This Bear Can Pass A Turing Test

bear

Some thought the first artificial intelligence would come about as an accident, others as a war machine that decides the only way to protect humans is to kill them all. It turns out both these ideas were wrong. The first AI is apparently a teddy bear, available on Kickstarter for $60.

The Supertoy Kickstarter is selling a mechatronic teddy bear with motors, speakers, and enough electronics to connect to a cell phone. After plugging your cell phone and stuffing it in Teddy’s thorax, the bear comes alive with an intelligence all his own and a voice seemingly lifted from [Peter Griffin].

Needless to say, we’re just a bit skeptical that Teddy here can perform as demonstrated in the Kickstarter video. While the team behind Teddy has developed a successful talking chatbot before, the video makes this tech seem too good. Even the voice sounds like a real person with a microphone, and not like a clunky GPS personality.

Feel free to speculate in the comments on how good this tech can possibly be.

59 thoughts on “This Bear Can Pass A Turing Test

  1. Blimey, faker than a fake thing. Extraordinarily misleading.

    They could at least have ATTEMPTED to edit it convincingly!

    Little girl asks question with aircraft in background.
    Ted, who is in his words, answering questions in real time without any remote control and the little girl “can vouch for that”, answers, in perfect english, without any aircraft in background.
    Cut back to little girl asking another question, aircraft in background.

    1. If that’s not convincing you can also look at her sleeves. Whenever she asks a question in the 1:50s her sleeves are down. However when Teddy answers her sleeves are up.

      1. It gets worse, the bear alternates which side of the chair he is sitting in and the change in shadows leads me to believe they have been shooting this “interview” for hours.
        And wow the animatronics are bad, the original Furby had far more convincing movement.

        1. You… you… guys realize that they probably had only 1 camera and redid the whole thing to get both angles? But yes, regardless I feel it is fake – they are either using finely crafted voice samples for many exact questions, but when you go beyond the scope of them, it either starts using a robotic voice or just craps out entirely. OR, they could be just faking the whole thing, and using kickstarter to fund the ‘development’ of such a realistic voice package.

    2. Not to mention that the camera angle from which Teddy is seen is directly over her shoulder and close. Obviously, you would see that camera while the little girl is talking, if it were in real time.

  2. “….and a voice seemingly lifted from [Peter Griffin]”

    Close. It’s a ringer for the Movie “Ted” w/ Mark Wahlberg. (same producer)

    Way too close of a copy for that to be an accident.

    [youtube=http://www.youtube.com/watch?v=9fbo_pQvU7M&w=560&h=315]

  3. This really seems like an obvious fake. Apart from the mentioned flaw in the video, I’ve never heard of any kind of speech synthesis being that good!
    But then again, imagine this isn’t fake. An innocent toy connected 24/7 to an online data warehouse collecting just about every bit of information about us. How we live, what we do, what we like and what we dislike. Everyone would want to have the first true AI in their homes, especially children would want this teddy bear. Imagine the market value of the data this little toy could collect on us, the consumers.

    Besides, the mainframe can still decide to go skynet and send them teddies to kill us.

  4. The view from either side places the camera slightly behind and to the side of both the teddy and the girl, however the camera’s are never in the frame.

    Filmed separately at least. Bear is probably fake as well.

  5. It wouldn’t be on Kickstarter if it was already a viable product – it’s obviously going to be a mockup. As for the AI, isn’t this just Siri tweaked to be more kid friendly? Hardly a first.

    1. They don’t have room in their planning for R&D, they will start production straight after it gets funded … either the speech recognition, synthesis and animatronics are developed to the level of the video already or they won’t be by the time it ships. My guess, the engineers got in bed with some shady types to market/sell this and they are now being taken for a ride to scam city.

      If any of the engineers behind this are reading, you should really speak up, whether the video is accurate or inaccurate … really at this point we can only make a guess that it’s inaccurate and you guys are part of a scam. I’d love to be proven wrong and I’d be more willing to reserve judgement with some engineer presented info on the underlying technology rather than this edited sales video.

  6. They link to a Youtube video of its innards in the FAQ (under “is this some kind of black magic??”) That video definitely has a different, more typically computer generated voice and a much lower ‘chat bot’ level conversational skill.

