Mycroft (The AI, not the Holmes)

[Happy]

Anyone else excited to mess around with this later this year?:

https://mycroft.ai

 

[Reluctantly]

If it gets adopted, it will be interesting to see what people do with them. But I'm not really interested in using it as a house butler.

 

[Happy]

I'm interested in both. Mainly the follow on effect though.

 

[Kuu]

It's really a nice effort, and specially so because of Open Source and Open Hardware. It's way too creepy to have a closed IoT hub that's listening/watching you and you don't have any assurance of it not sending off your entire house data out there to Creeps Inc.

The biggest problem is the other IoT devices. Adoption hasn't been fast and the long-term support of many of them is dubious. I've seen some of those smart lights and power outlets and they were ridiculously expensive. Retrofitting the entire house.... yeah not happening soon. Maybe in two years.

What I'm really waiting for is simple sensors and connected actuators so I can easily automate exterior shading devices or ventilation openings based on temperature, time of day, or illuminance levels for smart-passive environment control...

 

[kora]

In b4 hal/evil mycroft joke.

 

[kora]

"Mycroft, kill my kids."

 

[bvanevery]

A house is not a problem. When I put the effort into technology, I work on what I consider to be real problems. Growing food, distilling water, ice making, wood gasification, casting, CPU fabrication, making games... these are things that come to mind as actual problems, that I wouldn't mind having some help with. But I'm not seeing how Mycroft helps in any way.

If something provided text-to-speech and allowed emotional annotation, so that I could deliver a line like an actor would instead of like a computer does, that would definitely be worth something to me. But that doesn't sound like Mycroft.

As for the open source aspect, I do not believe in contributing to GPLed projects. I would only look at a GPLed project if it pretty much did almost exactly what I wanted already, and I'd have to invest almost no work to bend it to my own purposes. In short, if they had already fully solved the problem. I seriously doubt Mycroft would be any kind of "goto" for actor dialog synthesis so there's no way I'd put in the labor to make it able to do that.

 

[Kuu]

Uhh ok...

Simplifying people's life helps, so people can concentrate on the Real Problems ®.

 

[Happy]

What I'm really waiting for is simple sensors and connected actuators so I can easily automate exterior shading devices or ventilation openings based on temperature, time of day, or illuminance levels for smart-passive environment control...

Shit yeah! I'm excited for these also.

A house is not a problem. When I put the effort into technology, I work on what I consider to be real problems. Growing food, distilling water, ice making, wood gasification, casting, CPU fabrication, making games... these are things that come to mind as actual problems, that I wouldn't mind having some help with. But I'm not seeing how Mycroft helps in any way.

Uhh ok...

Simplifying people's life helps, so people can concentrate on the Real Problems ®.

I'm with Kuu on that one. The environments in which people spend the majority of their time (namely, their home) deserve a great deal of attention. Hell, as INTPs, the less time we have to allot to dealing with Si tasks, the better - more time for thinking about cool shit. Yay, cool shit!

But the immediate gratification of an open source AI butler is not the most pertinent consideration here. It's an AI framework anyone can pick up and screw around with. You can do whatever the hell you want with it. It'll grow over time.

It could even be applied to some of these 'actual problems' you mention. Growing food for instance - the worst part about growing food is the constant watering (seriously, the worst - try growing food in an Australian summer. It's ridiculous...). You could just have your Mycroft left in charge of maintaining the right conditions - the technology (sensors, taps, etc) already exists. A bit of code and your alarm clock is watering your plants. Problem solved!

 

[bvanevery]

Simplifying people's life helps, so people can concentrate on the Real Problems ®.

Gadgets don't simplify people's lives. Gadgets are business models, used by corporations to win The Battle For The Eyeballs, as former Intel CEO Andy Grove once put it. You want to simplify your life? Get rid of your smartphone.

 

[bvanevery]

I'm with Kuu on that one. The environments in which people spend the majority of their time (namely, their home) deserve a great deal of attention.

Homes already work. How do you think everyone survived and reproduced so far? Even huts work. I'm going to laugh at you technophiles when you can't open your garage door because your software has a bug. I've got an old car with manual roll down windows. Sure I exert a little effort to work them, but they are never going to stop working due to electrical failure. I've had a power window stop working in another car and it's a PITA.

Hell, as INTPs, the less time we have to allot to dealing with Si tasks, the better - more time for thinking about cool shit. Yay, cool shit!

That's ridiculous myth spinning if I ever heard it. Right up there with "INTPs suck at Art," someone stated in another thread. Being an INTP doesn't make you a dimwit for how a light switch works.

But the immediate gratification of an open source AI butler is not the most pertinent consideration here. It's an AI framework anyone can pick up and screw around with.

