DeepMind, a subsidiary of Google, is currently developing a generalized AI that is expected to surpass human intelligence. According to Nando de Freitas, a leading researcher in the Artificial Intelligence (AI) department, the Gato AI can already perform complex tasks such as writing poetry and assembling blocks. However, DeepMind still needs to expand the AI to compete with human intelligence. This statement was made in response to a report by The Next Web, which claimed that AI could never surpass human intelligence. De Freitas strongly disagreed with this statement and emphasized the need to make the models larger, safer, more efficient, and faster.
Many scientists warn against developing an AI that could surpass human intelligence or at least match it. This could potentially lead to the replacement of humans as the dominant life form on Earth. Therefore, safety is of the utmost importance in the development of superintelligent AI. DeepMind is working on a “red button” to mitigate the risks associated with superintelligent AI.
Despite the potential risks, there are predictions that a generalized superintelligent AI could be a reality by 2028. DeepMind’s development of a generalized AI that can perform complex tasks is a significant step towards this goal. However, it is crucial to ensure that the development of AI is done with caution and safety in mind.
