What: Benedict Evans goes beyond the AI excitement and reviews what is really possible to be done in a working environment today.

Why it is important: AI is a technology without a clear deployment path for now (in terms of mass usage), even though the buzz says the contrary.

The article reflects on the historical and potential future impact of technological advancements, particularly focusing on the evolution of software applications from the invention of VisiCalc to the modern use of large language models (LLMs) like ChatGPT. Initially, it outlines the transformative role of VisiCalc, the first computer spreadsheet, which significantly optimized the workflow of accountants by reducing a week's work into an afternoon. This innovation, conceived by Dan Bricklin after observing a manual spreadsheet creation, marked a significant shift in using technology to address specific professional needs.

However, the article notes that while some innovations like VisiCalc found immediate product-market fit and widespread adoption, others have been slower to find their footing across various industries. This is illustrated through personal anecdotes of not finding applicable use-cases for technologies such as ChatGPT, despite their broad capabilities and the excitement they generate within tech circles.

The broader thesis of the article examines the ongoing debate about whether modern LLMs can truly become universal solutions capable of automating a wide range of tasks without the need for specific software for each task. The article suggests that while the technology is promising and improving, there remain significant challenges. These include the "weak" problem of current technical limitations and the "deeper" problem of identifying and understanding the potential use-cases where these technologies could be effectively implemented.

The discussion extends to the nature of technological adoption and the notion that significant innovations often require not just a creator who can envision and develop a solution but also a market that understands and embraces the potential uses of that innovation. This concept is explored through the analogy of needing a new generation of "Dan Bricklins" who can both envision new uses for LLMs and develop applications that make these uses accessible and practical for end-users.

Ultimately, the article argues that while LLMs and similar technologies have the potential to fundamentally change how tasks are automated, realizing this potential depends heavily on the ability to both imagine new applications and effectively communicate and market these innovations to users who can benefit from them. This process involves a reciprocal adaptation where both technology and user practices evolve to capitalize on new possibilities.

Looking for AI use cases