I travelled over to the west coast and spent some time with the Artificial Intelligence team within Google at their headquarters just off Venice Beach in LA. Like all who visit that facility, I am constrained by an NDA in talking about what is going on. However in their bid to establish "Quantum Supremacy" the team, led by Hartmut Neven, talks not in terms of decades but in a timetable that is the technology equivalent of tomorrow. For the avoidance of doubt, the "tomorrow" that I refer to is the timeline for building and operating a universal quantum computer.I interpret "the technology equivalent of tomorrow" as being within two years. Check back here at that time.
No, Google is not going to succeed. This is not like self-driving cars, where it is clear that the technology is coming, as prototypes have proved feasibility. For that, computers just have to mimic what humans do, and have several advantages, such better sensors, faster reactions, and real-time access to maps.
Despite hundreds of millions of dollars in investment, there is still no convincing demonstration of quantum supremacy, or any proof that any method will scale.
Google is all about scale, so I am sure that its researchers have a story to tell their senior management. But it is covered by a non-disclosure agreement, so we do not know what it is.
You can bet that if Google ever achieves a universal quantum computer, or even just quantum supremacy, it will brag to everyone and attempt to collect a Nobel prize. If you do not hear anything in a couple of years, then they are not delivering on their promises.
Google also has a reputation for pretty ruthlessly killing products if they seem to be headed for a dead end. If you are right, they may not take two years to get to that decision.
ReplyDeleteThe company has never done an innovative thing in its life because it's full of coddled little pricks. The whole company is organized like some leftist cult and it overpays its employees like Adobe. They're a bubble like a lot of high tech and all they can offer is inferior services that nobody would pay for. They couldn't even put a paywall into YouTube or offer paid web hosting through Google Drive. Their cloud service just began to expand into business services and will attempt to compete with Amazon and Microsoft. They are always behind. I can see through their Glass and Cardboard and all I see is a one-trick pony.
DeleteOnce upon a time, not so long ago, there was a company which thought itself indispensable to the internet. Everyone believed this because the company said so, and had lots of magic paper called stocks which they convinced people were worth something. This company grew and grew, until one day someone realized, AOL was nothing but an overrated ISP with a proprietary search engine, and wasn't really worth much of the value of the magic paper they had sold so much of.
ReplyDeleteGoogle is AOL in all but name. Same sad story, same sad excuse of self importance for a company that doesn't really bring anything but the pretense of impartiality to web searches...the crucial word being 'pretense'.
Google even now happily wishes to invest heavily in AI technologies which (if they get what they wish for) would make their entire business superfluous. Any company so hell-bent on slitting its own financial wrists is not worth investing in.
Google is a company full of of coddled morons like Adobe. The difference is that Adobe makes software people actually need. They are hyped by Wall Street and the media but have no credentials.
DeleteGoogle is a total fifth-rate selling cardboard headsets and they couldn't even figure out how to put a paywall into YouTube. They are a sparse-matrix pagerank and balance algorithm recreating 90s office software in a browser. Their cloud service is a command-line and hasn't even expanded into enterprise and business services where Amazon completely clobbered them. Their coddled employees are cocky and overrated clowns. They are a major reason the outdated disaster called "the web" never improved and we still use unmaintainable HTML, styling and slow scripting languages (HTML 5 finally allows video embedding and WebGL but that's about it). I think they're a really awful company that couldn't learn to scale anything but search and that has not even got any better. It just retrieves quantity and not quality. It's actually a piece of garbage. They're a company sitting on billions dollars in cash deciding to get no return and they only invest in gimmicks. Self-driving cars are not going to work when the first one crashes. This is America and we don't do freedom anymore. These companies are bubbles that have been coddled by easy money and Wall Street hype. They just produce Darts and Gos.
ReplyDeleteDo you really think that self-driving cars will never get here? Sure, driving through Boston at rush hour is hard, but a lot of other routes are not. As sensors and computing power and map-searching get ever faster, I don't see how self-driving cars DON'T happen. Compared to where we've come in navigation and automobile technology, this doesn't seem like a huge additional technological step to me.
DeleteApparently, you can't read. It's not about technology but REGULATIONS. Ask Peter Thiel. Secondly, they do hype how flexible the system actually is. Many of the algorithms they use are not easily tweaked and it can't handle imperfect road conditions and weather. For instance, an optimized neural network is just an unintelligible tangle. It's hard to refine it and I have programed many myself. Computing power has absolutely nothing to do with it and only idiots trying to brute force all the problems think that it's the actual bottleneck. MIT has learning algorithms learning character recognition from single examples and not giant piles of moron data. I think full-scale AI is definitely possible but the present generation is too stupid to even get close. They can't even reverse engineer very small parts of rat brains.
DeleteNo, actually I read reasonably well. We disagree. No need to get nasty.
DeleteRoger, ilya Khan is not the best person to quote. The guy has absolutely no degrees or training in any STEM field, only in philosophy. His company Cambridge Quantum Computing made a big uproar in sept 2015 claiming that it had been bought by Grupo Arcano owned by Alberto Chang in Chile for 50 million dollars. But In the last month the Chilean government has pressed charges against Chang for fraud and pyramid schemes. Chang has fled Chile and is currently in Malta
ReplyDeleteROFL! I can't let you get away with such a stupid appeal to authority. When exactly did a STEM degree have much credibility? CS people sit around studying Godel and theoretical complexity. They create another context-free language every time they encounter a bad library. It's a complete embarrassment and people that go through these degree programs can't even code Boyer-Moore, let alone play around with fiber bundles. There isn't the slightest empirical evidence that they learn anything.
DeleteWindows bloats to over 100 million lines of code and runs slower on a machine a million times faster and larger. Windows 98 boots in 5 seconds on a modern machine and I have video to prove it. Alan Kay can run the whole Squeak IDE in 1.8 MB and a whole system in about 2.8 MB. It's a thoroughly discredited field and I'm astonished you can't understand that. The latest Visual Studio was 30 GB with emulators and it takes 5 minutes just to get to the start screen (on my i7).
Julia runs about as fast as C, which is about 50 times faster than Python, even though it has the same capability, yet guess what schools like MIT use for data science. My computer has a 3 TFLOP/s video card but a few adds on a webpage stutter and choke the whole system. Total clowns. They should shut down CS departments for being overly theoretical.
They should shut down CS departments for being overly theoretical.
DeleteAbsolutely agree. Bunch of mathematician wannabes. All Failures I tell you. Matt, you are talking to a programmer.
Thanks. I did not know that. I was quoting Khan for what Google is privately telling people. Yes, it could be distorted.
ReplyDeleteRoger, I totally agree with this.
ReplyDeletehttps://blog.cr.yp.to/20160516-quantum.html