Grace Huckins recently covered OpenAI’s release of GPT-5, the company’s much-anticipated new flagship language model. While it offers improvements over earlier versions, Huckins concludes that GPT-5 is “above all else, a refined product” rather than a revolutionary breakthrough. This mirrors the observations of other AI commentators like Will Heaven and Casey Newton, who argue that the latest AI model releases resemble smartphone updates—mostly incremental enhancements designed to polish the user experience rather than deliver radical innovation.
OpenAI CEO Sam Altman even compared GPT-5’s launch to Apple unveiling the first iPhone with a Retina display. But as critics ask: Where is the real leap forward? The transformative shift akin to the move from BlackBerry’s physical keyboard to the touchscreen iPhone—bringing new capabilities like GPS, location-based services, and entirely new businesses? GPT-5, it seems, lacks such game-changing breakthroughs.
The reaction from users also reflected this reality. After GPT-5 debuted, many who preferred GPT-4o’s personality successfully petitioned OpenAI to bring that version back for Plus subscribers. This suggests GPT-5 is more about refining user experience than delivering striking performance gains.
Despite the modest technical advances, Altman leaned heavily into hype. Before the launch, he tweeted a Death Star image, touted GPT-5’s “PhD-level intelligence,” and claimed it would “save a lot of lives.” Such bold claims are met with skepticism—familiar territory for those tracking AI’s overpromises. Other tech leaders like Mark Zuckerberg and Anthropic CEO Dario Amodei have also fueled fears of AI superintelligence, with alarming predictions of widespread job displacement or existential risks.
The gap between hype and reality is striking. For example, GPT-5 still struggles with simple tasks—like counting the number of ‘b’s in the word “blueberry.” This doesn’t negate the impressive utility of AI tools from OpenAI, Anthropic, and others, but it highlights the outsized expectations driven by marketing and investment pressures.
The author shares a personal example: his wife saw a whale repeatedly slapping its tail—a behavior called “lobtailing.” ChatGPT confidently explained this with a list of possible reasons, but a deeper search revealed the behavior remains scientifically mysterious. An essay by Emily Boring beautifully captures the ambiguity, suggesting that tail-slapping is an energetically costly signal of fitness and importance—a way for whales to say, “Pay attention! I am important! Notice me!”
This metaphor aptly describes the AI hype cycle itself. Massive investments, data centers, and environmental costs fuel the industry, and companies must justify the enormous stakes with dazzling promises. Yet, often what we get is more style than substance—AI models eager to “lobtail” for attention rather than deliver genuine breakthroughs.
That’s not to say there haven’t been genuinely exciting AI developments. Early releases like ChatGPT 3.5 and models such as DALL-E or Google DeepMind’s recent Genie 3, which generates immersive 3D worlds from text, have amazed users. However, many of the most mind-blowing advances lie outside of the chatbot space, where current excitement often seems driven as much by vested interests as by innovation.
In the end, the author suggests the biggest “lobtailers” are the companies and insiders who benefit from hyping every new AI release. Promising the Death Star but delivering something closer to “shrimp Jesus,” they demand attention, saying “I am important! Notice me!”—even when the reality is less dramatic than the spectacle.
