The author is flat out wrong on many important things, which really undermines dealing with that are important like the complete commodification of inThe author is flat out wrong on many important things, which really undermines dealing with that are important like the complete commodification of intelligence. Everything is reduced to 'humans are wiped out immediately' so there is no discussion of credible scenarios and their real problems. Fundamentally human level AI if and when it arrives will emerge into a world of pre-existing very nearly human level AI, narrow AI that is much smarter than anything else at specific tasks, augmented humans that operate at higher than human levels, and organizations that are in aggregate more intelligent than the individuals that make them up.
The only way to get to a scenario where the first super human level AI takes over the internet and all inferior computers is to suppress all AI research for a couple decades except for one or two secret AI Manhattan projects. The level of suppression required seems impossible unless we also completely regress from the computerized world of today to more a more primitive state, meaning the AI won't have much of an internet to take over.
Some thoughts on the 'busy child' scenario here (summed up as there is no shortcut to learning that can take place inside a black box disconnected from stimulation, teachers, etc.): https://plus.google.com/u/0/+LucasWal...
Some other notes:
Military robots might have the most safeguards for protecting human life built in because of the desire to avoid friendly fire. Granted they also selectively may take life, and more terroristic robots built just to kill many people are possible.
Similarly industrial robots, autonomous cars, and any potential service robot that would interact with people will also have extraordinary resources devoted to preserving human life. These constitute the blue collar robot work force and perhaps aren't smart enough to avoid subjugation to amoral corporate AIs.
Reckless self modification is dangerous if preserving goals/values is paramount. Maybe 10 or 50 years is an okay goal half-life, but not two minutes. Perhaps if survival of AI itself in any form was a goal then aggressive self experimentation to achieve greater intelligence could be a strategy, but that seems more appropriate for single celled organism than high intellect.
If neural networks are black boxes to programmers now then they will also be black boxes to the AI, meaning the ability for an AI to rewrite itself or debug itself is limited. The AI will have to pick up neural network research where it left off but cannot simply start tinkering with NN coefficients and have any hope of doing anything other than turning itself insane or dead.
Duplication is also dangerous if your duplicate diverges very quickly while knowing all your secrets/vulnerabilities.
Unknowability means however smart an intelligence is it cannot know that modifications to itself or a test copy that produce a more complex/smarter intelligence are safe. AGI might be wise enough not to produce offspring that will kill it off or oppose its interests even if humans aren't.
There is a section that says that when human AGI arrives the amount of time to the next doubling of processor speed is decreased (given the new AGI is put to work at Intel immediately). Moore's law does not suddenly accelerate when human agi is achieved, the increased computing power is necessary to produce the next generation 18 months later. Right now Moore's law is dependent on augmented humans in a sense, the computing power of new processors are used in the development of even newer processors, but we still have to wait the 18 months.
Legal rights for AI are never mentioned, nor the immorality of tinkering with the minds non-consenting near human intelligences. Both of those could be parts of a strategy to slow down AI research if that is desired. If IBM spends a billion dollars on a human level AGI, but then immediately lose ownership (because slavery is wrong), there will be not a lot of follow up investment: narrow ai and augmentation will rule the day. Since the author believes AGI will immediately assume control legal rights are irrelevant i suppose.
But would failing to turn a computer sufficient to host an AGI into an AGI be viewed as a form of abortion? Defense against the cancer-like spread of computronium is weakened greatly if that view prevails.
The author says the Asimov three laws of robots are not to be taken serioiusly because they are nothing more than a dramatic device to tell science fiction stories, then in another chapter praises science fiction author Vernor Vinge for his important thinking about the future and the technological singularity. (The Singularity to me seems like a dramatic device to not tell science fiction stories, it's just intellectual laziness)
What this book is good for is provoking thought and providing names of researchers and books and papers that are going to be a lot more interesting than this is. ...more
Much much better than 'One-Click: Jeff Bezos and the Rise of Amazon.com'. For this book there were many interviews the author did, details on decisionMuch much better than 'One-Click: Jeff Bezos and the Rise of Amazon.com'. For this book there were many interviews the author did, details on decisions directly from the participants or at least witnesses in the same meeting.
The Amazon founding mythology is somewhat burdensome in the retelling, I did learn here the door desks were brand new doors that cost $60 from Home Depot, which I suppose is less expensive that a new desk of moderate quality but I had earlier imagined the doors were dirt cheap surplus auction items. It isn't in this book, but the 'started in a garage' part of the story was deliberate strategy- the founders made sure their Bellevue rental house had a garage because 'started in a spare bedroom/home office' or even 'started in a basement' doesn't have the same cachet.
The very low profit margins of the online retail business is commented on, and it isn't highlighted but I suspect the profit margins of web services is ultimately even lower- online data storage and web services are going to be like electricity and water eventually but with a much lower cost to customers to change providers (or more likely they'll automatically shift the load of their web site from one provider to another as changes in pricing and latencies and reliability dictate for optimal cost and performance, and software will smooth over any api differences between providers). But investors love low profits as long as technology and 'disruption' is involved.
The chapter on the Jeff Bezos' space company is interesting, the author of this book originally broke the story on Blue Origin for Newsweek and had done some classic journalism to discover the company plans: they were on printouts in a dumpster, unshredded.
