Cheaper/faster/easier makes for step changes (and that's why even current-level LLMs are transformative)
A viral essay claims even weaker AI models create a qualitative leap by making thought processes 10x cheaper and easier.
A viral essay on LessWrong by user 'Ruby' presents a compelling framework for understanding AI's impact, arguing that the true transformation comes not from flashy, high-level task automation, but from making the basic building blocks of thought—memory, search, and summarization—radically cheaper and easier. The post draws a historical parallel: technologies like writing, the printing press, and modern transport didn't enable humans to do fundamentally *new* things, but made existing activities (communicating, remembering, traveling) orders of magnitude more efficient, leading to qualitative civilizational shifts.
Ruby applies this 'cheaper/faster/easier' lens to current Large Language Models (LLMs) like GPT-4 and Claude 3. While attention focuses on macro-tasks like coding or diagnosis, the essay highlights their power in automating constitutive mental tasks: writing notes (storing info), locating text (searching/recalling), and summarizing (processing info). The author contrasts the high-friction, effortful process of traditional note-taking and memory recall with the near-zero cost of using an LLM-powered 'Exobrain' to instantly record, search, and synthesize thoughts.
The core argument is that a sufficient quantitative reduction in the cost of these cognitive operations creates a qualitative change in human capability. By turning a task that required stopping, retrieving a device, and typing into a seamless, voice-activated command, LLMs remove the friction that previously prevented effective external memory systems. This enables the practical construction of personal AI assistants that fundamentally augment human cognition, not by being superhuman, but by being relentlessly available and efficient at the basics. The transformative potential lies in the compounding effect of automating these low-level processes across millions of users, reshaping how knowledge work is done.
- The essay frames LLM impact through historical 'step changes' where quantitative efficiency gains (like the printing press) lead to qualitative societal shifts.
- It argues current models (GPT-4, Claude 3) are transformative by making basic cognitive tasks—note-taking, search, summarization—10-100x cheaper and less effortful.
- The author's 'Exobrain' concept demonstrates this: using LLMs to create a seamless external memory system, overcoming the friction of traditional methods.
Why It Matters
It shifts the focus from waiting for AGI to leveraging today's AI to radically augment human cognition and productivity by automating mental grunt work.