liveBook · Manning

Automation

As explained in How Large Language Models Work

Automation is the use of technology to perform tasks or processes with reduced direct human labor.

Automation uses technology to perform tasks with less direct human labor. Historically, it has often removed difficult, physically damaging work while allowing labor to move elsewhere. section 9.1.2

Large language models change the concern by targeting knowledge work such as copywriting, visual arts, graphic design, and banking. Their broad usefulness makes every possible use difficult to test for safety. section 9.1.2 section 9.6

Automation does not produce one predictable economic outcome:

  • Jobs may disappear, change, or expand rather than simply vanish. section 9.1.2
  • New work may center on obtaining high-quality training data and applying human creativity. section 9.1.2
  • Broad knowledge-work automation raises unresolved concerns about safety, alignment, data rights, and future feedback loops. section 9.6

The ATM counter-example

ATMs automated many bank-teller tasks, but teller employment still doubled to approximately 600,000 between 1970 and 2010. Automating tasks did not automatically eliminate the occupation. section 9.1.2

How How Large Language Models Work covers it

How deep it goes
Compares historical labor changes with LLM risks, but stops short of predicting automation’s eventual economic outcome.
Examples
Uses farming, mining, assembly work, copywriting, banking, and ATMs to illustrate different automation effects.
What you can do after
Evaluate automation claims without assuming that task replacement directly predicts job loss or social benefit.
How it connects
Connects automation to LLMs, knowledge work, manual labor, training data, alignment, and feedback loops.

Also explained in

Compare how 19 books explain automation →

Related concepts

This explainer was generated by Manning's knowledge graph from the text of How Large Language Models Work. The section links point to the parts of the book it draws on.