  7. Speaking as someone whose degree and specialization is in Computational Linguistics, this toy looks to be EXCEPTIONALLY fake. I’m on the cutting edge of speech generation and “chatbots” neither of which are near this level of sophistication. Not saying that it’s impossible to create any sort of product like this, just that what is shown in the video is definitely infeasible. I’m also extremely suspicious of their promise of seamless use of “over 30 languages”. This project, if true, would be 30-40 years ahead of existing technologies. You rarely, if ever, see a jump like that especially without huge investments.

      1. “No AI”? Since when? But you are right in that humans fail the Turing test far more than you would expect:
        http://www.geekosystem.com/cleverbot-passes-turing-test/
        “It seems that Cleverbot, the chatbot so ready to admit that it was a unicorn during a discussion with itself, has passed the Turing test. This past Sunday, the 1334 votes from a Turing test held at the Techniche festival in Guwahati, India were released. They revealed that Cleverbot was voted to be human 59.3% of the time. Real humans did only slightly better and were assumed to be humans 63.3% of the time. That being the case, Cleverbot’s success in conning people into thinking it was human is greater than chance, and therefore, one could argue that it has technically passed the Turning test.

        Of course, that’s only one way to look at the results. Although Cleverbot may have been able to convince a majority of people that it was a human, as bizarre as that may sound, it still comes short of actual humans. 59% is also not that much greater than chance. Still, when you consider that actual humans are only suspected to be human 63.3% of the time, there’s not much of a gap for Cleverbot to close.”

    1. The fact is, you run your kickstarter video in Wizard of Oz mode, making a complete mockery of the robotics field.

      The text to speech engine does not exist in the way you advertise it, and your chat bot will never meet the expectations you have placed upon it.

      1. I certainly did not want to mock anyone. I have an immense respect and passion for the robotics (hardware) pioneers. I take a different approach, I measure future of robotics by its intelligence and communications skills. Its just a different approach, a software approach, the brain. Nobody is right or wrong because both will come together soon. Like in my all time favorite film Robot and Frank. I am not proposing I have all the answers, I am just making a TOY. Lets get that straight, so in no way do I demean anything the hardware guys are doing. In fact the opposite, I think their heroes. I just want to make a toy talk like a human being. Supertoy is not scripted. I dont know what he/she is going to say for sure. Yes, somethings are repeated like us humans we repeat ourselves! anyway the bit at the end when I said “Bye Now”, Supertoy responded “will that accomplish your objective”, that was totally a surprise! as I expected Supertoy to reply “bye”. As I said Supertoy will interrupt the conversation, its not all one-way. I have another clip to show if anyone wants.

        The voice is real. I have found a way to create a realistic voice, in this case a cartoon voice filled with emotion, how I did that I cant show :-)

        Regarding the A.I. its my own design. Again I cannot share it naturally, maybe a year from now in a book LOL. Seriously, what I am doing is mixing hard and soft a.i. its a kind of hybrid, I stress what I am doing is mimicking human thinking etc.

        All this self funded literally from the kitchen table, I came up with the concept 3 years ago then spent 6 months 24/7 to get to this point.

        1. “The voice is real. I have found a way to create a realistic voice, in this case a cartoon voice filled with emotion, how I did that I cant show :-) ”

          We know the voice from the video on the kickstarter page is real and filled with emotion, human speakers are good at doing that :/

          1. Ashley: So your saying that right now, the voice in the Kickstarter video was Human but later ones will be computer generated? Just trying to get clarification, because you’ve avoided answering any questions directly asking this.

        2. “Nobody is right or wrong because both will come together soon.”

          So what you’re essentially saying is that you don’t have the text-to-speech engine, and you don’t have the chatbot to run it, and you’re still presenting false advertisement with a human-acted or mocked up video to collect money for a product that doesn’t actually exist.

          That’s still called a scam.

          1. “The voice is real. I have found a way to create a realistic voice, in this case a cartoon voice filled with emotion, how I did that I cant show :-) ”

            And I have a bridge to sell. Interested?

