Scan barcode
sascha_z's review
2.0
Life 3.0 is best described as a downhill slope.
The book starts out with its climax – the title. “Life 3.0” is the author’s cleverly self-referential code for artificial intelligence (AI), more precisely beyond-human-level (or general) intelligence. In Tegmark’s view, this contrasts with low-level “Life 1.0” (where any kind of intelligence or “software” is genetically hard-wired and only changes with evolution) and human “Life 2.0” (whose software can be changed through learning and communication). By contrast, artificial (super-)intelligence is based on “hardware” that is no longer subject to evolution but may be designed at will. Just at whose will?
A hint at the potential opportunities and dangers inherent in unleashing superintelligence is given in the book’s engaging prelude. It is the story of a fictitious Google-style “Omega team”, who use AI to... well, take over the world. Rather than explicitly introducing this as a thought experiment, it dives straight into a world which, at least in the story’s early pages, might well be ours. That makes it genuinely scary and thought-provoking. It is bit quirky a start for a work of non-fiction but, ironically, by far the book’s best and most intriguing part.
It goes downhill from there. In the first two regular chapters, Tegmark lays out some introductory concepts, ranging from biological evolution and intelligence to complexity, memory, and learning. He does a decent job at juxtaposing biologically familiar terms and more general, artificially created counterparts. However, I did not feel this part was as inspiring as it might have been, and I did not feel I learned a whole lot about the mechanisms of machine learning etc.
The book then moves on to inspect the implications of AI upon the near future, such as consequences for the job market or weapons. This part probably is most relevant to the present-day discussion of AI. However, its presentation was also somewhat uninspiring, as if the author was keen to move on to his own favorite topics in the ensuing chapters. This is where the book starts to drift off completely into wild speculation:
How exactly will AI take over the world?
How will it rule it in the next 10,000 years?
And how will it perpetuate life throughout the cosmos over the next billion years?
I agree all these are conceivable scenarios, but I found them so utterly hypothetical and bleak that I prayed for the chapters to wrap up soon. They didn’t. Had these scenarios been presented by a gifted sci-fi writer, they might have caught my imagination. But he was ticking them off in the spirit of a checklist, rounded off by a dash of alarmism.
In the closing chapters, the problem of anchoring human-friendly goals in AI is discussed, along with some musings on consciousness. Especially the latter topic might have been tremendously interesting, had it not been treated in an equally superficial and dull fashion. However, the ultimate low point of the book comes with its fairly unnecessary epilogue – where the author brags at tedious length about setting up his own institute and how he got Elon Musk to fund it.
This is emblematic of the book’s weakness: It might well have been much more focused, presented with a strong narrative, and stripped of an annoying amount of name-dropping and self-advertising. Unfortunately, the abundant use of the word “amazing” does not make up for an inspired writing style.
The book starts out with its climax – the title. “Life 3.0” is the author’s cleverly self-referential code for artificial intelligence (AI), more precisely beyond-human-level (or general) intelligence. In Tegmark’s view, this contrasts with low-level “Life 1.0” (where any kind of intelligence or “software” is genetically hard-wired and only changes with evolution) and human “Life 2.0” (whose software can be changed through learning and communication). By contrast, artificial (super-)intelligence is based on “hardware” that is no longer subject to evolution but may be designed at will. Just at whose will?
A hint at the potential opportunities and dangers inherent in unleashing superintelligence is given in the book’s engaging prelude. It is the story of a fictitious Google-style “Omega team”, who use AI to... well, take over the world. Rather than explicitly introducing this as a thought experiment, it dives straight into a world which, at least in the story’s early pages, might well be ours. That makes it genuinely scary and thought-provoking. It is bit quirky a start for a work of non-fiction but, ironically, by far the book’s best and most intriguing part.
It goes downhill from there. In the first two regular chapters, Tegmark lays out some introductory concepts, ranging from biological evolution and intelligence to complexity, memory, and learning. He does a decent job at juxtaposing biologically familiar terms and more general, artificially created counterparts. However, I did not feel this part was as inspiring as it might have been, and I did not feel I learned a whole lot about the mechanisms of machine learning etc.
The book then moves on to inspect the implications of AI upon the near future, such as consequences for the job market or weapons. This part probably is most relevant to the present-day discussion of AI. However, its presentation was also somewhat uninspiring, as if the author was keen to move on to his own favorite topics in the ensuing chapters. This is where the book starts to drift off completely into wild speculation:
How exactly will AI take over the world?
How will it rule it in the next 10,000 years?
And how will it perpetuate life throughout the cosmos over the next billion years?
I agree all these are conceivable scenarios, but I found them so utterly hypothetical and bleak that I prayed for the chapters to wrap up soon. They didn’t. Had these scenarios been presented by a gifted sci-fi writer, they might have caught my imagination. But he was ticking them off in the spirit of a checklist, rounded off by a dash of alarmism.
In the closing chapters, the problem of anchoring human-friendly goals in AI is discussed, along with some musings on consciousness. Especially the latter topic might have been tremendously interesting, had it not been treated in an equally superficial and dull fashion. However, the ultimate low point of the book comes with its fairly unnecessary epilogue – where the author brags at tedious length about setting up his own institute and how he got Elon Musk to fund it.
This is emblematic of the book’s weakness: It might well have been much more focused, presented with a strong narrative, and stripped of an annoying amount of name-dropping and self-advertising. Unfortunately, the abundant use of the word “amazing” does not make up for an inspired writing style.
rihoke's review against another edition
4.0
I actually quite enjoyed reading this book. Got me thinking about the defining questions of the future. also gave me (a newcomer in the AI-studies field) a good overview on the problems we face today involving the ever evolving AI. I skipped the more speculative chapters, which was also "encouraged" by the author, for they didn't interest me as much.
stumpnugget's review against another edition
4.0
Excellent survey of the AI problem.
