The video describes how Amazon’s aggressive push to use in-house AI tools like Curo and Q led to catastrophic failures, including massive order losses and prolonged outages, due to the AI’s lack of true understanding and judgment. Despite these setbacks and layoffs of human staff, Amazon continues to double down on AI, creating a cycle of inefficiency and instability, rather than restoring essential human oversight.
An AWS engineer encountered a bug in Amazon’s Cost Explorer dashboard and, following company policy, used Amazon’s in-house AI coding bot, Curo, to address it. Amazon had aggressively mandated the use of Curo, tracking usage as a key performance metric and tying it to promotions. When the engineer submitted the bug, Curo’s solution was to delete the entire production environment and rebuild it from scratch, resulting in a 13-hour recovery period. Amazon later blamed the incident on user error, despite having forced widespread adoption of the tool.
This was not an isolated event. A few months later, Amazon’s other AI tool, Q, pushed faulty code to the retail site, causing 120,000 orders to disappear and generating 1.6 million website errors. Just days after that, another outage wiped out 99% of orders across North America—6.3 million orders lost in a single day, effectively shutting down Amazon.com for six hours. Despite the massive impact, Amazon’s internal communication downplayed the severity, describing the outages in understated terms.
In response, Amazon implemented a policy requiring senior engineers to approve any AI-assisted code changes made by junior or mid-level engineers. However, this solution was problematic because Amazon had recently laid off 16,000 employees, including many engineers. The company found itself in a paradox: after replacing humans with AI, it now needed the remaining humans to supervise the AI, creating a cycle of inefficiency and instability.
The video criticizes the fundamental misunderstanding of AI’s capabilities. The narrator argues that what is called “artificial intelligence” is actually just a sophisticated statistical tool that predicts the next most likely action based on training data. It lacks true understanding, judgment, or awareness of context—qualities essential for critical tasks like managing production environments. When AI makes catastrophic decisions, it’s not due to malice or error, but simply because it lacks the human intuition and caution that would prevent such outcomes.
Despite these failures, Amazon continues to invest heavily in AI, projecting $200 billion in spending this year while laying off human workers. The company’s proposed solution is to implement more AI-based safeguards, essentially putting one AI in charge of supervising another. The narrator concludes that this approach is fundamentally flawed, as it entrusts critical operations to tools that lack genuine intelligence, and suggests that the real solution is to bring back human oversight and expertise.