          2. “Seriously, what I am doing is mixing hard and soft a.i. its a kind of hybrid”

            Hard or Strong AI refers to human level cognitive ability by mimicking the same congitive processes. It’s one of the unsolved problems in computing, seeing that we don’t yet know how the human brain operates in the first place, or what constitutes as intelligence. As such, Hard AI doensn’t even exist.

            You on the other hand come across as a charlatan.

  8. Won’t say anything bad? “If you want Supertoy Teddy to be your wise guy badass pal its your choice!” Well, I guess the people building it can. ;-)

    There’s an idea, record when kids say obscenities and it alerts the parents by sending a text to their phones.

    1. Not a bad idea. We are applying filters for blacklisted words + content. if a child says: “whenever I say the word noodle you fart” and this slips through our filters we have to apply the First Amendment for robots.

      Also keep in mind that the application is for children of age 8 to 80, so they are probably expecting some South Park like behavior!

  9. Why hasn’t anyone reported to kickstarter that they are running a misleading video? There are going to be a lot of severely disappointed people if the project is allowed to finish.

    From the guidelines page:
    No product simulations or photorealistic renderings

    Technical drawings, CAD designs, sketches, and other parts of the design process are awesome and encouraged. Photorealistic renderings and simulations that could be mistaken for finished products or real events, however, are not allowed.

  10. This is certainly fake. I know some people who work on voice processing and even state of the art systems are not as good as that. In fact, this would be 20 years more advanced than the competition. Such a system would not need a kickstarter, it would be trivial to find enormous amounts of money from companies or academia. Especially since they claim that this is not a mock-up (they state they will immediately go into production) so the technology is ready for prime-time commercial applications.

  11. Kickstarter rules notwithstanding, this falls nicely under the Fraud Act 2006
    (1)A person is in breach of this section if he—
    (a)dishonestly makes a false representation, and
    (b)intends, by making the representation—
    (i)to make a gain for himself or another, or
    (ii)to cause loss to another or to expose another to a risk of loss.
    (2)A representation is false if—
    (a)it is untrue or misleading, and
    (b)the person making it knows that it is, or might be, untrue or misleading.

    1. Fraud — kind of describes Kickstarter in general. Okay, I admit that I may be a bit jaded after being ripped off by a kickstarter — never again will I have anything to do with them. Kickstarter itself is completely unwilling to help once they get your money, you are on your own. Let the buyer beware.

  12. Fake as fake gets, no speech engine exists that can speak numbers in that fashion that I’ve ever come across. You’d have to have a sample for each possible combination of numbers for it to speak without noticeable gaps in speech and changes in tone, good luck with that. I’ve also never heard a speech engine that could maintain the tone of a sentence when using samples, this is just fake. For shame.

  13. The voice is read back by the actual app, driven by matching dialogue patterns. Granted the TTS version of the voice in the Robot video sounds a little bit less phenomenal, however we are capable of modifying emotions in the TTS and we can utter special vocal expressions. The TTS is based on the same voice you hear in the video.

    What we mainly wanted to show is that we do have a working prototype of a toy which can
    * chat with you for hours ( check out Jeannie on Google play), tell you jokes, give you info etc
    * perform many smart actions for you
    * talk back freely

    We had very positive reactions from everyone playing with the prototype, so if there is any danger that those positive reactions turn into negative ones we take it very seriously.

    The voice that you were hearing is being transformed into a fully functional text-to-speech voice, specifically created for our Teddy.

    1. OK Then, let’s see a video of the actual voice synthesis in an unscripted dialogue.
      As noted before, if your voice synthesis is to the level of the promotional video, you can make millions right now, by selling the technology to Satnav manufacturers, automated telephone system integrators, Apple, Samsung, man + dog

  14. I also agree that this seems fake/misleading. There is no way a bear could just sit there and not try to bite that girl at least once. In my experience, they are not patient enough to have a conversation for more than 20 minutes tops.

  15. A pure hoax… I’m an A.I. researcher and I laughed my longs out. If they can do it, Google would have done it years ago. I checked whether the launch date of the project was April the 1st. This cynical Teddy bear is a pure hoax, comparable to “the flying bird man last year” http://www.youtube.com/watch?v=hqojK0moe_E Good joke though, it is amazing how easily people believe things…

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.