I started reading science fiction when my parents subscribed me to the Tom Swift series of books, and I'm still reading the genre today.
We still don't have faster-than-light travel, and flying cars are (still) just around the corner. But we can now talk to our machines and have something
reasonable come back.
It's still amazing to me that essentially the same algorithm I set as a Code Kata can carry out a perfectly good (and helpful) conversation.
Because they are so good, it's easy to forget that there's no original thought taking place inside an LLM; the output you get is basically a statistical sampling across their training input.
That's perfectly fine when you want to learn something that was fairly common at the time the LLM was trained (or fed RAG data). But ask it to teach you something new, and the quality drops, often to the point where the answers are basically spurious.
LLMs can only teach you the past. That's why I strongly feel that books are vitally important to our future.
A book represents a major commitment by many people to curate and present knowledge. Unlike a quick post on the web, a book is long-lived, and the author takes care with every detail. Books go through intense reviews: the author works with a development editor who refines structure and flow, technical editors who look for errors, and copy editors who polish the writing (and sometimes even the code). Books are laid out to make them easier to read. And even though you can get them electronically, they can also be read on paper. Call me old-fashioned, but a paper book invites you to immerse yourself in its content in a way that no screen can match.
That’s why the Bookshelf does not use AI-written content in our books. Sure, we use AI to help make the occasional cover image, and some authors use AI to check their writing style, but the words you read, and the thought behind them, are entirely human-generated. That isn’t a cheap process, but I believe the value it produces makes it worth the cost.
And, because irony is priceless, here’s a 40% discount on five of our machine learning and AI titles—all written by lovely people.
