it did not. google own deepmind, but the restrictions on data usage are pretty tight.
> I expect a HN to understand how this is usually impossible. Correlating data may seem hard to the layman, but Google is experience at data analysis.
yes, but then again, i don't expect you to know what data it is that deepmind are working with, and what their primary motives are. whilst i don't know for certain, i heavily doubt the data is leaving deepmind.
> who are trying to find subtle correlations in the medical data
that is not what they are doing.
> That's not the only reason to not want your medical data spread around. A better reason might be not wanting your data in more places where it could be stolen or leaked.
oh please. i trust google far more than my own government for data protection; we're talking about an orginisation who leaves sensitive literally just lying around (http://news.bbc.co.uk/2/hi/uk/7449927.stm).
> The "pathetic" part is the complete disregard for the people that generated that data. It doesn't matter if Google is talented at what they do, or if the data might be used for something that you might think is useful. I may even agree about the utility of the data, but if you aren't getting explicit informed consent from the people involved, you have a serious ethics problem.
an aside; it's funny that people are far more concerned about their medical data than they are their personal history. my medical data doesn't define me, but my actions do. if only such zeal existed for governments indiscriminate collection of personal data. oh well.
but back on topic, the data deepmind has (and i want to make that distinction that google does not have it, because they do not). it is anonymised sufficiently such that short of heavy analysis (which i concede is probably possible) very few inferences about the identity of the patient can be drawn. further, deepmind has absolutely no interest in correlating this data with people, and less than zero interest in targeting adverts to patients.
We're talking about a corporation that had employees access personal data to show off to teenagers[0].
No one can be trusted with data, therefore it needs to be restricted to only those who need access. The NHS probably needs data about patients, Google does not and the unspecified third-party contractor Google has employed to store the data definitely does not.
[0] http://techcrunch.com/2010/09/14/google-engineer-spying-fire...
DeepMind IS Google. Asserting that it isn't doesn't magically make them a separate legal entity.
> restrictions on data usage
I really don't care what the restrictions are today. The data exists under Google's control.
> i don't expect you to know
Arrogant insults are not a useful form of rhetoric. I read the actual agreement, and while it wasn't anywhere as detailed as I would have liked, it was fairly clear about the data that would be sent.
> primary motives
Again, this is irrelevant.
> that is not what they are doing
That's (part of) what DeepMind does.
> trust google
You may trust Google, but trust isn't transitive. You don't get to make that decision for other people.
> more concerned
You don't know that; you should really stop assuming you know what other people's opinions are.
> about their medical data than they are their personal history
Both of those are important. Where people place their concern will depend on their personal situation, which might or might not include medical data.
> my medical data doesn't define me
You're projecting your own situation on others. For someone with HIV (which is one of that data points being sent to Google), that medical fact may play a very defining role in their life.
> if only such zeal existed for governments
It does. Why would you think that someone that is against Google gathering more data is fine with the government doing the same? Both of those groups need to be defended against.
> an orginisation who leaves sensitive literally just lying around
Pointing to problems at other orginizations doesn't excuse making new problems.
> anonymised sufficiently
Then prove it; the agreement only listed [name (including initials0, address, NHS number, photographs or videos] as "identifying" information and [full postcode, date of birth, telephone number, e-mail address] as identifying in combination. It requires that those be sent encrypted (why not hashed? this means they can be decrypted).
> deepmind has absolutely no interest
Who cares what their interest is today. They may not have the choice, there interests can change, etc. The argument that everyone should "trust us, we're not greedy and have only good motives" has always been stupid, and Google themselves have a poor history of respecting user data.
in the context of this data, yes, there is significant distinction between the two.
> Arrogant insults are not a useful form of rhetoric.
sorry, i'm not trying to be arrogant.
> You may trust Google, but trust isn't transitive.
true.
> You're projecting your own situation on others. For someone with HIV (which is one of that data points being sent to Google), that medical fact may play a very defining role in their life.
i'd argue that letting yourself be defined by your HIV status is a poor metric to live ones life by, but you're right, that's not my call to make.
> It does. Why would you think that someone that is against Google gathering more data is fine with the government doing the same? Both of those groups need to be defended against.
playing the cynical card about governments vs corporations - a corp has no incentive in using my data in the same way a government has ala dragnet surveilence, to which there is a resounding response of apathy toward. of course the HN crowd care about these things, but we're a somewhat isolated bubble of people compared to the rest of society.. although that's another discussion.
> Then prove it;
of course, i cannot do that. and my feelings toward a group of people can't dictate how others should feel toward them - i know. but the team at deepmind are looking to do some amazing things. and this article from vice screams ignorance. maybe deepmind could be more transparent about their work. that's not my call though.
> Who cares what their interest is today.
it matters significantly - they can't keep the data after sept 2017, so if for some reason the direction of the group changes, there is only so much they can do with it. but it would set the field back decades if anyone at deepmind were to do anything so irresponsible as what you might be suggesting.
Did you miss the disclosures about that surveillance over the last few years, which indicate that governments are taking data from from those corporate databases? With programs like (but not limited to) Prism, giving data to a corporation may effectively be the same giving it to governments.
> anyone at deepmind were to do
You need to consider a threat model that includes more than that one group. What about everyone else at Google, or anybody with access to Google's internal networks, etc. Security is a process that has to be applied at all layers.
The data "got to" Google when it was sent to Google DeepMind.
> anonymised
I expect a HN to understand how this is usually impossible. Correlating data may seem hard to the layman, but Google is experience at data analysis.
Arguing that Google's AI division - who are trying to find subtle correlations in the medical data - somehow cannot correlate those same records back to their "personally identifiable information" laughable.
> use it to
That's not the only reason to not want your medical data spread around. A better reason might be not wanting your data in more places where it could be stolen or leaked.
edit:
> pathetic
The "pathetic" part is the complete disregard for the people that generated that data. It doesn't matter if Google is talented at what they do, or if the data might be used for something that you might think is useful. I may even agree about the utility of the data, but if you aren't getting explicit informed consent from the people involved, you have a serious ethics problem.