Just come across this, thought it might make an interesting topic for discussion as it's been a bit quiet around here...

http://www.myjibo.com

I'm trying to work out if I think it's cool or if I think my iPhone could do a lot of that stuff....

Cheers

Cris

I don't think I really 'get it'. It sounds like a mobile phone for the most part.

It mentioned that it makes seniors living alone more independent. How? Is that just by being a voice activated phone?

It also says that it can proactively help with everyday tasks. What do they mean here? If Jibo can help with the dishes or laundry, sign me up. I don't see how it can though.

Of course, I'm the curmudgeon that doesn't get the fascination with smart phones either

The more I think about it the worse the idea gets.

Unless there is a Jibo in every room of the house, or he could move between rooms. I don't see the point.

Can you make a cool video and slap it on kickstarter for just about any idea ???

Cheers

Cris

I assumed it was mobile. If not, yikes.

The fact you have to carry it around would be a pain and then it requires charging probably quite frequently too.

As for the comparison with phones, then yes apart from the movement there's not that much more it can do.

The pitch was good though.

The video conference aspect has me rather interested. If they can keep the cost in that realm, I could see a lot of people using it for that aspect.

Beyond that, I see it mostly as an example of where Siri, Google Now and Cortana can continue to grow and evolve.

Sounds like a very expensive electronic cork board for pinning family messages to. Personally, I prefer SMS for that sort of thing.

Agreed. There's not a lot of utility in this thing, considering it costs so much. I don't know why you'd get this in addition to a smartphone. For the price, you could buy two Nexus 7 tablets.

And how does the family share the Jibo? That little girl has it in her blanket fort, so now grandma doesn't know that whatshername is picking her up to go grocery shopping. Are we supposed to buy multiple Jibos for the home? There's countless things I'd rather do with a thousand dollars.

BTW, I can't help but imagine a different ending to that video, where the Jibo wishes the girl goodnight, then the heart turns to an angry looking red dot as Jibo plots its revenge on the family that's enslaved it.

That triggered something which made me finally understand why I felt uncomfortable about the Jibo when I first skimmed its web page yesterday. Now I can put my finger on the issue:

Voice-interactive computer systems are difficult to get working correctly. All of the existing systems like Siri/Cortana have severe problems which they're working to correct, a little at a time. Issues with mistakes in the actual speech recognition, the inability to perform all of the actions that you might want or expect them to do, and especially the inability to glean the context of the conversation, are all still sticking points which are going to take a lot of time, research, retry, and coding work to get around. Right now, that limits what I can do with those systems. Right now, I use Siri pretty much only for setting alarms, because that's one of the few things where it's easier and quicker with Siri. And even then she gets it wrong sometimes!

This is a hard problem to solve, and the people working on Siri, Cortana, Google Now, and Jibo, are all working separately, in competition with each other, rather than pooling their knowledge. If they were working together, we could advance the state of the art in speech recognition much faster. But even if they were working together, it's still a hard set of problems to solve and it will take time to get things right. The fact that we've got these things on our phones now, and we're slowly easing into being able to sometimes use these things, is helping them advance the state of the art a little at a time. The only reason I'm using Siri right now is because she came with my phone, not because I set out to "buy Siri". I wanted the phone for other reasons, Siri is just an extra feature. One with a very narrow range of usefulness that they're working to expand, a little at a time.

So my particular problem with Jibo is the price tag: It feels like they're promising that the extra expense will result in better voice recognition. As if somehow the higher price tag magically makes Jibo's voice recognition better than Siri. But that's not how it works.

I'm sure that's not what Jibo's makers intended. I'm sure they are a hard working team who wants nothing more than to deliver the most amazing voice-interactive experience the world has ever seen. They want to create something awesome, and they have their own vision which is different from Apple's or Microsoft's. My point here is just that I had a specific negative reaction to seeing their web page and their price tag, and it was a subtle thing, and it took a while to understand why and be able to put into words. My initial reaction was that I couldn't imagine spending that much money to help them fund their voice recognition research.

I've got nothing against spending extra money for a particular product of course. I bought the quite-expensive empeg because it was, and still is 15 years later, the best possible in-car MP3 player. Though it was expensive, at the time I bought it, it solved a problem that no other product could at the time (all my songs in the car), and it did so in such an advanced and forward-thinking way that today I still prefer its feature set to any current product which can solve the same problem (like... my phone ).

And after writing all of that, I go back to the page and double check the price...

I had originally misread the page on my initial skim and thought the price was over $2000 USD. I think maybe I had misread the number of backers as a price tag or something. Seeing that the price is actually $500-$600 USD is more in line with what I would expect for something like that. Still, pretty expensive for a glorified Furby whose speech recognition isn't going to work nearly as well as their demo video implies.

I'll be interested in seeing some hands on tests of the final shipping product. If it can get past the Kinect's problem of false-activating when people in the room are having normal conversations, it could be a fun development platform.

My main problem with all voice recognition programs is that I still feel weird talking to my device with anyone else around. Even if it's a simple search, it seems strange to speak into my phone like that. Maybe that'll become a thing of the past, though.

That said, have you tried Google Now, Tony? I certainly use it for way more than alarms (which I do too). It's quite good at understanding things, and even has a mechanism to correct misheard words through voice alone.

