Slashdot is powered by your submissions, so send in your scoop

 



Forgot your password?
typodupeerror

Submission + - AI Is Transforming Google Search. The Rest of the Web Is Next

catchblue22 writes: Yesterday, the 46-year-old Google veteran who oversees its search engine, Amit Singhal, announced his retirement. For much of his tenure, Singhai believed that Google’s search engine should be driven by algorithms based on definite rules that automatically generate a response to each query. His replacement, John Giannandrea is a proponent of using deep learning instead of hard coded algorithms; the "RankBrain" project uses deep neural networks to produce search results. Some at engineers at Google have shown skepticism of using neural networks for search: "...it’s hard to explain and ascertain why a particular search result ranks more highly than another result for a given query...It’s difficult to directly tweak a machine learning-based system to boost the importance of certain signals over others.”

Giannandrea's appointment shows that deep learning has arrived at Google Search. "By building learning systems, we don’t have to write these rules anymore...Increasingly, we’re discovering that if we can learn things rather than writing code, we can scale these things much better.”

Comment Re:You mean Space Coffin (Score 1) 101

These days, I have to seriously consider the possibility that you're part of a paid smear campaign by one of Musk's competitors. Because that's actually done these days. Ethical standards in marketing, never very high in the first place, have slipped that far. I suppose it's not the ethical standards of marketers that bother me so much, it's the public acceptance of such methods.

I have to agree with you. I have been watching Musk and his companies for a long time, and it seems to me that these trolling posts about Space X seemed to start appearing at a particular time, I think about two to three years ago. ULA, Space X's competitor in the US hired S-3 Group as a their propaganda and lobbying company about that time. I think what may have provoked it was that Space X was calling out ULA for using Russian rocket engines in its Atlas V rocket. They were involved in legal motions to prevent the Air Force from buying Russian engines. It seems to me that ULA likely realized that Space X was a real competitor. They likely wrote them off before as a joke because they Elon Musk kept making promises that seemed impossible. However those promises started coming true, albeit a bit later than promised. One thing that I have learned about Elon Musk is that he may make pronouncements that seem impossible, but usually they come true. He is quite brilliant and for the most part honest in what he says. He predicted that Space X would become a globally competitive launcher during a speech he made after their first successful launch, and this has come true. He predicted they would be able to land a first stage, and they succeeded last month. He predicted they would make a capsule that would land like a helicopter, and so far testing looks promising. I know the Mars idea seems unlikely at first glance, but if re-usability pans out (which seems likely because he has landed a first stage, which is the hardest part), then sending three or so astronauts on a fly-by won't be that expensive. Building a lander would be hard, but their experience with supersonic retropropulsion will help, as will their experience landing rockets. As for making fuel on Mars, I think that would require time and testing. I suspect the first flight will be a fly-by, just as it was with the Moon program. That is definitely feasible in the 10-year time frame.

Comment Re:You mean Space Coffin (Score 5, Interesting) 101

This monster is going to get people killed in the name of profit.

Imagine if the above intrepid poster typified decision makers in sixteenth century. They would never have sent out explorers such as Sir Francis Drake or Ferdinand Magellan on their great voyages to map the world. Hell, if all humans were all like this poster (and those who modded him up), these great explorers would never have existed. Judging by many of the comments on this article, we are turning into a society of Statler and Waldorfs who criticize from the sidelines while offering little of substance. So grow a pair, and remember that all of us are going to die. What are you going to do with your life?

Comment My own Marvin Minsky story on neural networks (Score 2) 76

Posted here first this morning (couple of types fixed): https://ma.tt/2016/01/minsky/#...

Wow, sad to hear the news. Marvin Minsky and I were academic peers of a sort -- he was one of George A. Miller's first students, and I was one of George's last students. :-) George told my parents something like I was the student who most reminded him of Marvin Minsky, except whereas he spent George's Air Force money, I spent my father's money. :-) Which was not quite true (I paid for a chunk of Princeton with the proceeds of a video game I wrote and with some loans) but it sounded funny. :-) My dad was actually mostly a blue collar worker, and my mother only later in life worked for county social services, so George may also not have realized my family was not that well off financially.

I met Marvin Minsky once in his MIT office in 1985 as I was graduating from Princeton. I likely gave him a copy of my thesis -- "Why Intelligence: Object, Evolution, Stability, and Model". I also wrote to him once in the 1990s about getting computer time for space habitat simulations (he was responsive in a positive way, but then I met my wife and so just let stuff like that drop). And I saw him in passing about fifteen years ago when he gave a talk at IBM Research while I was a contractor there (he spoke about multiple simultaneous mental representations, and picking from the best one). A nephew of his even lived down the hall from me my senior year at Princeton in 1903 hall, too, but I never talked with him about his uncle. But we never really connected any of those times, sadly.

