Jacob Coxon, a former researcher at Anthropic, has raised serious concerns about the future of humanity in light of artificial intelligence advancements. Speaking to the BBC after his resignation, Coxon mentioned that many in the AI field are “genuinely frightened” about the potential for human extinction. His words come at a time when safety concerns regarding AI technology are growing rapidly, sparking a viral response to his post about the dangers of AI. The 27-year-old urged for a slowdown in AI development, advocating for industry-wide regulations and independent monitoring of AI model advancements.
Coxon emphasized that while developing AI technology itself is not in question, the associated risks are “serious” and pose a significant threat on an international scale. He painted a vivid picture of a potential future where a swarm of bots could function like a supercomputer, capable of taking control of the internet. “I think this scenario could become a reality in just six months to a year,” he warned. His peers in the industry share these fears, with some even contemplating buying land in safer regions as a precaution against potential AI disasters.
In response to Coxon’s alarming claims, a spokesperson from Anthropic stated that the company has always been transparent about the dual nature of AI, which could bring both enormous benefits and unprecedented risks. They highlighted that the firm has been a leader in understanding AI models and was the first to publish a framework aimed at mitigating risks tied to AI development. The spokesperson further noted that they are actively addressing issues related to AI misalignment.
Coxon’s concerns echo sentiments from other notable figures in the field, including Evan Hubinger, a fellow scientist at Anthropic, and Geoffrey Hinton, a computer scientist often dubbed the “Godfather of AI.” Hinton mentioned that while it’s difficult to assign a probability to the risk of AI leading to human extinction, the apprehensions surrounding it are not unreasonable. He compared asking Coxon about these risks to consulting an air conditioning technician about climate change – underscoring the need for proper context when discussing such high-stakes topics.
Last week, during a conference hosted by Goldman Sachs, Anthropic’s CEO Dario Amodei dismissed fears of AI leading to the end of humanity as unfounded. However, his comments did not sit well with many, as they felt the seriousness of the concerns was being overlooked. Huang, another prominent figure in the field, described the notion that AI could end humanity as “complete nonsense.” Meanwhile, OpenAI CEO Sam Altman assured the public that such an occurrence would not transpire this year, citing ongoing safety concerns within the industry.
The UK government has also been hesitant to endorse the idea of a “kill switch” for dangerous AI systems, adding another layer of complexity to the ongoing debate about AI safety. As experts and researchers continue to grapple with the implications of AI technology, one thing is certain: the conversation around its potential risks will only intensify. Will we be able to harness AI’s benefits while keeping its dangers at bay? Only time will tell…
Kaynak: Orijinal Haber