Artificial Intelligence

Burtonbartijn reviewed Artificial Intelligence by Melanie Mitchell

None

4 stars

Een goed en diepgravend boek over AI voor de geinteresseerde leek. Deels geschiedenis, deels uitleg over de verschillende vormen van AI. Hier en daar nog best technisch om te volgen. Maar ook zeker veel ruimte voor ethische en filosofische vragen. Waar "Slim, slimmer, slimst" (van Bennie Mols) een hele leuke en bruikbare, maar wat simpele inleiding is, gaat dit boek echt veel meer de diepte in maar blijft (zij het met enige moeite) goed te volgen.

Aaron reviewed Artificial Intelligence by Melanie Mitchell

Artificial Intelligence

4 stars

Mitchell, a computer scientist, has written an engaging book about AI, including both its opportunities and problems. She comes down as more skeptical (or at least restrained) than AI enthusiasts, although she clearly sees benefits to the technology. More than any other book I have read, Mitchell skillfully walks non-technological people (like this reader) through technological explanations of what AI is doing. There are plenty of great visual examples in the book, demonstrating how AI can make elementary mistakes (identifying images of random dots as different animals, for instance).

More than anything else, reading this book gave me tremendous respect for what the human mind can accomplish. We use words like "intelligence" to describe what machines are doing, but nothing comes close to what the human mind can do easily.

Near the end of the book, Mitchell quotes the economist Sendhil Mullainathan: "We should be afraid. Not of intelligent machines. …

Mitchell, a computer scientist, has written an engaging book about AI, including both its opportunities and problems. She comes down as more skeptical (or at least restrained) than AI enthusiasts, although she clearly sees benefits to the technology. More than any other book I have read, Mitchell skillfully walks non-technological people (like this reader) through technological explanations of what AI is doing. There are plenty of great visual examples in the book, demonstrating how AI can make elementary mistakes (identifying images of random dots as different animals, for instance).

More than anything else, reading this book gave me tremendous respect for what the human mind can accomplish. We use words like "intelligence" to describe what machines are doing, but nothing comes close to what the human mind can do easily.

Near the end of the book, Mitchell quotes the economist Sendhil Mullainathan: "We should be afraid. Not of intelligent machines. But of machines making decisions that they do not have the intelligence to make. I am far more afraid of machine stupidity than machine intelligence. … Machines can make many many good decisions and then one day fail spectacularly on a tail event that did not appear in their training data." (page 279) The unrestrained hype surrounding AI may have the unfortunate effect of making us too trusting, and thereby blind to Mullainathan's advice.

rsf92@lectura.social reviewed Artificial Intelligence by Melanie Mitchell

Tit for tat

4 stars

[ESP] Es bastante interesante, aunque se pueda haber quedado un poco atrasado en su análisis de los chatbots, pues chatGPT es posterior a este libro.

Me ha parecido muy interesante el concepto de que la inteligencia va estrechamente ligada a la capacidad de "ejecutar" simulaciones en nuestro cerebro sobre sucesos que podrían ocurrir.

Como punto más negativo, me parece que es poco crítica ideas bastante estúpidas, como la singularidad, la superinteligencia y la ley de rendimientos crecientes.

[ENG] Pretty interesting book, even though it's now a little oudated in it's analysis of chatbots, since it came out before chatGPT.

I found the idea that intelligence is related to the ability to "run" simulations of possible events in out brain very interesting.

The most negative point to me is that it's not that critical with truly idiotic ideas like the singularity, superintelligence and the law of incrementing returns.

giantrobot reviewed Artificial Intelligence by Melanie Mitchell

Review of 'Artificial Intelligence' on 'Goodreads'

5 stars

An excellent and detailed look at where we are right now with artificial intelligence. Despite all the incredible progress AI has made so far, this also delves into showing how far away we still are from general-purpose AI and why that is such a challenge in itself.

Taylor reviewed Artificial Intelligence by Melanie Mitchell

Review of 'Artificial Intelligence' on 'Goodreads'

3 stars

A great breakdown of the current state of AI, detailing its strengths and weaknesses.

The author continuously highlights what we do well as humans and where AI falls short, even when competing against children. She covers AI strengths by discussing programs like AlphaGo and IBM Watson. She also summarizes the recent rise in basic AI services, beginning in the mid-2000’s with translation, auto-subtitles, virtual assistants, facial recognition in pictures, up to today’s standard for self-driving vehicles.

It’s very dry and overly technical in some chapters, especially the first few. I almost tapped out during her explanation of multi-layered neural networks, backpropagation, and perceptrons. But it gets better the further you go and does an excellent job breaking down the differences between specific and general intelligence.