Back on topic: I also don't understand the use of this device as a camera. So all your photos will be taken from table height?

Yes, other members of my household have Google Now on their Android phones, and I see it as being pretty much equal to Siri in a very generalized sort of sense. There are minor differences and advantages, but the big picture of context is still something they haven't succeeded at tackling on any of these platforms.

I remember years ago, when speech recognition was in its infancy, an example of a conversation that would be understandable to a human being, but impossible for speech recognition in terms of contextual understanding.

She: Hungry?
He: I've got a McDonalds coupon.
She: The keys are on the table.

I think it will be quite a while before the joining of AI and speech recognition reaches the stage where a computer can understand something like that.

tanstaafl.

Agreed, that's an example of something I wouldn't expect any time soon.

But context problems still get in the way of even using simple voice commands on current devices to do the things they're advertised as being able to do. Such as take short dictation messages for sending texts. This is something they are touted as being able to do. The ability to understand simple things like the difference between when I want to spell out a word or a special character, versus when I want it to try to voice-recognize the words I'm trying to speak, is still problematic. I had a devil of a time trying to dictate a carriage return recently, for example.

Almost all encounters I have with voice recognition end up like Teri Garr in that TOS episode with the typewriter. As much as I wish computer voice recognition were magic, it's still got those same old problems.

I still find it weird and slightly pointless talking to my phone. I never ever use Siri, just terrible for anything useful.

Not only am I left feeling pretty stupid but I don't actually get what I want.

My normal problem is in the car, where I suspect the back ground noise makes this task difficult, a command like "Play Dire Straits" will end up with some random song playing and never ever anything by Dire Straits.

With this device I assume it will be a mix of having to carry the damn thing around with you and plugging it in, and wanting to drown it in the kitchen sink out of utter frustration! I think the cute animations it plays will only fool me for a very short time!

Cheers

Cris

I use Siri regularly, but only for things like this:

- "timer, 10 minutes"
- "remind me tomorrow at 10 to pick up blah"
- "where is Eryl"
- "wake me at 9"

It works very well for me for those sort of things, it is hopeless in the car. It needs the fast speech recognition that Google have to make it more responsive (there are some improvements in iOS 8, but still not as good as Google in that respect).

OK, fair enough. I haven't used Siri myself so I can't really compare. I've also had the same annoyances you have. I often run into that issue of not being able to do a carriage return. I do feel like Google Now is at least making strides towards the goals you're laying out.

There have been a few instances where I really have been able to use a little conversational language with it. For example, there was an instance where we were on vacation in Bethany Beach, DE, and we were concerned about our basement flooding (we've had issues with that when it rains really hard). It was pretty cool to be able to have a conversation like this:

Me: Will it rain in Sterling, Virginia today?
GN: No, it's 85 degrees and sunny in Sterling, Virginia.
Me: What about tomorrow?
GN: Yes, it will be cloudy with a chance of thunderstorms
Me: How about Bethany Beach, Delaware?
GN: No, it will be sunny and 88 degrees in Bethany Beach, DE tomorrow.

While it's important to fix the mistakes that these voice recognition services make**, I think the real hurdle is making it understand when we make mistakes. At the moment, it's still essential to formulate exactly what you're going to say when you initiate voice recognition on any of these platforms. You can't stammer through a word or start a word over. It's very strict.


** I said "Halicarnassus" into my phone last night and it thought I said "how much are nurses"

I'm a pretty heavy user of Siri, and did get the 4s specifically for it. Mostly for reminders, but it's also very useful when riding a motorcycle (directions/text messaging). All 3 companies are making some good advances in the voice recognition area, including the context aspect. Android L is gaining the exact reminder features that had me buy a 4s, and Cortana has added some nice features in that area.

Though the Jibo reminded me more of the other aspects of a personal assistant, distilling a flood of data down to the important and relevant parts. I see Google Now ahead here currently, though Microsoft has made some interesting advances with Cortana, and Apple is starting to close the gap with changes in iOS 7 and 8.

There is a creepy feeling aspect that comes from these features though. Tony and I used to go to a particular lunch spot frequently. My iPhone started showing traffic info there automatically around lunch time every Monday. The creepiness is starting to wear off for me, and I'm curious to see where this goes in the future. Especially for families.

Two problems I saw:

1) The voice recognition seems to be overly contrived, even moreso than usual technology demonstrations. I'm sure anybody who views the takeaway example with a critical mind can see how that would be incredibly hard to do for real.


2) Following on from that, are they promising a product that can do that out-of-the-box or is it an example of what a developer could do with the hardware? Will they will claim the "hard stuff" like the takeaway scene isn't actually their responsibility and a takeaway app developer needs to do all of that, they will just let them listen for certain keywords or something.

Better context understanding, and self learning to gain personal assistance capabilities coming in the next generation:

http://www.wired.com/2014/08/viv/

Speaks a bit about where current systems are and how they work, and where Viv advances the current generation.

If it gets hacked, things get real creepy real fast

I did think that actually. Imagine if it could actually move around your house too!

Cheers

Cris

See attached picture of the conversation I had yesterday. Note she gets the answer wrong, despite seeming to understand. I have no idea what she thought I was referring to when I said "that". She looked like she knew, but she didn't.