Also, we're doomed.
Maybe we aren't.
But, yeah, we probably are.
Also, we're doomed.
Maybe we aren't.
But, yeah, we probably are.
mxbh35's review
Good book but got sick of the topic. Too optimistic or too pessimistic, don’t know what to think and this is all hypothetical
jjzsilva9's review against another edition
5.0
Very interesting and covers a broad range of topics. Gave me several existential crises.
bookishwelshie's review against another edition
4.0
An important quote to remember when approaching the subject of AI (from one of my favourite writers Yuval Noah Harari) is: ”If you hear a scenario about the world in 2050 and it sounds like science fiction, it is probably wrong; but if you hear a scenario about the world in 2050 and it does not sound like science fiction, it is certainly wrong.”
I decided to pick up a copy of this after reading Harari’s “21 Lessons For The 21st Century” earlier this year.
Life 3.0 is an engaging read, written for a general audience (so people new to the subject can pick it up) covering the many different scenarios we could face in the future as AI technology advances. A certain few chapters cover the basics, with Tegmark encouraging those familiar with the topic to skip these, but I decided to read the entire book despite my familiarity with (and quite the avid interest in) the subject. Even though it is intended for a general audience; I’d say a decent knowledge of computers/tech and physics is needed to be able to follow it properly, otherwise some parts may seem quite convoluted and confusing. Quite a lot of equations and scientific diagrams are used.
The prologue is a hypothetical “Prometheus” project by a team called the Omegas. It outlines the ways a superhuman AI could impact our world and gives us a best case scenario. I feel as though it isn’t quite what would happen in reality, but it is an intriguing concept all the same and (of course) hard to predict.
This is referred back to in later chapters, as it is explored and analysed in greater detail. A scenario given here, about one of the Prometheus’ team’s deceased wife being seemingly “brought back from the dead” as an AI generated computer simulation, feels like something straight out of the television show “Black Mirror”. (I personally do not like too many references to “Black Mirror” when talking about the future, as it seems overused and lazy now BUT it is a comparison most people will “get”).
Tegmark states that life is divided into three stages which are as follows:
- Life 1.0 (biological stage): evolves its hardware and software
- Life 2.0 (cultural stage): evolves its hardware, designs much of its software
- Life 3.0 (technological stage): designs its hardware and software
Throughout the book, Tegmark makes his feelings known that it is only a matter of time before we invent truly intelligent machines, but doesn’t really say exactly when this might happen. All we really know is that technology has been advancing so rapidly within these past few decades that it is truly fascinating to ruminate over what could be next, perhaps sooner than we think.
One slight critique I have of it is the overuse of quite lengthy foot-noting every few pages.
I do feel as though most of the information given here, I have read or head about about before in some other form; in other books, countless articles and videos. So for that reason it’s a 3.5 stars overall from me, rounded up to a 4 mostly for the way it was written. I liked the way Tegmark explained everything and his input on the AI topic as a whole, as he appears to be very passionate about it.
I am in need of reading Nick Bostrom’s “Superintelligence” in the very near future.
I decided to pick up a copy of this after reading Harari’s “21 Lessons For The 21st Century” earlier this year.
Life 3.0 is an engaging read, written for a general audience (so people new to the subject can pick it up) covering the many different scenarios we could face in the future as AI technology advances. A certain few chapters cover the basics, with Tegmark encouraging those familiar with the topic to skip these, but I decided to read the entire book despite my familiarity with (and quite the avid interest in) the subject. Even though it is intended for a general audience; I’d say a decent knowledge of computers/tech and physics is needed to be able to follow it properly, otherwise some parts may seem quite convoluted and confusing. Quite a lot of equations and scientific diagrams are used.
The prologue is a hypothetical “Prometheus” project by a team called the Omegas. It outlines the ways a superhuman AI could impact our world and gives us a best case scenario. I feel as though it isn’t quite what would happen in reality, but it is an intriguing concept all the same and (of course) hard to predict.
This is referred back to in later chapters, as it is explored and analysed in greater detail. A scenario given here, about one of the Prometheus’ team’s deceased wife being seemingly “brought back from the dead” as an AI generated computer simulation, feels like something straight out of the television show “Black Mirror”. (I personally do not like too many references to “Black Mirror” when talking about the future, as it seems overused and lazy now BUT it is a comparison most people will “get”).
Tegmark states that life is divided into three stages which are as follows:
- Life 1.0 (biological stage): evolves its hardware and software
- Life 2.0 (cultural stage): evolves its hardware, designs much of its software
- Life 3.0 (technological stage): designs its hardware and software
Throughout the book, Tegmark makes his feelings known that it is only a matter of time before we invent truly intelligent machines, but doesn’t really say exactly when this might happen. All we really know is that technology has been advancing so rapidly within these past few decades that it is truly fascinating to ruminate over what could be next, perhaps sooner than we think.
One slight critique I have of it is the overuse of quite lengthy foot-noting every few pages.
I do feel as though most of the information given here, I have read or head about about before in some other form; in other books, countless articles and videos. So for that reason it’s a 3.5 stars overall from me, rounded up to a 4 mostly for the way it was written. I liked the way Tegmark explained everything and his input on the AI topic as a whole, as he appears to be very passionate about it.
I am in need of reading Nick Bostrom’s “Superintelligence” in the very near future.