5 ms·
In the case of the human personal assistant the governing terms originate with the person being assisted, and so the data and the ways it can be used remain in
by Normal_gaussian 10y ago
In the case of the human personal assistant the governing terms originate with the person being assisted, and so the data and the ways it can be used remain in control of the owner.
Tech companies are inverting this control, which has very dangerous implications for those they assist.
There are ways to keep control with the user, however these require a level of architecture inversion the consumers haven't fought for yet.
- AndrewKemendo 10y agoIn the case of the human personal assistant the governing terms originate with the person being assisted, and so the data and the ways it can be used remain in control of the owner. Well, the data and ways it can be used are also in control of the user now, they just all agree to it - even without knowing it in many cases. My question is: how do we talk to consumers to show them the benefits while also informing them of how it works? There are ways to keep control with the user, however these require a level of architecture inversion the consumers haven't fought for yet. I'd be curious to know how you see this working - especially in the case of providing a benefit from network effects. The benefit from these systems is using the inputs of other people, and those by definition release control. Beyond that case, making each user train their own CNN is an impossible consumer demand. The way facebook does it with face tagging and google does it with number/letter recognition, everyone is training the CNN - crowd-sourced neural net training.
- Normal_gaussian 10y agoYou seem to be arguing as though the current situation is preferable and consumers just need to be convinced of that. Firstly, the user is not in control of the contract. They have no realistic means of modification or exerting pressure, nor resisting or requiring changes. Take it or leave it is not control. Secondly. You are likely aware of the permissions systems being introduced on modern smartphones. This is an interesting step towards inverting the control of the device back towards the user. Similar steps can be made in other areas. CNN training is certainly a different action to usage, and has different privacy implications, so they should obtain permissions independently. Note how these systems are not permission forever, but permission for now. Handling this sort of data access is a pain, but it isn't actually hard.
- AndrewKemendo 10y agoYou seem to be arguing as though the current situation is preferable and consumers just need to be convinced of that. My whole stance is that it's unknown if it's preferable because people don't know what the long term costs are - as no one has told them about it. Clearly they are gaining value from utilizing these systems or they wouldn't be using them. What isn't understood is if they would still use them if they understood the potential long range privacy implications. Firstly, the user is not in control of the contract. They have no realistic means of modification or exerting pressure, nor resisting or requiring changes. Take it or leave it is not control. In the case of private services, it most certainly is control. Irrespective of that however, these systems aren't basic services or things that people have a right to, so your control argument is in the wrong context. Facebook et al. are frivolities - not needs.