Well we've now seen this same story for Amazon, Google, and Apple. I guess Microsoft's big scandal is next with Cortana? There's really nothing left to say. How tech people understand things is...
Well we've now seen this same story for Amazon, Google, and Apple. I guess Microsoft's big scandal is next with Cortana?
There's really nothing left to say. How tech people understand things is different than how lay people understand things. Machine learning obviously requires training data, but many assume it's magic.
The problem is probably best solved by improving transparency, offering opt-outs, or my preference: offering incentivized opt-ins. Apple could offer some free iCloud usage if you join their "training program" or such.
The only issue with this strategy from a machine learning standpoint is that you'll get a biased dataset: people likely to opt-in to this kind of program are not necessarily representative of the...
or my preference: offering incentivized opt-ins. Apple could offer some free iCloud usage if you join their "training program" or such.
The only issue with this strategy from a machine learning standpoint is that you'll get a biased dataset: people likely to opt-in to this kind of program are not necessarily representative of the whole, and their Siri queries might not reflect overall Siri queries very well.
I don't think that's a very big problem, but it shows how fast things can get tricky when you don't have all the data you need in machine learning, which is most of the time.
Yeah, that is true. I think it's just something you have to accept. It happens quite a bit in tech already. For example, privacy-focused people are quick to opt-out of analytics. As a result they...
Yeah, that is true. I think it's just something you have to accept. It happens quite a bit in tech already.
For example, privacy-focused people are quick to opt-out of analytics. As a result they are also underrepresented when it comes to developers determining important issues for their users. That's a problem for topics like deciding to provide Linux support for applications, as these users appear even smaller than they already are.
I've been making more of an effort to leave analytics enabled where possible. As long as the company is reputable, and has a reasonable privacy policy. eg. I've left on Mozilla's "Health Report" in Firefox.
I’m glad that they’re a) suspending the reviews pending their investigation into the issue and b) plainly stating, without equivocation, that they’ll be offering an opt-out in the future. I wish...
I’m glad that they’re a) suspending the reviews pending their investigation into the issue and b) plainly stating, without equivocation, that they’ll be offering an opt-out in the future. I wish other companies like Amazon and Google had reacted similarly.
Having the program be opt-in would be even better, but I do understand that they need data and making it opt-in would probably limit the amount of data they collected, as well as the types of people it was collected from. And for this type of thing, a large cross section of people is extremely important.
I’m happy to see that Apple is moving on this. Now that Apple includes privacy as part of its value offering, it has no choice but to respond to public pressure on privacy concerns.
I’m happy to see that Apple is moving on this. Now that Apple includes privacy as part of its value offering, it has no choice but to respond to public pressure on privacy concerns.
Well we've now seen this same story for Amazon, Google, and Apple. I guess Microsoft's big scandal is next with Cortana?
There's really nothing left to say. How tech people understand things is different than how lay people understand things. Machine learning obviously requires training data, but many assume it's magic.
The problem is probably best solved by improving transparency, offering opt-outs, or my preference: offering incentivized opt-ins. Apple could offer some free iCloud usage if you join their "training program" or such.
The only issue with this strategy from a machine learning standpoint is that you'll get a biased dataset: people likely to opt-in to this kind of program are not necessarily representative of the whole, and their Siri queries might not reflect overall Siri queries very well.
I don't think that's a very big problem, but it shows how fast things can get tricky when you don't have all the data you need in machine learning, which is most of the time.
Yeah, that is true. I think it's just something you have to accept. It happens quite a bit in tech already.
For example, privacy-focused people are quick to opt-out of analytics. As a result they are also underrepresented when it comes to developers determining important issues for their users. That's a problem for topics like deciding to provide Linux support for applications, as these users appear even smaller than they already are.
I've been making more of an effort to leave analytics enabled where possible. As long as the company is reputable, and has a reasonable privacy policy. eg. I've left on Mozilla's "Health Report" in Firefox.
I’m glad that they’re a) suspending the reviews pending their investigation into the issue and b) plainly stating, without equivocation, that they’ll be offering an opt-out in the future. I wish other companies like Amazon and Google had reacted similarly.
Having the program be opt-in would be even better, but I do understand that they need data and making it opt-in would probably limit the amount of data they collected, as well as the types of people it was collected from. And for this type of thing, a large cross section of people is extremely important.
I’m happy to see that Apple is moving on this. Now that Apple includes privacy as part of its value offering, it has no choice but to respond to public pressure on privacy concerns.