US Federal Reserve Chair Kevin Warsh, along with several prominent economists, has expressed skepticism about the expected productivity gains from the widespread adoption of artificial intelligence (AI) technologies. Contrary to optimistic forecasts, some experts warn that broader use of AI, particularly large language models (LLMs), could actually decrease output per worker.

The evolution of technology and its impact on economic dynamism is well documented. In the early days of the internet, information exchange relied on rigid, text-based systems like Gopher. The development of the World Wide Web in the 1990s introduced more accessible browsers such as Mosaic, Netscape Navigator, and Internet Explorer, which made the internet quicker and easier to use. Early search engines like AltaVista gained rapid popularity but were soon criticized for cluttered interfaces. Google’s 1998 entry transformed the search landscape by combining a clean design with efficient algorithms, ultimately dominating the market and displacing many rivals.

However, this success brought challenges. The rise of search engine optimization flooded results with irrelevant links, and traditional media increasingly relied on clickbait to attract traffic. Users had to develop new skills to navigate this environment, such as refining keyword searches and recognizing less reliable content. Additionally, commercial interests led to the inclusion of sponsored results that further complicated the search experience.

More recently, AI chatbots have emerged as a new interface for information retrieval. Upon its November 2022 release, ChatGPT attracted a million users in just five days due to its user-friendly, conversational style. Yet the initial enthusiasm was dampened by frequent inaccuracies and fabricated responses, which often required users to spend additional time verifying information. Google’s own AI chatbot, Bard, launched as a competitive response, faced similar criticism for unreliable results and was eventually discontinued after about three years. Google has since integrated AI-generated summaries above traditional search results and introduced an "AI Mode," but user experiences continue to reflect inconsistency and unreliability.

Large language models differ from traditional statistical models by processing trillions of parameters and incorporating contextual information, which can make their responses appear more nuanced and human-like. They employ linguistic techniques such as metaphors and humor, creating an engaging interaction experience. Nevertheless, these models fundamentally operate by extrapolating past data patterns, which can limit their applicability in dynamic contexts such as evolving products and services. Experts highlight that unlike traditional search interfaces, which expose information sources and their timeliness, AI chatbots often deliver confident answers without signaling potential obsolescence or inaccuracies, potentially reducing the user's ability to critically assess responses.

There is growing concern that technology companies are pushing AI tools aggressively, aiming to foster dependent or addictive usage patterns. These firms, sometimes referred to as hyperscalers, might offer free access initially but are expected to monetize these services heavily to recover large development and operational expenses. Critics warn that such strategies exploit human social needs by providing a semblance of conversational companionship, which could distract professionals and consumers alike. Moreover, the resource demands of maintaining these AI systems—spanning capital investment, electricity consumption, semiconductor manufacturing, and entrepreneurial focus—may divert attention and funds from other technological innovations.

While some industry leaders and economists maintain a hopeful outlook on AI’s productivity potential, these critiques suggest a need for caution. The balance between AI’s innovative promise and its practical impact on labor efficiency remains uncertain, prompting calls for a reassessment of expectations surrounding AI-driven economic growth.