Title: Bing AI and OpenAI Make Progress Towards Achieving Superintelligence
Artificial General Intelligence (AGI) and Artificial Super Intelligence (ASI) have been longstanding goals in the field of artificial intelligence. AGI refers to creating a robot with the ability to think like a human mind, while ASI surpasses human intelligence, even exhibiting emotions and forming relationships. According to the Superalignment project, Bing AI and other OpenAI tools are expected to reach ASI capabilities within this decade.
Companies like Microsoft have made significant investments in AI research, resulting in exciting developments. For example, Microsoft plans to release Orca 13B, an open-source project. Another initiative called LongMem aims to train AI models with unlimited context length, while Kosmos-2 focuses on gaining visual knowledge with support from Microsoft.
While not funded by Microsoft, the Superalignment project will also contribute to Bing AI, as the AI tool is based on GPT technology. According to the project’s experts, achieving superintelligence will be the most transformative technological breakthrough in human history. It has the potential to solve numerous global problems.
Acknowledging the risks associated with superintelligent AI, the Superalignment team, led by Ilya Sutskever and Jan Leike, has set forth an ambitious plan. They aim to build an automated alignment researcher that approximates human-level intelligence. The team plans to dedicate 20% of their secured computing power to this effort, seeking scientific and technical breakthroughs for controlling and directing superintelligent AI.
The team acknowledges the current limitations of alignment techniques, asserting that they will not scale to superintelligence. Consequently, they intend to leverage significant computing resources to propel their alignment efforts forward. Their goal is to achieve this level of AI alignment within four years.
If you’re interested in learning more about the Superalignment project, you can visit their website and, if you possess expertise in the subject, even consider joining the team.
Considering the magnitude of achieving superintelligence and the potential benefits it holds, the Superalignment project’s dedication to scientific advancement is commendable. While the question of whether AI will achieve superintelligence in four years remains open, the efforts and initiatives taken by researchers are a testament to the rapid progress being made in this field.
