New Delhi: Amidst the rapidly increasing capabilities of Artificial Intelligence i.e. AI, the debate regarding its dangers has intensified. Meanwhile, Google DeepMind's AI safety researcher Bilal Chughtai has expressed concern by resigning. He said that AI may have the ability to destroy humans and we are running out of time to avoid it.
Concern expressed about the speed of AI
Bilal Chughtai gave the reason for his resignation in a social media post. He said that while working at Google, he has closely seen the development of AI. According to him, he is very worried about the current direction of this technology.
He told that when he started work on AI in the beginning of 2022, these systems were quite limited, but in a few years the capability of AI has increased rapidly. AI agents are now being able to work on complex problems and in some cases take action autonomously.
“Things are about to get even crazier”
Chughtai believes that in the coming time, such superintelligent AI systems can be developed, which will be more capable than humans in almost every field. The biggest danger can arise when such systems go beyond human control and start taking dangerous decisions.
He warned that such a situation could reach the point of permanently weakening human power and control or threatening the existence of humanity.
Another DeepMind researcher resigned
Josh Engels, associated with Google DeepMind's AGI safety team, has also resigned. He feared that advanced AI would cause huge losses in the next few years. Engels is now associated with METR, a non-profit organization that assesses the capabilities and risks of AI models.
Appealed for cooperation from AI companies
Chughtai also expressed concern about the intense competition among AI companies. He said that companies should work together to keep the pace of development such that the society can handle it. Also, more work is necessary on AI alignment, so that AI systems work in accordance with human interests. According to him, at present, it is not possible to understand and control security at the speed with which AI capabilities are increasing.