Matthew Berman, in a recent video analysis, challenged the conventional wisdom that ever-smarter AI models are the primary drivers of value and adoption. Drawing insights from an exclusive article in The Information titled "How OpenAI's Organizational Problems Hurt ChatGPT," alongside interviews with OpenAI leaders Greg Brockman and Sam Altman, Berman dissected a critical shift in the AI landscape: the diminishing returns of raw model intelligence for the majority of users. He argues that for widespread application, particularly in the consumer sphere, the velocity and seamless integration of AI outputs now supersede incremental gains in intellectual capability.
The core of Berman’s argument stems from observations within OpenAI itself. In prior years, every significant upgrade to ChatGPT’s underlying AI model resulted in a surge in usage, as users quickly discovered new, useful applications. However, "over the past year, some OpenAI staffers noticed a concerning change in the way people who used ChatGPT were reacting to improvements in the chatbot." Even as ChatGPT attracted more users, "improvements to the underlying AI model’s intelligence—and the in-depth research or calculations it could suddenly handle—didn’t seem to matter to most people using the chatbot, several employees said." This trend left employees "scratching their heads," indicating a fundamental disconnect between internal development metrics and external user perception.
For the vast majority of use cases, a "PhD-level intelligence in my pocket" is already more than sufficient. The critical factor for most users is not whether the model can solve complex mathematical problems gold-medal performance at the 2025 International Mathematical Olympiad, but whether it can deliver a useful, accurate response quickly and without friction. Berman emphasizes this point, stating, "Speed matters more than getting the absolute best response. As long as the response is right for 99% of my personal use cases, I want the fastest response possible." This prioritization of speed over maximal accuracy aligns closely with consumer expectations, especially in an era accustomed to instant gratification from services like Google Search.
The implications for developers and companies are profound. Rather than relentlessly pursuing marginal intelligence gains, the focus should pivot to the scaffolding, implementation, and deployment side of AI. This involves optimizing for speed, ease of use, and integration into existing workflows and daily routines. The real value, Berman contends, lies in making AI accessible and effortless, not necessarily in making it infinitesimally smarter.
This shift in priorities has created internal tensions at OpenAI. Fidji Simo, the new CEO of Applications at OpenAI, reportedly believes "products aren't the goal themselves," echoing a sentiment that OpenAI remains primarily a research-focused company. However, Greg Brockman, OpenAI's President, revealed in an interview a difficult reality: "We did not have enough compute to keep that going. And so we made some very painful decisions to take a bunch of compute from research and move it to our deployment to try to be able to meet the demand. And that was really sacrificing the future for the present." This candid admission underscores the immense pressure to allocate finite, expensive compute resources between cutting-edge research and the immediate demands of popular products.
The strategic importance of brand and distribution cannot be overstated. Sam Altman, OpenAI's CEO, recognized that "people really want to use one AI platform." He drew a parallel to smartphone usage, where individuals prefer to use the same device at home and work. This consumer familiarity with ChatGPT translates directly into an advantage in the enterprise space. "The strength of ChatGPT consumer is really helping us win in the enterprise," Altman noted, highlighting that enterprise decision-makers, already familiar with ChatGPT in their personal lives, are more inclined to adopt it professionally. This brand recognition, akin to Google's dominance in search, gives OpenAI a significant moat that transcends raw technical benchmarks.
Ultimately, the "big lie" isn't that AI models aren't getting smarter, but that this continued intelligence growth is the sole or even primary metric for success in the current market. For most users, the intelligence ceiling has been met, and the new frontier is about how seamlessly and rapidly that intelligence can be applied. The battleground for AI leadership is shifting from pure model intelligence to the ubiquity of integration, the responsiveness of the interface, and the power of brand recognition.
