Hacker News new | comments | show | ask | jobs | submit login
Ask HN: Best books on AI
18 points by forgotmysn on July 25, 2017 | hide | past | web | favorite | 8 comments
It feels like the whole world wants to make sure the whole world knows how they feel about AI/AGI. I'm curious what the best books are to get a better understanding of AI than I can from just reading Medium and Techcrunch posts?

If you want to have a clear grasp of the AI/AGI challenge, "SuperIntelligence" by Oxford philosopher/mathematician Nick Bostrom is a must. https://www.amazon.co.uk/d/Books/Superintelligence-Dangers-S...

Now if you have a bit of a math/engineering background and want to understand how modern AI really works, the two best books/shortcuts to go from beginner to expert are: - Ian Goodfellow's "Deep Learning" http://www.deeplearningbook.org/ - Richard Sutton's "Reinforcement Learning: An Introduction" http://incompleteideas.net/sutton/book/

I found the first chapter of Deep Learning by Ian Goodfellow et. al. to do a good job going over the basic structure of AI as well as some useful history. The book is published online for free by MIT as well, which is great. www.deeplearningbook.org

My recommendation would be to go back to the original sources. Hardly anyone (on Medium or Techcrunch, that is) would know about Arthur Samuel's checkers playing program. His paper, "Some Studies in Machine Learning Using the Game of Checkers", can be found in the compilation, "Computers and Thought" by Feigenbaum and Feldman.

Another good resource is the three-volume set, "The Handbook of Artificial Intelligence ", also by Feigenbaum. It's a very thorough catalog of programs developed in the 60s and 70s that illustrate various techniques used in AI programs.

In the words of Alan Kay, "the past is vast".

Artificial Intelligence: Case Studies in Common Lisp, Peter Norvig. It is also a really good book on programming in general.

Norvig also wrote the classic Artificial Intelligence: A Modern Approach

I'm more a fan of his Common Lisp book because it has running code instead of pseudo-code and that code is simpler than the Common Lisp code he wrote for A Modern Approach.

I'm a fan of the AIMA book, too. But the Lisp code for it kind of failed to excite me...

I don't think Norvig was all that excited about it either (or at least whatever excitement there was may have worn off): http://norvig.com/python-lisp.html

Though I guess it is probably better if he describes it himself, https://news.ycombinator.com/item?id=1803815

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact