The Alignment Problem: How Can Artificial Intelligence Learn Human Values?

Much more interesting than the title suggests, this is about ai in general and the risks it possess. Very readable.

Algorithm are used in every branch of society. In this thoroughly researched book, the author addresses the alignment problem, meaning how automata does what we intend it to do.

Provides a good historical review - and articulates the problem - but I bought it for the subtitle "How can artificial intelligence learn human values?" - this it does not address. authors like Gary Marcus take a view which is contrarian but at least provides a possible avenue

The AI alignment problem is not that the AI might draw politically uncomfortable conclusions from its data, nor is it that certain forms of insurance might be mispriced.The problem with misaligned AI is that it will kill everyone on the planet, and destroy everything in the universe.The author seems to not have noticed this, despite writing nearly 400 pages of human-interest stories about people interested in the alignment problem, many of whom understand this perfectly well.If you're the sort of person who likes reading what the New York Times has to say about particle physics, this is the book for you.If you've got some ability to think, maybe try Nick Bostrom's "Superintelligence", which this author has clearly read, and equally clearly failed to understand.

Really interesting with a lot of old and new examples.I really like the writing style and the way the author told the 'stories'.

This book gives a very good overview of the different science fields and applications behind the development of AI however fails quite significantly in the aspects that I consider to be more relevant with regards to AI - philosophy, ethics and morality. It was entertaining but decisively unfulfilling.

Quite thorough, definitely picked up some important principles that I wasn't aware of beforehand.

Gute Erläuterung des Themas und angenehm zu lesen.

Great read!