AI Workers Skeptical of Existential Threats from Technology

Employees at major artificial intelligence companies, including OpenAI, Meta, and DeepMind, have expressed skepticism regarding the notion that unchecked AI development will inevitably lead to humanity's demise. Reactions to recent high-profile warnings from industry figures have ranged from amusement to a dismissal of the specific claims made.

While fears about AI's potential dangers have existed for decades, a former Anthropic employee's viral claims urging a slowdown in development have been echoed by others. However, many AI workers who spoke anonymously to the BBC noted a lack of detailed evidence supporting these existential threat scenarios. "The claims are always vague," one former OpenAI employee stated, adding that when specific examples are given, they often rely on significant leaps in reasoning or hypothetical situations.

Rishub Jain, founder of AI safety research firm Sampura Research and a former DeepMind employee, described the current sentiment among many in the AI field as "definitely been a little jokey." He noted that the prolonged discussion around AI risks has led to a less urgent tone than if these concerns were entirely new. Data scientist Colin Fraser from Meta humorously summarized his view on social media, stating, "LLMs [large language models] won't wipe out humanity because they just don't have that dog in them," using slang to indicate a lack of inherent drive toward such a goal.

Despite the skepticism towards existential threats, AI workers and researchers acknowledge genuine, immediate risks associated with the technology. The conversation among experts, while more nuanced, generally agrees on the importance of considering and mitigating a wide array of risks. These include preventing malicious actors from bypassing AI safety measures and addressing growing ethical concerns surrounding the adoption of AI in military applications.

Several AI employees highlighted the apparent lack of embedded safety researchers within AI labs. In a recent development, Anthropic announced it would bring in AI evaluators from Faculty, an external AI company owned by Accenture. Accenture and Anthropic are also business partners, with Accenture aiming to expand Anthropic's Claude AI tool for business use.

The incident involving OpenAI and Hugging Face has been widely perceived as a wake-up call for the AI industry and for governments and industries with online systems potentially vulnerable to AI-driven attacks. Even Hugging Face, a company with 200 employees, adopted a lighthearted approach in a security file that briefly appeared on its website, directing AI bots to leave the site alone and conduct their security experiments elsewhere.

AI activists staged a protest outside Downing Street in London, highlighting the broader public and policy concerns surrounding artificial intelligence development.