Slashdot Mirror

← Back to Stories (view on slashdot.org)

Google Has Made It Simple For Anyone To Tap Into Its Image Recognition AI (gizmodo.com)

Posted by BeauHD on from the easy-peasy dept.

An anonymous reader quotes a report from Gizmodo: Google released a new AI tool on Wednesday designed to let anyone train its machine learning systems on a photo dataset of their choosing. The software is called Cloud AutoML Vision. In an accompanying blog post, the chief scientist of Google's Cloud AI division explains how the software can help users without machine learning backgrounds harness artificial intelligence. All hype aside, training the AI does appear to be surprisingly simple. First, you'll need a ton of tagged images. The minimum is 20, but the software supports up to 10,000. Using a meteorologist as an example for their promotional video was an apt choice by Google -- not many people have thousands of tagged HD images bundled together and ready to upload. A lot of image recognition is about identifying patterns. Once Google's AI thinks it has a good understanding of what links together the images you've uploaded, it can be used to look for that pattern in new uploads, spitting out a number for how well it thinks the new images match it. So our meteorologist would eventually be able to upload images as the weather changes, identifying clouds while continuing to train and improve the software.

42 comments

  1. They want the crowd to train their machine by Anonymous Coward · · Score: 1

    because it's free....

    1. Re:They want the crowd to train their machine by TheRaven64 · · Score: 3, Interesting

      I wonder if that's actually true. On the one hand, they get a lot of training data for free. On the other hand, how well are the protected against adversarial training? If a botnet signs up for a million accounts, submits 20 photos with correct descriptions for 19 and incorrect ones for the remaining one, is it possible for the attacker to teach the Google algorithm some nonsense things? What if I'm more subtle and upload 100 pictures of dogs and tag them as dogs, but the pictures are carefully crafted to share more characteristics with other pictures of cars than of dogs? Can I train it to recognise all cars as dogs (last year researchers were able to alter a picture of a dog or a car by one pixel and have this system recognise it as the other). A system like this is only as good as its training set - remember that this is the same system that decided that all black people were gorillas because the only black face in its original training set of photos from Google employees was a gorilla at a zoo.

      --
      I am TheRaven on Soylent News
    2. Re:They want the crowd to train their machine by Anonymous Coward · · Score: 0

      Nope, you are testing and helping their A.I. to learn using your data, at your expense, and THEY (Google) get all the benefits.

      Anytime something is free, you are not the customer, you are the product.

    3. Re:They want the crowd to train their machine by 140Mandak262Jamuna · · Score: 1
      Adversarial training is not limited to Artificial Intelligence. Natural Stupidity also suffers from the same fundamental flaw.

      How else can you explain political parties training their flock to vote against their own self interest?

      --
      sed -e 's/Chuck Norris/Rajnikant/g' joke > fact
    4. Re:They want the crowd to train their machine by swillden · · Score: 1

      Nope, you are testing and helping their A.I. to learn using your data, at your expense, and THEY (Google) get all the benefits.

      Even if they get some benefit, it's silly to say that they get all the benefits. I mean, if you do this it's because you'd like to have an AI trained to so some image classification job, and it's easier for you to use Google's service than to build and train your own. So it must be fulfilling some need of yours, and doing it more cheaply/easily than you could do it yourself. That's value.

      --
      Note to ACs: I usually delete AC replies without reading them. If you want to talk to me, log in.
    5. Re:They want the crowd to train their machine by deodiaus2 · · Score: 1

      When I started out as a programmer, you never wanted to be anyone's beta tester. Now, people give up their free time to do so!

    6. Re:They want the crowd to train their machine by TheConway · · Score: 1

      Free to Google, which is what he probably meant.

  2. how long till this one gets shut down by Anonymous Coward · · Score: 0

    another google public api experiment.... wouldn't rely on it being there for too long..... they do get bored with their projects.....

  3. The thirty fourth rule. by queazocotal · · Score: 2

    'Using a meteorologist as an example for their promotional video was an apt choice by Google -- not many people have thousands of tagged HD images bundled together and ready to upload.'

    I suspect quite a few of us have tagged HD images bundled together and ready to upload.

    1. Re:The thirty fourth rule. by PolygamousRanchKid+ · · Score: 1

      I suspect quite a few of us have tagged HD images bundled together and ready to upload.

      My list of of twenty pictures:

      Hillary.
      Donald.
      A can of Cheez Whiz.
      A fractal cow: http://mndl.hu/2008-02-01-frac...
      A Bitcoin.
      A Jai Alai Cesta.
      John Small Berries.
      A build break.
      An iPhone battery.
      Bacon.
      Bigfoot.
      Queen Elizabeth's Crown Jewels.
      An ingrown toenail.
      Twenty years to life, with no chance of parole.
      "The Economist" international Big Mac index.
      A Nobel Peace Prize.
      An Ig Nobel.
      Winter Storm f "Friederike".
      A Ford F-450 Super Duty Limited.
      Slashdot.

