An international team of scientists from Germany and the UK investigated whether artificial intelligence (AI) models could break out of human control and pose a threat to humanity. The study was published on: website Association for Computational Linguistics (ACL).
The experts focused on skills called emergent abilities, which are skills that the AI develops independently rather than being deliberately coded into a program by the neural network’s creators.
The team experimented with four different AI models, giving them tasks that had previously been identified as urgent. The scientists wanted to find out whether the AIs were creating new ways to accomplish the tasks or simply working in complex ways within the confines of their own code.
For all four models, the ability to follow instructions, memorization, and linguistic proficiency accounted for all reported abilities. The scientists found no evidence of independent behavior in the neural networks involved.
“There were concerns that as models got bigger, they would solve new problems that we can’t even imagine today. This raises the danger that these larger models could acquire dangerous capabilities, including reasoning and planning.
Our research shows that fears that the model will spin out of control and do something completely unexpected, innovative and potentially dangerous are unfounded,” said co-author Tayyar Madabushi.
Scientists added that today’s AI models do not have such intelligence. However, this does not mean that they are completely harmless and do not pose a threat to humans.
Previous researchers taught Robots increase work safety by almost half, accounting for human negligence.
What are you thinking?
Source: Gazeta
Barbara Dickson is a seasoned writer for “Social Bites”. She keeps readers informed on the latest news and trends, providing in-depth coverage and analysis on a variety of topics.