineers from OpenAI’s previous alignment teams, as well as researchers from other divisions across the company.

Superalignment will be utilising approximately 20% of the secured compute over the coming four years to aid them in resolving the issue of superintelligence alignment.

While superintelligence could provide solutions to numerous prominent global issues, it can potentially act as a double-edged sword and cause more harm to humanity. This called for the formation of the Superalignment team as superintelligence is believed to emerge within this decade.

“Currently, we don’t have a solution for steering or controlling a potentially superintelligent AI and preventing it from going rogue. Our current techniques for aligning AI, such as reinforcement and learning from human feedback, rely on humans’ ability to supervise AI,” Sutskever and Leike wrote.

“But humans won’t be able to reliably supervise AI systems much smarter than us, and so our current alignment techniques will not scale to superintelligence. We need new scientific and technical breakthroughs,” added Sutskever and Leike.

With the formation of Superalignment, OpenAI aims to construct a roughly human level alignment researcher. From there, it can utilise copious amounts of compute to expand their efforts and in turn align superintelligence.

Curating the first automated alignment research will see OpenAI’s efforts in constructing an AI system that can aid in the evaluation of other AI systems, as well as automating the search for problematic practices.

“While this new team will focus on the machine learning challenges of aligning superintelligent AI systems with human intent, there are related sociotechnical problems on which we are actively engaging with interdisciplinary experts to make sure our technical solutions consider broader human and societal concerns,” Sutskever and Leike concluded.

Related articles: 
AI chatbot powered by OpenAI’s GPT jumps on board Snap
LLaMA for dummies: 101 on Meta’s new AI research tool touted to outperform OpenAI
73% of consumers trust content by generative AI: Here's why they shouldn't