I think most people are writing the rabbit r1 by Teenage Engineering as "another AI no one asked for" that might be true... HOWEVER LAM "Large Action Model" sounds interesting. An assistant to be able to use apps, not use APIs... literally use the apps by clicking buttons & such. But, the reason it sounds interesting is that fleeting hope it might, at long last, help me to avoid the little UI bugs/failings that drive me nuts in many apps.
Which begs the question: why not just fix the UIs? 1/
And really "just fix your UI" ... and lets go beyond "fix" ... because what I really mean is the difference between programs that manage to simply work-- and those rare programs designed thoughtfully, that really save you time, that really make life better.
I don't think throwing AI at poor UI will ever be a real solution. But I guess we need to watch as AI is thrown at every long standing problem for the next two years.
Thinking about LAM I also had to wonder, again, why is it a gadget? 2/
@futurebird ...but my reaction to this was that it seems like a lot of excessive computing power thrown at something that shouldn't be that hard a problem, for a variety of reasons--there are so much more efficient ways to do this. You could have real reliable API support for the third-party service, you could have a simple bot that is not some monster AI system... this is just terrible overkill for what sounds like not great results.
The same issue that leads to UI problems leads to badly designed inconsistent APIs.
Personally I think there should be some minimal API threshold all apps need to meet... as part of privacy regulations I dream about ... the kind of privacy regulations that would make silicone valley VC types cry and thrash on the ground.
I should always be able to ask any app "what are you doing?" or "stop" or "who are you sending that data to?" etc.
Consistent, effective APIs aren't just for interoperability, they are also key to real accessibility.
Accessibility is the difference between apps that are fun toys for some people some of the time... and apps that can really be integrated into people's lives.
These companies want us integrate, but they don't want to wipe their feet before coming in the house.
They won't treat our data with circumspection, won't play nice with other apps, won't explain what they are doing.
@futurebird@mattmcirvin that would be a huge, expensive rearchitrcture of the cloud and Internet infrastructure, but would ultimately be good for security too. Much easier to say "sorry Mx Hacker, I can't send that data to your server" when you have a clearly defined authorization of which data can go to which servers and enforce that at multiple layers.
In a way the practical need for some security with things like payments and logins has forced "the industry" to self adopt a few notions of standards around logins and authentication. They had to because otherwise they don't get paid.
But there could be more to that. And apps that can't meet the threshold are probably not apps that we want floating around and using people's data.
@mattmcirvin@futurebird Yeah, but we live in a world where those APIs don't exist, because the companies that make the software either don't care, or actively don't want to make their software available that way. It's an adversarial situation.
I wonder what the near-future AIs will bring us in efficiency. The locally-runnable ones don't seem to bad to me, and Apple in particular seems to be interested in squeezing these things onto phones.
Why is this another little object I need to have? This will sound hypocritical since I often complain about how everything can't just be another app on the phone.
Purpose-built devices have a place, they are a kind of material language, their buttons, their form factors help us to interact with the dataspace.
There is a reason why mechanical keyboards are resurgent in the age of the "universal, single button, touch-screen rounded-rectangle device" 3/
So, after reading a few reviews of the $200 little oddity, I still have the same takeaway... I wish it were a music player, oooh made of heavy metal & make that wheel more like a fidget toy.
There is something to be said for a device that's fun to hold. That does something simple, that isn't in my personal business.
The LAM idea seems like it could just be another app.
But any device with buttons and knobs gets me a little excited. There must be a better excuse to have them. 4/4
Edit: my hope-- just as much 'work' went into the boot loader and someone hacks together a cloud music player or something more useful out of the hardware.
@Cefr@hazelnot Did these little sneaks just make a different app for the affordable music player they had in the works and try to hop on the bandwagon? I want to see the PCB bet it has room…a gap even… for the headphone jack they quietly removed LOL
@futurebird@hazelnot That was my thinking, Teenage Engineering probably had this slated as a music toy of some sort that never made it to production. They just shed some buttons and knobs for Rabbit R1.
@Cefr@futurebird@hazelnot What else was it going to be? OSs are millions of hours of work / billions of dollars worth. I don't get why people this is in a reveal?
@futurebird@Cefr@hazelnot Yeah, seriously, this is dirt cheap. I think people have got way out of wack in their expectations of what sort of hardware you can get at that price. This is a cheap, minimal device to get their server-based AI stuff out from behind a launch-the-app friction.
If it worked it would be pretty cool. It won't work though. The tech isn't there yet.
@hazelnot@Cefr@futurebird Well, the things it actually does are mostly server-based, so I'm not sure what the problem is? I mean, what's the next shocking reveal, their servers run Linux?
Don't get me wrong, this thing won't work because AI isn't good enough to do what they say it will do. To me, that's the bit that galling, not calling their not-entirely-original software stack an ‘OS’.
Add comment