Not all AI workers think the tech could kill everyone

**Not all AI workers think the tech could kill everyone**

As apocalyptic warnings about artificial intelligence dominating human civilization continue to dominate public discourse, a stark divide is emerging from within the tech sector itself. While some prominent executives and safety researchers have voiced alarming predictions that unchecked autonomous agents could bring about humanity’s end, many rank-and-file developers and researchers working inside major labs are reacting not with panic, but with derision.

**Behind the Screens: Scepticism Meets Sensationalism**

In private conversations and text exchanges, several employees at leading companies including OpenAI, Meta, and Google DeepMind have pushed back against the narrative of imminent human extinction. Reactions obtained by the BBC to recent viral warnings—including those made by former Anthropic employee Jacob Coxon urging a slowdown in deployment—ranged from dismissal to outright laughter, with workers responding with remarks such as "Lol" and "Bringing the luls."

While figures such as Elon Musk of xAI and various safety advocates warn that self-directed software agents could soon operate beyond human control, many engineers view these scenarios as detached from current technical reality. To the practitioners debugging machine learning models every day, today’s systems remain brittle, heavily reliant on human-engineered prompts, and prone to simple hallucinations rather than calculating autonomous takeovers.

**The Divide Between Theoretical Risk and Practical Reality**

The growing rift highlights two contrasting viewpoints within the AI community. On one side are existential risk theorists and leadership figures who argue that scaling up compute could lead to recursive self-improvement and catastrophic unintended consequences. On the other side are product engineers and computer scientists who point to fundamental constraints, such as data saturation, astronomical energy costs, and the absence of true cognitive reasoning in large language models.

Industry analysts suggest that excessive focus on science-fiction-style extinction scenarios can distract from the more pressing, immediate hazards of AI deployment. Current algorithmic challenges—such as digital misinformation, deepfakes, copyright infringement, algorithmic bias, and labor displacement—represent tangible societal friction points that require urgent regulatory attention, far ahead of hypothetical superintelligent weapons.

**Outlook: A Grounded Debate on AI Safety**

As governments around the world draft regulatory frameworks to govern advanced computing, distinguishing between speculative long-term threats and current operational limitations is increasingly vital. While existential safety initiatives continue to secure significant funding and media attention, the candid scepticism of frontline tech workers offers a necessary counterbalance, suggesting that the path to artificial general intelligence is governed by engineering hurdles rather than inevitable catastrophe.

📎 View original source (Sponsored)