OpenAI Forms Task Force To Prevent An Artificial Intelligence Catastrophe

OpenAI has assembled a team with the aim of averting an artificial intelligence (AI) doomsday, as the organization expresses apprehension regarding the potential annihilation of humanity caused by renegade AI technology.

As per OpenAI, AI possesses the capacity to evolve into superintelligence, which may represent the most momentous technological leap ever accomplished by humankind, facilitating solutions to pivotal global dilemmas.

However, Ilya Sutskever and Jan Leike of OpenAI caution that humans are ill-equipped to grapple with technology that surpasses their own cognitive abilities.

The peril of a superintelligent AI

The OpenAI team acknowledges the impending peril of a superintelligent AI and concedes that presently they lack a solution to control or redirect its conduct. Existing methodologies for aligning AI, such as reinforcement learning from human feedback, rely on human oversight. Nevertheless, as AI systems surpass humans in intelligence, this supervision becomes unreliable.

Mr. Sutskever, the co-founder and chief scientist of OpenAI, alongside Mr. Leike, the head of OpenAI’s alignment department, are establishing a group of researchers and engineers tasked with surmounting the technical obstacles posed by superintelligence. They have set a four-year deadline for the team to achieve this objective.

While the prospect of humanity’s demise is alarming, the leaders at OpenAI maintain an optimistic outlook on discovering a solution. They firmly believe that with focused and concerted endeavors, the predicament can be resolved. Encouraging concepts and preliminary experiments, coupled with measurable progress and the ability to empirically study these challenges using current AI models, contribute to their sanguinity.

Mr. Sutskever and Mr. Leike emphasize their unwavering dedication to disseminating the outcomes of their work widely. OpenAI is actively recruiting research engineers, scientists, and managers who possess an interest in preventing the dominion or eradication of humanity by AI. Furthermore, OpenAI has announced the widespread availability of GPT-4, their most capable AI model, with the intention of making it more accessible to developers.

Paulo Montenegro
Categories: Robots
Tags: AI (Artificial Intelligence), OpenAI