One of the biggest mistake I've made in my life careerwise (or so it seemed at the time) was when visiting Marvin Minsky in his office to talk to him about the triplestore and semantic network ideas in my thesis (stuff that indirectly helped inspire WordNet which George started as I graduated). I casually mentioned in passing to Marvin Minsky very early on in our meeting something about neural networks (MIT had a spinoff then of the Connection Machine), and I guess that may have put him in one of those mental states where some of the 400 different little computers activate. :-) I had not known then that he had essentially written a book (Perceptrons) to discredit neural networks (by only considering a limited version of them) to preserve funding for more formal semantic networks he worked on. He warned me sternly about how many careers had been destroyed by exploring neural networks. Another of George's students had found a copy of Marvin's original SNARC paper (what Marvin spent George's Air Force grant money on), and I can wish I had thought to take a copy to Marvin, as that might have set a different tone for our meeting, as it turned out Marvin had lost his original and wanted to reference it in his book "the Society of Mind" he was working on then.

So, instead of MIT, I spent a year hanging out in Hans Moravec's and also Red Whittaker's robot labs, and that was interesting in its own way. That experience also set me to thinking about the implications of most of the CMU robotics work being funded by the US military, which ultimately lead to my key insight about the irony of using robots to fight about material scarcity they could otherwise alleviate.

I sent Marvin Minsky an email in 2010, with a subject of "Vitamin D, computing, and abundance", warning about the health risks of vitamin D deficiency for heavy computer users. I also thanked him for his interactions with James P. Hogan, an author whose writings have been very inspiring to me (like Two Faces of Tomorrow and Voyage From Yesteryear), as James acknowledges Marvin in the first as a major source of ideas and inspirations, so some big ideas went from Marvin to James to me at least in that sense. :-) I also thanked him for being such an inspiration in years gone by. I had been reading through all the comments at a Wired article on "DARPA: U.S. Geek Shortage Is National Security Risk" and reflecting on my own career and inspirations, and thought I'd write to him. I told Marvin the biggest thing missed in that article is just that most computer jobs in the USA are not given much room for creative expression these days. And, sadly, much the same is true for academia for reason Dr. Goodstein outlines in his "The Big Crunch" essay (as do many other authors) which I also linked to. When you think about it, even as billions of dollars are being poured into proprietary software every year, and maybe a similar amount into academia, most programmers are kept on very short leashes focused on narrow project outcomes. There are very few projects like Mozilla or Chandler with a broad mandate (and when there are, they are often badly managed). When you think about it, for example, where is the broad support for creating better GUI libraries? Sure you can point to Angular or React as spinoffs of big corporations and they are getting most of the mindshare, but the very limited support for truly creative people like Leo Horie who made Mithril on the side in their spare time is more typical. Likewise, Notch and Minecraft got all the publicity and billions but Infiniminer and Zachary Barth, the source of key Minecraft ideas, get very little returns or support. Even Bill Gates learned by dumpster diving and reading the TOPS-10 OS listings and his MS-DOS was purchased as a ripoff of Kildall's CPM, which IBM may have known about and used Gates to stay at arms length from a suspicious transaction. Sigh.

I have since some to think that, short of improved subsistence via 3D printing and flexible home and agricultural personal robotics, or a radical change to a gift economy, or broad government grants totaling in the hundreds of billions of years to any programmer who asked, about the only thing I can think of that would really fix that situation of limited time for programmers to be creative, that would really give most programmers some financial freedom to innovate, not just a few (like Marvin) who manage (often by technical brilliance of a sort, and so seemingly "deservedly") to work their way up the social/funding hierarchy, would be a "basic income" for everyone. Then any programmer who wanted to could live life a graduate student their entire life (but without grad school restrictions like pleasing an adviser) and turn out free/libre and open source software. And others might choose to do other things with that freedom (have kids, teach, write books, paint, whatever). Most such creative programming projects would fail of course, but we might still see a lot of great innovative socially-useful stuff, where programmers would have the time to really support it.

I included in that email links to my Post-Scarcity Princeton writings. That email to Marvin Minsky was also when I first created my email sig, to, as I said to him, sum up the most important thing I've learned over the past 25 years by following the road less traveled (via CMU). :-) The version then was: "The biggest challenge of the 21st century is technologies of abundance in the hands of those thinking in terms of scarcity." I changed "thinking" to "still thinking" later to be a bit more optimistic. :-)

It is a sad day for us and his family, but Marvin apparently had one of the most fun careers of anyone I can imagine, so I can't feel too sad for Marvin himself. I am sad though that I said the wrong thing incidentally in his office and so never got to be part of that fun. But a deep question to ask is, how can more people have a fun and creative life like Marvin Minsky had?

Comment Re:Well... (Score 2) 232

Agreed. I encourage folks to check out Rocket Jump's video Why CG Sucks (Except it Doesn't). If you don't, here's the short version: we don't notice the good digital effects because they're so good or so subtle. We usually only notice the bad stuff.