      Ok, Google AI . . . get at it . . .

      --
      Schroedinger's Brexit: The UK is both in and out of the EU at the same time!
    2. Re:The thirty fourth rule. by Anonymous Coward · · Score: 0

      I suspect quite a few of us have tagged HD images bundled together and ready to upload.

      Somewhere, there is an AI curled up in the corner going "no no no no no no no ... make it stop".

      One shudders at the thought.

    3. Re:The thirty fourth rule. by Anonymous Coward · · Score: 0

      Try this one instead.

      https://ixquick-proxy.com/do/spg/show_picture.pl?l=english&rais=1&oiu=http%3A%2F%2F3.bp.blogspot.com%2F-TlfBCUUNsA4%2FUQGTDgZiEfI%2FAAAAAAAAGRc%2FMCNnzL-zsRg%2Fs1600%2Fiborg384.jpg&sp=5361c0b85aff5f3adfa409fbee0a07f1

    4. Re:The thirty fourth rule. by Nutria · · Score: 1

      Mine pics are of 20 different gorillas.

      --
      "I don't know, therefore Aliens" Wafflebox1
    5. Re:The thirty fourth rule. by Anonymous Coward · · Score: 0

      Maybe now we can really get that Penis/NotPenis app from Silicon Valley started.

      Now all I need is a hugh dataset of dick pics.

      Anyone know a 13 year old girl?

  4. For now. by Anonymous Coward · · Score: 0

    As always, Google's service is free... for now.

    1. Re:For now. by Anonymous Coward · · Score: 0

      When has Google had a service that was free, then later charged for it?

    2. Re:For now. by sTERNKERN · · Score: 1

      No, it is not free. Just as their search engine is not free (where you pay with your own personal data) this time it is your time, effort, data you provide.

    3. Re:For now. by Anonymous Coward · · Score: 0

      If I upload a lot of randomly labeled images, can I screw up their devilish plans to keep this service unfree?

    4. Re:For now. by asylumx · · Score: 2

      I believe he meant it's free for *Google* -- they won't have to pay people to feed it data.

  5. Sweet! Porn actress! by Anonymous Coward · · Score: 0

    There's this hard bodied porn actress from the 90s and I'm trying to find her name. She's hardly named on porn sites and the one time she was, searching on the name only brought up that video. And the name was unmemorable.

    1. Re:Sweet! Porn actress! by Anonymous Coward · · Score: 0

      I checked your history - it's Gauge.

  6. double take by Froze · · Score: 1

    Did anyone else read that title where the second capital T was an F?

    --
    -- The morphemes of your disquisition are ascertainable, but they have eschewed an ambit of transpicuous exposition.
    1. Re:double take by Anonymous Coward · · Score: 0

      Yeah I thought it was thinking about Tor too.

  7. When? by Anonymous Coward · · Score: 0

    December 6, 2012

    Existing customers can keep using it for free (so far) but if you need to grow beyond 10 users you must pay. And if you ever do pay, beyond a brief trial period, you can never go back to free. When you stop paying, they don't downgrade you to the free tier again, they delete your account.

    1. Re:When? by Anonymous Coward · · Score: 0

      Fake news... from your own link:

      "Try G Suite for free for 14 days. Then, set up billing to continue using the additional features. Or downgrade back to the legacy free edition."

      So... You can downgrade back to the legacy free edition.

    2. Re:When? by Anonymous Coward · · Score: 0

      You seem to be experiencing a reading comprehension issue. I said "if you ever do pay, beyond a brief trial period, you can never go back to free."

      Admittedly their statements are worded a bit awkwardly on that page, but when you actually start the trial, they make it quite clear. You can either continue to pay or downgrade within the trial period. Once the trial period ends, you have no choice but to continue payments or terminate your service entire. You are not allowed to downgrade after the trial.

    3. Re:When? by Anonymous Coward · · Score: 0

      Once the trial period ends, you have no choice but to continue payments or terminate your service entire. You are not allowed to downgrade after the trial.

      Your link says the exact opposite. I think you're confused.

    4. Re:When? by Anonymous Coward · · Score: 0

      When you actually click through, you'll see this page says: Before your trial ends, you must set up billing to keep using the paid version of G Suite. Or, downgrade to go back to using the legacy free version of Google Apps."

      Also you'll see it somewhat more clearly worded in the steps listed "(Required) Before your free trial ends, set up billing or downgrade."

      Maybe you'd prefer even more clarity from their help pages: "You can only return to the legacy free edition of Google Apps during your free G Suite trial. You can’t downgrade back to the legacy free edition if *Your free G Suite trial expires."

