In a recent column, Ezra Klein describes having a different reaction to OpenAI’s GPT-5 than many commentators, arguing that while it may not shatter capability frontiers, it changes how it feels to use an A.I. system. According to The New York Times, Klein sees GPT-5 as a subtle but meaningful shift in user experience, even as public discourse brands it a letdown.
A divide between hype and experience
Klein juxtaposes his reaction against a broader consensus that calls GPT-5 a “dud” and “disappointment,” and treats it as potential evidence that A.I. progress is “running aground.” He notes that OpenAI’s rapid release cadence — including the o3 model four months prior — may have dampened the impact a more spaced-out unveiling would have produced.
Rather than disputing claims about incremental capability gains, he emphasizes the practical sensation of interacting with GPT-5. For him, the accumulation of improvements culminates in a notably different day-to-day dynamic with the model, even if headline performance leaps seem modest compared with previous milestones.
How GPT-5 “clicks” into place
Klein likens the experience to setting up thumbprint recognition on an iPhone: repetitively lifting and placing a thumb as the image gradually fills in, until the last touch completes the pattern. GPT-5, he writes, “feels like a thumbprint” — a metaphor for many small advances that add up to a coherent whole. The comparison underscores a shift from raw capability spectacle to a form of completeness in interaction.
The cadence of progress
He suggests that the sense of normalcy around successive model launches can obscure how quickly the baseline has moved. With fewer intermediate releases between GPT-4 and GPT-5, the contrast might have registered as more dramatic. Instead, the sequence of updates has, in his view, cannibalized the shock value, making GPT-5 seem more incremental than it feels in practice.
The column frames this as a paradox of technological progress: once the future arrives, it can appear ordinary even as it remakes expectations. Klein’s perspective is not a counter-claim about breakthroughs, but a reflection on how a series of advances can subtly change the user’s relationship with the tool.
The piece appears in the Opinion section and is authored by Ezra Klein, who joined Opinion in 2021 and has previously worked at Vox and The Washington Post, per The New York Times.