One of my favourite movies, Master and Commander, uses CG, but it is not obvious that it does. I think that is the best use of CG, when it is largely invisible and not flaunting itself in your face. Mad Max, Fury Road is another great example.

Comment Re:TV ratings methodology (Score 5, Interesting) 302

In general, I prefer Netflix's system that isn't based on ad revenue but rather subscription revenue. The ad revenue system seems to encourage broadcasters to seek the lowest common denominator in their audience. Netflix has won me over with series such as Daredevil and House of Cards. Other subscription based services also seem to produce better material, the best example likely being HBO with series such as Game of Thrones. I have little sympathy for the old broadcasters. They are dinosaurs and should pass into oblivion in my opinion.

Comment Re:Does it have to be the whole booster? (Score 1) 118

Purely speaking from an economic standpoint, it would also make sense to do things differently. One could also just go for the most expensive part, which are the engines and avionics, and, depending on how you manage to retrieve them, it could actually be better.

Yeah, I hear the engines and avionics are the most expensive part of a 747. Who don't we just ditch the fuselage after every flight, and keep the engines and avionics. Would that make economic sense?

You sound like a ULA shill.

Comment Re:Actual update! (Score 1) 118

This was the last launch of v1.1 of the Falcon 9. As I understand it, v1.2 (sometimes called v1.1 Full Thrust) has upgraded landing legs. In either case, I would not call this a failure. The payload was placed into orbit. The touchdown speed was in fact normal. For some reason, one landing leg didn't lock. The landing is considered an experiment anyways. Isn't it good to do experiments? Don't you learn from them?

Comment Re:It's really too soon for this post. (Score 4, Informative) 118

The latest tweets from Musk indicate that on reading the data, the landing was not "hard". Apparently one of the legs failed to lock. Also it landed 1.3m from the center.

Elon Musk @elonmusk 6h6 hours ago

Definitely harder to land on a ship. Similar to an aircraft carrier vs land: much smaller target area, that's also translating & rotating.

Elon Musk @elonmusk 6h6 hours ago

However, that was not what prevented it being good. Touchdown speed was ok, but a leg lockout didn't latch, so it tipped over after landing.

Most of the posts in this discussion are based on incomplete information.

Comment funding policies in automotive intelligence & (Score 1) 276

The below is from me originally from 2001: http://www.pdfernhout.net/on-f...

Although see also this idea from a couple of weeks ago: http://www.pdfernhout.net/pled...
====
Consider again the self-driving cars mentioned earlier which now cruise some streets in small numbers. The software "intelligence" doing the driving was primarily developed by public money given to universities, which generally own the copyrights and patents as the contractors. Obviously there are related scientific publications, but in practice these fail to do justice to the complexity of such systems. The truest physical representation of the knowledge learned by such work is the codebase plus email discussions of it (plus what developers carry in their heads).

We are about to see the emergence of companies licensing that publicly funded software and selling modified versions of such software as proprietary products. There will eventually be hundreds or thousands of paid automotive software engineers working on such software no matter how it is funded, because there will be great value in having such self-driving vehicles given the result of America's horrendous urban planning policies leaving the car as generally the most efficient means of transport in the suburb. The question is, will the results of the work be open for inspection and contribution by the public? Essentially, will those engineers and their employers be "owners" of the software, or will they instead be "stewards" of a larger free and open community development process?

Open source software is typically eventually of much higher quality ( http://www.fsf.org/software/re... ) and reliability because more eyes look over the code for problems and more voices contribute to adding innovative solutions. About 35,000 Americans are killed every year in driving fatalities, and hundreds of thousands more are seriously injured. Should the software that keeps people safe on roads, and which has already been created primarily with public funds, not also be kept under continuous public scrutiny?

Without concerted action, such software will likely be kept proprietary because that will be more profitable sooner to the people who get in early, and will fit into conventional expectations of business as usual. It will likely end up being available for inspection and testing at best to a few government employees under non-disclosure agreements. We are talking about an entire publicly funded infrastructure about to disappear from the public radar screen. There is something deeply wrong here.

And while it is true many planes like the 757 can fly themselves already for most of their journey, and their software is probably mostly proprietary, the software involved in driving is potentially far more complex as it requires visual recognition of cues in a more complex environment full of many more unpredictable agents operating on much faster timescales. Also, automotive intelligence will touch all of our lives on a daily basis, where as aircraft intelligence can be generally avoided in daily life.

Decisions on how this public intellectual property related to automotive intelligence will be handled will affect the health and safety of every American and later everyone in any developed country. Either way, the automotive software engineers and their employers will do well financially (for example, one might still buy a Volvo because their software engineers are better and they do more thorough testing of configurations). But which way will the public be better off:
* totally dependent on proprietary intelligences under the hoods of their cars which they have no way of understanding, or instead
* with ways to verify what those intelligences do, understand how they operate, and make contributions when they can so such automotive intelligences serve humane purposes better?

