
The current AI boom functions as a corporate "slot machine" driven by leadership's fear of obsolescence rather than genuine utility. Companies like Uber and Walmart have faced massive, unsustainable token costs, blowing through annual budgets in months for trivial tasks like summarizing emails. These generative models are structurally inefficient, requiring a full re-read of context for every generated word, which creates a reinforcement loop of intermittent rewards that keeps engineers hooked. Despite the hype, AI remains largely unprofitable and prone to "confidently wrong" outputs. OpenAI’s push for an IPO reflects an existential need to secure capital as enterprise customers sober up to the reality that AI often creates more problems than it solves. Ultimately, the technology serves as a costly, high-volume generator that provides value only in narrow, verifiable tasks requiring constant human oversight.
Sign in to continue reading, translating and more.
Open full episode in Podwise