Not all employees of major firms working on artificial intelligence think the technology spells doom for humanity. In fact, many who have been part of companies like OpenAI, Meta, and DeepMind express skepticism about the notion that unchecked AI could result in widespread destruction. During casual chats and text exchanges, their responses range from laughter to disbelief. “Lol,” said one engineer. “Haaaaaa,” another chimed in. There’s a certain levity in their voices, suggesting they aren’t losing sleep over AI taking over the world.
Take Jain, for example, who remarked, “I didn’t just wake up last week thinking, ‘Oh no, AI is going to kill everyone.’” It seems the idea of AI leading to mass extinction is taken with a grain of salt by many in the field. Colin Fraser, a data scientist at Meta, even took to social media to assert that there’s no substantial evidence indicating AI models would inevitably pursue goals that could lead to human demise. He humorously pointed out that large language models (LLMs) “won’t wipe out humanity because they just don’t have that dog in them.”
Now, when we talk about experts in the AI arena, the conversations are often nuanced. While there’s a general agreement on the importance of considering a variety of risks associated with AI, it’s not all doom and gloom. Sure, there are ethical concerns about military applications of AI, but many tech professionals emphasize the need for proper guardrails to mitigate potential failures. The discourse isn’t just about fearing a silicon uprising; it’s about understanding the complexities and responsibilities that come with developing such technologies.
But what fuels these concerns about AI threatening humanity? Well, uncontrolled AI could lead to unforeseen consequences, a phenomenon sometimes referred to as ‘silicon species.’ The reality is, numerous AI employees who spoke with the BBC mentioned that they had yet to see any safety researchers embedded in AI labs. It’s a bit alarming, right? Just this past Friday, Anthropic announced plans to bring in external AI evaluators from Faculty, a company owned by Accenture. While that sounds like a step in the right direction, details on when these evaluators will actually join the team remain vague.
And let’s not forget about the OpenAI-Hugging Face incident, which many have dubbed a “wake-up call.” It has stirred discussions across the industry about the need for more robust safety measures. “Go get your high score there, no need to hack us,” has become a bit of a mantra among tech insiders as they ponder the future of AI.
So, what’s next? With the rapidly evolving landscape of artificial intelligence, it seems the conversation will only continue to grow. Are we heading toward a future where AI is a partner rather than a threat? Only time will tell…
Kaynak: Orijinal Haber
