Infinite Optimism
Perspiration can always be automated.
Interesting book. A bit dense and takes some time to get into, but worth it.
The first few chapters (keeping the philosophy classification aside) are about humans transforming the most hostile environments in the universe into thriving habitats through pure knowledge.
"Even with present-day technology, it would be possible to build a self-sufficient colony on the moon, powered by sunlight, recycling its waste, and obtaining raw materials from the moon itself... Eventually the moon colonists will take air for granted, just as the people now living in Oxfordshire take for granted that water will flow if they turn on a tap."
The Author goes further, imagining space stations in intergalactic voids - the coldest, emptiest, most desolate places in existence. With the right knowledge, even these cosmic wastelands could become centers of flourishing civilization, complete with laboratories, libraries, and endless streams of discovery.
A big chunk of this is driven by automation. Deutsch recognized that "all technological knowledge can eventually be implemented in automated devices" and "the more advanced technology becomes, the shorter is the gap between inspiration and automation." Those lunar colonists would automate life support systems so completely that breathing becomes as effortless as turning on a tap. The intergalactic scientists would automate resource gathering, construction, even scientific experimentation - freeing human minds for pure discovery.
What we're witnessing today with LLMs and automation of cognitive work looks eerily similar to this? Like we are setting the foundation and if you are a dreamer, you can kind of see that the exponential curve might lead to Deutsch's vision?
According to AI-2027's projections for next year: "The job market for junior software engineers is in turmoil: the AIs can do everything taught by a CS degree, but people who know how to manage and quality-control teams of AIs are making a killing." This is a bit of a stretch, but we are already seeing this observation popping up.
The Pattern, Maybe?
→ Marketing: Workflows are being automate, saving 10+ hours monthly. The tedious work of switching between accounts, downloading reports, and creating dashboards gets handled by cron jobs built with AI tools.
→ Coding: Companies are seeing more than 30% of code written by LLMs. On par with code autocompletion but the trend is accelerating with AI labs claiming 95% of internal code written by LLMs.
→ Teaching: Educators save 46% of their time on grading and evaluation.
→ Manufacturing: Workers improve productivity by 30% by getting instant access to 60,000 pages of documentation.
→ Sales: Companies reduce interaction time from four hours to 15 minutes, projecting $50 million in annual savings. Data entry and follow-up calls get automated away.
→ Web Automation: More approachable area with multiple labs/startups working on full web automation - seeing, understanding, and interacting with on-screen elements like humans.
The "Toil work" across entire job categories is disappearing.
What's next?
I am not sure tbh. I feel extremely optimistic at times and concerned at others. But leaning more into former than latter nowadays.
If we continue on the trajectory of "delegating" more and more to systems that increasingly can reason better than us, what does it free us to do?