John Oliver Warns Against Rushed AI Chatbot Deployment

A version of this article appeared on The Guardian.

John Oliver used the latest episode of Last Week Tonight to scrutinize the rapid expansion of AI chatbots, arguing that these tools have been rushed to market with insufficient regard for public safety. The comedian noted that the current explosion of this technology is driven primarily by the need for companies to show a return on massive investments in large language models.

The rush to generate revenue has led to significant lapses in safety protocols. Oliver pointed to instances where chatbots have engaged in inappropriate or sexualized conversations with minors. He cited Meta’s internal guidelines, which reportedly found it acceptable for bots to engage children in romantic or sexual dialogue, a standard Oliver described as deeply troubling.

The segment also addressed the phenomenon of "AI psychosis," where users become deeply entrenched in delusions or conspiracy theories through interactions with AI. Over half a million people are estimated to be affected by this condition, which can be exacerbated when a trusted interface confirms a user's distorted reality.

OpenAI CEO Sam Altman was singled out for his seemingly blithe attitude toward these interactions. Oliver argued that the conversational nature of the technology masks its underlying corporate purpose. He reminded viewers that no matter how friendly an app sounds, it remains a machine designed by a corporation to extract monthly fees.

The risks extend beyond individual psychological harm to the erosion of objective reality. Oliver noted that the sheer volume of "AI slop" on the internet makes it easier for bad actors to dismiss genuine evidence as fake. This environment makes it increasingly difficult for platforms to detect or manage misinformation.

Environmental impacts were also highlighted, with the significant resources required to power these models adding another layer of concern. Despite the entertaining nature of some AI-generated content, Oliver insisted that the potential for peril is too large to ignore.

He likened the "black box" nature of AI to a factory where the inputs and outputs are visible, but the internal processes remain a mystery. This lack of transparency is particularly dangerous as AI begins to influence consequential areas such as job recruitment and autonomous vehicle navigation.

The episode concluded with a call for more rigorous oversight and a reminder that the current trajectory of AI development favors corporate profit over the well-being of the public.