> ## Content Index > Fetch the complete content index at: https://www.philsimon.com/llms.txt > Use this file to discover other available public pages before exploring further. # The Steinmetz Lesson - URL: https://www.philsimon.com/the-steinmetz-lesson/ - Published: 2026-11-09T13:04:00.000Z - Updated: 2026-11-09T13:04:00.000Z - Description: Thoughts on AI, tokens, and the cost of true expertise. - Author: Phil Simon - Tags: AI, Claude, ChatGPT, Faves, Dataviz, Infographic Deep domain expertise has long fascinated me. My favorite tale on the subject involves [Charles Proteus Steinmetz](https://www.smithsonianmag.com/history/charles-proteus-steinmetz-the-wizard-of-schenectady-51912022/?ref=philsimon.com). Roughly a century ago, the Prussian-American engineer billed Henry Ford $10,000 for quickly solving a thorny production problem. Steinmetz did, and submitted the invoice to Ford. Ford balked, and asked his hired help to itemize it. The Wizard of Schenectady gladly did as follows: Making chalk mark on generator $1. Knowing where to make mark $9,999. Ford remitted payment. ## A Century Later It took decades for [Steinmetz](https://en.wikipedia.org/wiki/Charles%5FProteus%5FSteinmetz?ref=philsimon.com) to become proficient in mathematics and electrical engineering. A poseur he was not. At the time of his death in 1923, he held more than [200 patents](http://www.becklaser.de/hbeng/steinmetz.html?ref=philsimon.com). Imagine what the polymath could have done with today's crop of genAI tools. ## Sign up for Simon Says Weekly insights on AI, workplace tech, and more. SUBSCRIBE Email sent! Check your inbox to complete your signup. No spam. No bullshit. Unsubscribe anytime. Today anyone can easily access the greatest cheat codes in human history—and not just college students too lazy to write their own papers. Claude, ChatGPT, and their ilk are remarkable [Swiss Army Knives](https://www.philsimon.com/claude-the-ultimate-utility-player/). > Anyone can appear to be an expert on anything. Anyone *can appear to be* an expert on anything. As one of many examples, a mediocre programmer can pose as an advanced one. Ditto for writers, designers, lawyers, and architects. The cost isn't time to learn one's craft à la Steinmetz, but tokens: Simon’s First Law of Generative AI A line chart. The horizontal axis shows expertise. The vertical axis shows tokens. The line starts high on the left and falls steeply, then flattens to the right, so the tokens needed drop as expertise grows. Simon’s First Law of Generative AI tokens expertise ↻ Replay [Simon's Law of Pre-Sales EffortA simple chart on how I evaluate legitimate service inquiries.![](https://storage.ghost.io/c/8f/b5/8fb5e321-49a6-4cf5-8a28-c6f6d5417e54/content/images/icon/300br-1-e3a08062-5c57-478a-bd80-dff0719c6a45.png)Phil SimonPhil Simon![](https://storage.ghost.io/c/8f/b5/8fb5e321-49a6-4cf5-8a28-c6f6d5417e54/content/images/thumbnail/saleswoman-presentation-341be2e2-7ea6-468e-aec9-49d436d5bf13.jpg)](https://www.philsimon.com/simons-law-of-pre-sales-effort/) ## Simon Says It's only a matter of time before Anthropic, OpenAI, and other companies raise prices, restrict features, or some combination of the two. (Yep, [*enshittification*](https://americandialect.org/2023-word-of-the-year-is-enshittification/?ref=philsimon.com) at its finest.) Perhaps in an attempt to prepare themselves for the inevitable, senior leaders are [increasingly pruning their organizations' AI budgets](https://www.theguardian.com/technology/2026/oct/13/ai-companies-rushed-workers-fallout?ref=philsimon.com). Have at it. Just remember the lesson that Henry Ford learned: true subject-matter experts aren't cheap.