Recent books and brief AI reflections

We’re running a live experiment with very big levers and very little control.

Recent books and brief AI reflections
Rabuda hanging out in the wall overseeing the garden

I didn’t make time for writing this week, so I wanted to drop recommendations on some recent books I’ve read and am reading.

I’ll end with a brief current take on AI.

AI Engineering — Chip Huyen

https://learning.oreilly.com/library/view/ai-engineering/9781098166298/

Been out for less than a year and already feels a bit outdated, but it very deliberately covers principles instead of tactics to try and keep it as fresh as possible.

This is not a book for “How do I use Cursor?”, but rather a book for the basics of “How do I build Cursor?”

I’m still wrestling with it, but it covers the basics for any AI Engineer.

How AI Works — Ronald T. Kneusel

https://nostarch.com/how-ai-works

It’s a bit basic, but that’s the point. I love this author, and his other 2 books from No Starch Press, Math for Deep Learning and Practical Deep Learning, are my favorites on these topics.

Honestly, I’m unsure how necessary it is to understand how AI works in order to use it effectively.

But me? I learned math just so I could understand this thing better, so I’d rather err on the side of really understanding the fundamentals and principles, because the abstractions on top are certainly changing very rapidly.

The Thinking Machine — Stephen Witt

https://www.penguinrandomhouse.com/books/757558/the-thinking-machine-by-stephen-witt/

The story of Nvidia and its founder Jensen Huang. Honestly, quite a disturbing book, and the fact that Jensen’s p(doom), or opinion on probability that AI is an existential risk, is 0, is quite concerning.

That said, it’s nice to get some anecdotal views on the chip war, and also the type of no-holds-barred disputes that happen at the Silicon Valley top companies, if only to see if that’s for you.. or not.

Chip War — Chris Miller

https://www.simonandschuster.com/books/Chip-War/Chris-Miller/9781982172008

This is the most unsettling book of all. It talks about how technology in general and AI has driven the demand for chips through the roof, how that’s all dependent on a few countries: Hong Kong, South Korea, Singapore, US, and China, how the machine building those chips is only manufactured by a single company out of the Netherlands, and other factoids embroiled in geopolitical webs.

Definitely recommended read.

Brief AI reflections

I’m coming to some principles when it comes to AI. I don’t know if these are it, but here’s what I’m currently gravitating towards:

1) AI is a big deal

This is not bitcoin, metaverse, VR, or 3D. This is iPhone with Social Media, Internet, and TV.

Whether you like it or not, it’s here to stay, it’s transformative, and it’ll carry you with it.

2) AI is socio-political

Jobs. Energy. Taxes. Economies. Countries. Wars.

Where there is power, there is socio-political implications, and this is probably the biggest unbalancing power generation we had since the atomic bomb. Yes, more than the computer and the internet.

Understanding AI implications requires multi-disciplinary expertise that most of us simply don’t have, and opportunities for well-reasoned dialogue that most of us don’t have either.

3) AI is changing fast

The gap between the people good at this and the people bad at this is increasing unlike anything I’ve ever seen before. I have no expectations of it stopping.

While being an expert in some topics allowed you years of learning and, then, years of relying on that expertise, with AI that timeline is measured in months and, sometimes, weeks.

The gap between the “haves” and “have-nots” of AI expertise is already abysmal. And I think it will continue to increase.

Well, what does all of this mean, Dui?

Honestly, I don’t know. And I don’t think anybody does either.

We’re running a live experiment with very big levers and very little control — hoping either for a good outcome, or that we’ll have time to iterate our way out of a bad one.

But that train isn’t stopping anytime soon.

So buckle up.