381–390 of 594 posts
What I'd like to see is some good open source software that can compete with Google Assistant, Siri, and Alexa. It looks like there are some promising projects, but nothing turn-key yet. I'd like to be able to simply apt-get a package, and have voice recognition on my box.
The hard parts of the Assistant, Alexa, etc., are basically everything else.
But we've seen that Google is happy to turn over massive amounts of customer data to government without a warrant and without alerting customers to the practice, which makes the technology seem ominous.
First the GPS, then the microphone, then the camera, accelerometers, 3D touch sensors, etc. Gait, affect, and all sorts of factors will be able to predict criminal behavior before it happens.
Let's hope the next generation of tech giants will take customer privacy and freedom seriously and avoid the dark patterns and privacy violations of the current era.
Only now, when it's likely too late, can we actually get a glimpse of the sort of Orwellian dystopia that so many have warned about in decades past.
Earlier quoted context omitted.
This right here. When I first heard about android, I imagined an open system I could tinker with like I can on my PC. Instead I get the current evolution. I want a 3rd party. The same thing comes to my every day of usage, I'm still on win 7, and short of leaving Linux(yes I should have already), I can't upgrade without becoming a product. I want what we all imagined and dreamt of, and I pay for it. What I won't do, i…
Learning* not leaving Linux. Too late to edit. Damn you fat thumbs.
It's not that complicated. Use a common distro like Ubuntu and everything's documented online.
Earlier quoted context omitted.
I fear you will be unable to recognize when that burger was your choice and when it was a reaction. You probably won't notice. And that is harmless. I also fear you will be unable to notice in which areas of life and information the distinction between choice and reaction is harmless and which it isn't. Of course, I'm not talking about "You" you, but just people. Me as well. I feel we are widening the field of uncons…
I'm sorry but that just sounds like blind fear mongering. What you're saying is vague and doesn't really mean much. It's like saying we shouldn't use prescription glasses, or medication, or cars, because it's not "us". Humans invent all these tools and systems to improve and optimize our life. Make our vision better. Make our health better. Make us move around faster. In the case of AI, make us do perform certain thi…
In your metaphor, you are implicitly paying the secretary, so the secretary is incentivized to maintain your interests.
How much have you paid Google for its free services?
Your metaphor is inapplicable. You don't have a secretary telling you these things; you have a salesman trying to sell you things, and the salesman is getting smarter every day while you aren't. Not the same thing at all.
Earlier quoted context omitted.
One tiny caveat being that, Google (and others as well, to be fair) will be able to indirectly collect data even on the privacy-aware part of the population who don't use Google much. The simplest example being, even if you don't use GMail, still some part of your emails inevitably end up in GMail inboxes. Now also consider this: just being a guest in a Google-stuffed house means you are under surveillance. So no, it…
Sure, and to illustrate your point, I have an email very similar to someone else's. I very frequently get their emails (invoices, church events, travel itineraries, purchase receipts). Google thinks they're _my_ trips, and updates me about flight times. I think this example serves both our points. To your point, it's totally leaked this other person's info into my "google world" because I'm on gmail. On the other han…
Fun story there. Because Google's internal privacy safeguards are so strict, the people working on features like that can't look for example emails to train their ML models with.
They can only look at emails that were explicitly sent to them in order to improve the feature (and almost no one forwards along positive nor negative examples). What they can do across the email corpus is run jobs that return aggregate stats, where each stat must be coarse enough that it is infeasible to trace back to original users (often 100k+ users per data point).
So, AFAIK, training & testing models under these safeguards is more or less done blind. Build a model with the few examples you do have, and then run it against the corpus. If you see numbers change, you have no idea if that's good or bad, since you can't actually inspect the run.
(at least, this is the way it was a few years ago)
Earlier quoted context omitted.
There are all kinds of interesting thought experiments. What happens when a classifier innocently discovers that the best classification is by race? Do we care? How about if we remove race but it happens to discover that four features which are very strongly correlated to race are the best way to classify?
There are ways to account for that. A model can be fit to race, and then you only predict "on top" of race (meaning residuals). You use that model, which is independent of race.
Machine learning is also very opaque.
What I think is interesting is that many of us nerds have probably innocuously fantasized about having a Star Trek-like AI assistant with us, but now that they're taking the first steps towards that, we're starting to realize that in order for it to do everything for us, it has to know everything about us, too.
But does it? Does it have to know your birthday? (leave alone the fact that bdays are somehow part of a superkey for your identity).
Why should it know my residence, my spouse, or my CC# (with Apple's TouchID maybe it won't need to)?
Google's concept of AI is too creepy for me. It can be useful without being creepy. They're not even trying to make it less creepy.
Overlay on this the subtext that NSA and other tla's are monitoring all this (leave alone other countries). While I may trust Google, I don't trust them to not be forced to collude with the government.
Google Now has done the same for me, told me how long it will take to get to a bar I frequent. My reaction was quite exactly "Oh that's neat, thanks!" and I went and had a great burger that night. Totally OK for me Google. I respect that people have different privacy thresholds, but I think the fact that it's different for everyone is being lost in articles like this.
I met a good friend and current employer because Google Now told me about an event I might be interested in a couple of years ago. Of course it also once told me how long it would take to get to an ex's house from my current girlfriend's place. How about an 'OK, funny once' command?
(eugenics through large scale suggestions anyone? ;>)