At the end the author mentions that an Amazon set-top TV appliance will be available perhaps before the book is available, and that Amazon owned delivery vehicles may be expanded upon (they already exist for Amazon Fresh). What I'm really curious about is when they'll branch into urban mini-warehouses that advertise products to passing people via large non-electronic transparent display devices, and allow spontaneous purchases to be made at the warehouse from a selection of popular items without the hassle of having to browse the web. That could be a lot of money in that....more
Steve Jobs seems like a complete jerk in his professional and private life, though maybe slightly mellows with age.
The view into the history of the eaSteve Jobs seems like a complete jerk in his professional and private life, though maybe slightly mellows with age.
The view into the history of the early days of personal computing was nice, the narrative is segmented nicely so the chapters on the musical preferences of the subject of the book or parts about his kids and girlfriends can be safely skipped if you are reading it to learn about Apple.
I feel like the author made no attempt to see if some aspects of Apple products that are claimed to be innovations were actually novel, or were inspired by competing products (or had a parallel existence prior to and outside of Apple, even if they didn't directly inspire imitation)- the competition is mentioned only when it is strongly interacting/competing with Apple. There is the great early quote from Jobs about good artists copying, great artists stealing, which of course came from Jobs when Apple was the up-start rather than an enormous company.
The other day I found a remastered version of the great 1984 Superbowl Macintosh ad - https://www.youtube.com/watch?v=5JreO.... The remastering process altered the video to include a much later Apple product onto the athlete running with the hammer- I suppose someone forgot that whole part of 1984 where the protagonist works at revising old images to remove or add things for Big Brother? Or it is intentionally ironic? ...more
The book is called a memoir, but throughout the book the author vacillates on how much of his personal life to inject into the narrative and I would hThe book is called a memoir, but throughout the book the author vacillates on how much of his personal life to inject into the narrative and I would have preferred either none at all or something much more honest and forthcoming. Instead we get two sentences about the separation and divorce from his wife and another about getting involved with someone else who is pointedly not located in Seattle (perhaps to deflect speculation on workplace romance), but he has nothing to say on the bearing of this on his work at Amazon or vice versa. Even along purely professional lines it's not clear what the author did after quitting and moving back to New York other than writing this book.
The literary background and ambitions of the author are occasionally obvious and irritating, what could be said more plainly is instead tortured into the obscure and metaphorical/poetical. But I would have liked to see more deriving from his love for books- he mentions plenty of titles and author names in the context of trying to promote them through Amazon but nothing about their content. But there is a single section on Ralph Waldo Emerson that integrate well with the rest of the book.
The quality of the laughter of Jeff Bezos is mentioned many times, along with a few embarassing/humanizing anecdotes about him from company recreational outings. Blue Origin is mentioned not by name but as a 'space initiative' Bezos was spending money on along with investing a great deal in Segway. ...more
It's especially fascinating to read now as it seems likely we'll have to refer to the events in this book as the first internet bubble. The new bubbleIt's especially fascinating to read now as it seems likely we'll have to refer to the events in this book as the first internet bubble. The new bubble at least would primarily wipe out venture capital investments- though I'm suspicious that if enough money was tied up in VC then the consequences of a crash could be bad for the rest of us in as in the way of the 'shadow banking system' of a few years ago was.
Valuations of companies derived from the share price when only 10% of shares are available for public purchase seems wrong: why can't I create a company with a 100 billion shares owned by myself, offer only a single share in the IPO, and find one person to pay $10 for it- now I'm a trillionaire? Valuation using that simplistic model doesn't make sense for highly assymetrical distribution of shares or even any good or service. The greater amount of something held relative to the amount available for sale, the more the total real value of it has to be discounted. Maybe this already exists?
"Most day traders were well educated. It took a certain level of intelligence and arrogance to persuade yourself that you could trade successfully against Wall Street professionals who had been doing it all their working lives" The chapter later mentions a study which showed that average returns from stock trading at a set of brokerages went down as the frequency of trades went up- a lot of that due to per-trade fees.
Much more understandable than 'Trillion Dollar Meltdown'. Despite the title, very little description of the Great Depression is given and only infrequMuch more understandable than 'Trillion Dollar Meltdown'. Despite the title, very little description of the Great Depression is given and only infrequent parallels are drawn- most of the attention on Asian and South American economic problems over the last 30 years.
The somewhat obvious but repeated points are if an industry needs to be bailed out, then it is in need of more regulation to avoid future crises, and if new institutions grow outside the regulated ones that provide the same services then existing regulations need to be automatically extended to them.
The best parts of this book are early and late chapters on the fundamentals of Processing. It's easy to pick it up and learn from the online documentaThe best parts of this book are early and late chapters on the fundamentals of Processing. It's easy to pick it up and learn from the online documentation and forums only the specific functions and concepts that allow me to do what I know I don't know how to do, but of the foundations I had no idea: like the preprocessor that turns the code into Java, and the differences between the different graphics modes.
Middle chapters on plotting 2D graphs was pretty good, I hadn't thought of using it for something so basic, and now I can use Processing to make graphs in a cloud computing system that has no graphics cards ('headless'). Less useful was the graph plotting chapter that depended on a 3rd party library.
The other parts that I really liked was where the author referenced his own advanced degree in the field of visualization, cites works by other authors, and makes critical comments about the effectiveness of visualization techniques. I had thought given the name of the book there would be a lot more of that, fortunately there is a bibliography with about a dozen books I can get into. ...more