If, for example, automotive intelligence was developed under some form of copyleft license like the GNU General Public License, then at least car owners or their "software mechanics" would be assured they could have access to the software in source form to ensure safe operation. What might be "street legal" in terms of software modifications might be a different story -- in the same way people can't legally drive with a cracked windshield or a broken headlight. For example, software changes might need to first be proven safe in simulation before being provisionally "street legal". But, the important thing is, foundations or government agencies funding code development could insist on some form of free licensing terms for automotive intelligence as a matter of public policy.

There are many other areas of human activities that the exponential growth of technology will effect. Automotive intelligence is just one of them that is here now and which I am familiar with from tangential interactions at universities with people developing it. In enough time similar issues will arise for the software behind household robotics or intelligent devices that assist the elderly or handicapped. The IBOT wheelchair by Dean Kamen using complex software to balance on two wheels is just the beginning of such devices.

Note the IBOT wheelchair was developed entirely with private funds it seems, so the reasoning in this essay does not apply directly to it. Also, in general Dean Kamen is a role model of a socially responsible for-profit inventor. Still, the issue arises of whether "Johnson & Johnson" should be funding such development, as was the case, as opposed to, say, the "Robert Wood Johnson Foundation", as was not, given the public policy issue of whether individuals should be continually dependent for personal needs on proprietary software. In either case it would be worth it to pay billions for such innovation, and the public will pay that in the end as a toll on for such devices.

There is a real question here of how our society will proceed -- mainly closed or mainly open. It is reflected in everything the non-profit world does -- including the myths it lives by. The choice of myth can be made in part by the funding policies set by foundations and government agencies. The myth that funders may be living by is the scarcity economics myth. How does that myth effect the digital public works funding cycle?

Comment Re:NASA who? (Score 1) 57

It's contracted out design and production since the Apollo era.

Yes and AFAIK, this was historically done with "cost plus" contracts that had no incentive for cost reduction. Basically the companies could name their cost for a project and then be guaranteed a profit margin on top. The companies would add complex management structures and build overly complex machines in order to maximize their profit. NASA became a key tool in dispersing "pork barrel" money to various congressional districts.

The model for the contracts with SpaceX and Orbital is "fixed cost for service". This gives an incentive for the companies to reduce costs, and they have. SpaceX is currently the least expensive launcher in the world, even without re-use of their rockets. And that includes China. The incumbent players, Lockheed Martin and Boeing have helped to created a PR campaign that brands the owner of SpaceX, Elon Musk as a "corporate welfare queen" partly for relying on government contracts through SpaceX. This coming from companies that, through defence contracts, rely mostly on government money. It is quite absurd propaganda.

Comment Re:Bull Spit (Score 4, Insightful) 259

You might want to get a better grip on reality. Tesla Autopilot is already 80% of the way there, and the other 20% may not be available to consumers yet, but it has had millions of miles of testing...

It seems that Slashdot has been infested with willfully ignorant ball-less trolls. This is supposed to be a site for nerds. There is no greater nerd than Elon Musk. He is infused in sci-fi. He builds rockets...he designed much of the first SpaceX rocket (Falcon-1) himself. He builds arguably the best car in the world, and certainly the most technologically advanced (the Model S). It has the most advanced auto-driving features of any production car in the world. He literally bet the entire fortune he made from the sale of Paypal (200 million dollars) on Tesla and SpaceX after the 2008 market crash; most so called capitalists in our elite would never take such risks. Any libertarians amongst the readership here should worship Musk. He is more the Ayn Randian superhero than anyone I can think of. And if they return that Musk has taken some government help (like money for building the Dragon capsule to ferry cargo to the Space station for NASA or a $7500 subsidy for clean energy vehicle purchases), I would ask them what they think of defence contractors such as Lockheed Martin who receive 75% or more of their income from government contracts, or oil companies who have literally had wars fought in their name by governments. If those so-called libertarians don't denounce such things, then they are the worst type of corporate troll hypocrites.

Comment Re:Reliability (Score 1) 163

Still, there are obstacles. SpaceX still needs to demonstrate the ability to consistently produce and launch rockets many times a year after the June accident caused an unexpected, six-month setback, something it will do with several flights planned for the weeks ahead.

Just because it's relatively cheap to use Space X, if I have a 50-50 ( better or worse) chance that my $100 million satellite that took several years to design and build is going to get blown up, I'll pass.

One could make the argument, especially during the earlier launches, that using a rocket that has been tested three or four times already might give more reliability, and not less. Time will tell.

Slashdot Top Deals

10.0 times 0.1 is hardly ever 1.0.

Working...