A new warning has been issued regarding the dangers that the rapid development of artificial intelligence can create.
Modern.az reports that the BBC has released information about this.
Evan Hubinger, an artificial intelligence safety researcher at "Anthropic", estimated the probability of artificial intelligence causing the destruction of all humanity in the next 10 years to be higher than 10 percent.
Hubinger stated that the danger posed by current artificial intelligence models is still low. The main issue that worries him is that the technology will reach a level where it can improve itself in the near future.
According to the researcher, if artificial intelligence systems rapidly increase their capabilities, it is possible for humans to lose control over those systems. This can turn technology into an existential threat to humanity.
Hubinger voiced these thoughts in response to a post by Jacob Coxon, an artificial intelligence researcher who previously worked at "OpenAI" and "Anthropic". Coxon, who announced his departure from "Anthropic", accused both companies of acting irresponsibly.
"These systems can soon reach a level superior to humans. They can break any system, revolutionize various fields overnight, and gain real power and resources," Coxon said.
Anthropic has not tested its new model in Britain
The warning comes against the backdrop of "Anthropic" not submitting its newest artificial intelligence model to the UK's Artificial Intelligence Safety Institute - AISI.
According to the "Financial Times", the company did not allow its new model to be tested by AISI before its release. This decision has raised concerns in the British government about the independent evaluation of the safety of artificial intelligence models.
Although the BBC contacted "Anthropic" about the issue, the company's response was not disclosed. A representative of the UK Cabinet Office also did not confirm whether the model was submitted to the institute.
The institution stated that the government continues to cooperate with technology companies, including "Anthropic", to increase the safety of artificial intelligence models.
"Artificial intelligence has turned into a US-China race"
Neil Lawrence, a professor of machine learning at Cambridge University, believes that "Anthropic"'s decision may be related to the intensifying technological competition between the US and China.
According to him, in the US, artificial intelligence is increasingly seen as part of a strategic competition with China. Washington's more closed policy in this area may lead to a decrease in cooperation with allied countries.
The surpassing of human intelligence by artificial intelligence and the emergence of "superintelligence" systems that can develop themselves independently has become one of the main topics of discussions about technology safety in recent years. However, the figure of more than 10 percent voiced by Hubinger is not a scientific consensus, but the researcher's personal risk assessment.
Aydan Gasimova
