alt.hn

8/22/2026 at 9:07:26 AM

Embedded AI

https://nostarch.com/embedded-ai

by 0x54MUR41

8/22/2026 at 2:15:46 PM

Has anyone read this yet? I'm extremely skeptical of any book published after ~2024 just because of how much AI slop is in physical books these days.

Looks interesting, but I'd like to see some reviews on the content first before buying.

by seamossfet

8/22/2026 at 3:06:56 PM

Bill Pollock (who is nonstarch press) is a very serious editor so I would not worry about that with NSP books.

Of course it’s a small operation, and so isn’t under the same pressures of a big operation. When you’re a boutique operation your only differentiation is quality. Your rule of thumb is unfortunately probably pretty good in the general case.

by gumby

8/22/2026 at 3:20:50 PM

The editor is not the one producing the text. For this book, the sample chapter seems to make heavy use of LLM output. Pangram agrees. I guess that for books about AI it's fair game? Eat your own dog food so to speak. People who don't like AI are not the target audience.

by matherial

8/22/2026 at 4:26:29 PM

> extremely skeptical of any book published after ~2024

The problem is with that 'any':

> Drawing on more than 30 years of embedded engineering experience, David Such takes you through the complete engineering process

Would David Such ruin his reputation after 30 years building it?

by mdp2021

8/22/2026 at 4:36:45 PM

Few reputations remain unruined now.

by thatjoeoverthr

8/22/2026 at 3:55:16 PM

kind of the expensive side.

honest question, how is it better than posting toc on chatgpt and asking it to explain it to you

by dominotw

8/22/2026 at 4:16:34 PM

> honest question, how is [reading a book] better than posting toc on chatgpt and asking it to explain it to you

Since you're currently asking humans to explain this to you and not LLMs, you've already got your answer.

by schaefer

8/22/2026 at 4:23:58 PM

Have you noticed, as part of the over-process you imply, that LLMs absorb knowledge but do not produce it?

by mdp2021

8/22/2026 at 4:01:25 PM

Nice to see NSP cover this space. Curious if it goes into quantization for real hardware targets, or stays at the framework level.

by chenzheng