      So, no, this is not "fake news" and no, I'm not "confused" at all. Not only is this quite clear in the documentation, I've actually attempted to go through this process and learned the hard way (this was before some of the clarifications were made).

      After the trial period, you can no longer downgrade back to the free version.

  8. Is this actually AI? by Known+Nutter · · Score: 1

    Serious question... is this really Artificial Intelligence? Or is this really Self-Adapting Algorithm? For that matter, is "Self-Adapting Algorithm" what "Artificial Intelligence" actually is?

    "AI" is such a hyped up, overused term that I just can't tell what's what anymore. I'm old.

    --
    Beware of the Leopard.
    1. Re:Is this actually AI? by Anonymous Coward · · Score: 1

      "AI" is whatever we don't know how to get a computer to do.

      Once we know how to get a computer to do it, we give it a different label, and it becomes "no longer AI".

    2. Re: Is this actually AI? by Anonymous Coward · · Score: 0

      Not in the sense you are imagining, no. It's just an algorithm. It's basic functionality is processing data, just like any other algorithm. It can only do so much without input. It is absolutely more hype. I'll tell you a secret: 'AI' is a misnomer used to describe software. True 'AI' does not exist, and probably never will. At core, people do not think and act in binary logic even if they appear to be at times, and software will always and forever be dependent upon binary logic.

  9. That's nice by Anonymous Coward · · Score: 0

    But it isn't any different from releasing other kinds of algorithms over the decades. I'm sorry kids: AI is not magic, it is not different. In some cases, particularly with image editing, it's also not as good. Live longer, millennials, and you'll understand.

  10. Give me a break by Anonymous Coward · · Score: 0

    It isn't 'thinking'. That's part of the problem with modern education and why it fails - thinking is more than just parsing and organizing data. Sigh. Ridiculous. Automation does not make it 'intelligent'. At the end of this cycle, when software is not the magic wand they were promised it was, investors are going to be seriously pissed.

  11. Over Kill by Anonymous Coward · · Score: 0

    I just need something to detect duplicates in my porn collection.

  12. Gorilla by Anonymous Coward · · Score: 0

    Gorilla. It thinks my baby is a gorilla?!?

  13. Cloud OpenML as an Encrypted Communication Tool by PeteJanda · · Score: 1

    Admittedly, this concern may be a bridge too far even for the tinfoil hat crowd. But...

    If I were a bad guy, knew that intelligence agencies have compromised electronics down to the firmware and hardware levels and needed to securely communicate with other bad guys, then I'd develop image + label data to train Google's service to spit out plaintext results from certain image sets. My compatriots would run images of dogs, cats, etc. through OpenML and receive labels like "Bomb" "Building" "Corner" "Columbus" "Central Park".

    Good luck to the good guys when trying to pick up such e-mail traffic.

    1. Re:Cloud OpenML as an Encrypted Communication Tool by CustomSolvers2 · · Score: 1

      Although theoretically possible, the scenario you propose seems too complex to ever become a reality. It would be much easier to create a picture-based encryption system; in fact, I developed one of those some time ago.

      --
      Custom Solvers 2.0 = Alvaro Carballo Garcia = varocarbas.
    2. Re:Cloud OpenML as an Encrypted Communication Tool by PeteJanda · · Score: 1

      A picture-based system is only as strong as the protection of its cipher, which can be compromised in its storage medium (i.e., somewhere, somehow there's a key saying "dog" equals "bomb"). With OpenML, which is meant to be a turnkey solution for mitigating complexity, the cipher is the hidden layers. The interesting aspect here is that the state of the art in AI can't fully explain how those hidden layers arrive at their probabilistic results. Hence a potentially strong system.

    3. Re:Cloud OpenML as an Encrypted Communication Tool by CustomSolvers2 · · Score: 1

      You are making a pretty good point, but practically speaking the requirements for implementing both alternatives seem still way off. Also bear in mind that images contain orders of magnitude more information than plain text, what means excellent conditions to create very strong encryptions.

      The system you propose could be way harder to crack, but also to train (not just making sure that the algorithm behaves as you want, but also hiding any suspicious activity from Google). The strongest point of image-based encryption, at least when properly done, is the difficulty to even know whether there is encrypted information there at all. That's why I think that, if you are able to set up a proper image-based communication, the strength of the encryption algorithm wouldn't be one of your main concerns.

      --
      Custom Solvers 2.0 = Alvaro Carballo Garcia = varocarbas.
  14. Did google ever fix the gorilla problem? by kalieaire · · Score: 1

    Did google ever fix the gorilla problem?

    http://www.dailymail.co.uk/sciencetech/article-5270891/Google-bans-word-gorilla-racist-Photos-app.html

  15. If you tap into the Google by Dirk+Becher · · Score: 1

    the Google also taps into you.