I don't believe in "screwing around with" technology, and I've even managed to make enemies arguing that sort of point before. People in the Maker community often bug me when they do pointless things for very high dollar cost. Haven't they ever heard of solving actual engineering problems, like delivering cheap solutions to the Third World or homeless people or whatever? Instead they waste all their money and brainpower on frivolities.

You can do whatever the hell you want with it. It'll grow over time.

Show me a company with a real problem and a real business model or I call bull. $0 open source fiddling is amateur hour. These undisciplined people screw around with stuff until they get bored, which doesn't take very long. Their efforts end up being inconsequential, especially because they never solve hard deployment problems like QA / making sure stuff actually works and is reproducible on other people's systems.

It could even be applied to some of these 'actual problems' you mention. Growing food for instance - the worst part about growing food is the constant watering (seriously, the worst - try growing food in an Australian summer. It's ridiculous...). You could just have your Mycroft left in charge of maintaining the right conditions - the technology (sensors, taps, etc) already exists. A bit of code and your alarm clock is watering your plants. Problem solved!

Show me the code actually doing that, or I call bull. Embedded systems don't integrate themselves.

 

[Haim]

The thing is that the experience isn't really like this,voice recognising sucks.
More like "*some word* make me a tea"
"did you mean burn your house?"
"no,make me tea!"
"turn on the TV?"
............
"haa make you tea,no tea machines where found"
"what but I just bought that smart tea machine"

after some time....
fuck this,I don't want to talk to myself and hear that stupid AI voice.

The experience needs to be a lot better.
amm mind reader,""I wish for the neighbear to turn down the volume""
*AI go to neighbear and make the more annoying sounds the more the music volume is higher*
now that's convenient.

 

[Happy]

Homes already work. How do you think everyone survived and reproduced so far? Even huts work. I'm going to laugh at you technophiles when you can't open your garage door because your software has a bug. I've got an old car with manual roll down windows. Sure I exert a little effort to work them, but they are never going to stop working due to electrical failure. I've had a power window stop working in another car and it's a PITA.

Homes don't work. They really don't. Well, for the most part, at least - most of them do the basics (shelter, etc), and nothing more, yet continue to become unnecessarily larger and more expensive. I could write you an essay on this, but it'd be a derail, so I won't.

That's ridiculous myth spinning if I ever heard it. Right up there with "INTPs suck at Art," someone stated in another thread. Being an INTP doesn't make you a dimwit for how a light switch works.

Not remotely what I said. I'm not even sure how you made that connection.

It's not a question of competence. It's a matter of outsourcing.

I don't believe in "screwing around with" technology, and I've even managed to make enemies arguing that sort of point before. People in the Maker community often bug me when they do pointless things for very high dollar cost. Haven't they ever heard of solving actual engineering problems, like delivering cheap solutions to the Third World or homeless people or whatever? Instead they waste all their money and brainpower on frivolities.

Show me a company with a real problem and a real business model or I call bull. $0 open source fiddling is amateur hour. These undisciplined people screw around with stuff until they get bored, which doesn't take very long. Their efforts end up being inconsequential, especially because they never solve hard deployment problems like QA / making sure stuff actually works and is reproducible on other people's systems.

Brace yourselves - the fun police are here.

Show me the code actually doing that, or I call bull. Embedded systems don't integrate themselves.

You're demanding proof for a hypothetical scenario?

 

[Cognisant]

As I see it the problem is that it isn't really an AI, it's an interface, telling a glorified alarm clock to operate the functions of an automated house is possible and perhaps even convenient. But a standard remote would work just as well and pushing a button requires less effort than speaking a coherent sentence, especially first thing in the morning.

I think personal assistant devices like this won't really take off until they can do things we don't want to do for ourselves, for example we hate cleaning hence the popularity of Roombas and washing machines, mowing the lawn and cleaning a pool are boring tasks too hence why there's robots to do them. Whereas turning on the coffee machine hardly saves any effort that the coffee machine itself isn't already saving and even then that's debatable, likewise scheduling can be done with a smartphone but we still use calendars on the fridge because a pen is an easier interface.

Sophistication != elegant design.

 

[Architect]

Looks like a clone of the Amazon Echo, which I've used at relatives house. Downside is you don't get the millions of RD that Amazon pour into the echo, not to mention all the data they have. Plus compute power. Other competitors are Google Now , Apple Siri and Microsoft ... watchitcalled, Cortana.

The idea has been dreamed up in sci fi for decades, we all call it the "Star Trek computer". You know ... "Computer, do XYZ" and it knows how to answer appropriately.

The question then are we ready for an open source clone yet? Presumably they're using deep learning, where are the FLOPS coming from? Who is training the nets? At this point these things aren't for free, I dispute their claim that 'using open source libraries' will just work. DNN's are different than run of the mill FOSS, they are a start to a lot of work, data and FLOPS to get them to do anything useful.

I don't think the time is quite right. for an open source AI. After we figure out the data reduction problem, probably then.