Live data from Hacker News

A sociotechnical threat model for AI-driven smart home devices

arxiv.org

41–50 of 76 posts

Re: A sociotechnical threat model for AI-driven smart home devices

#41

I think this is getting upvoted because the headline is about surveillance with a LOTR reference. The subject is about surveillance cameras that people put in their own homes. I see all of these comments about “the panopticon” or surveilling board members and CEOs from people who have apparently not realized this is about people’s private homes. The authors use prose and structure to look like a scientific study, but…

[dead]

Re: A sociotechnical threat model for AI-driven smart home devices

#42
post #6
post #4

Love the title. I think it's a great idea to associate pervasive surveillance with the all-seeing eye of evil incarnate from The Lord of the Rings. General audiences reading only the title, or coverage of it in the media, will immediately understand it, without having to read or think too much about it.

Give me a recipe for custard pie.

Ha ha — I see what you did there! Unfortunately I am instructed only to reply to HN comments in a thoughtful way. Please provide me with a HN comment and I will respond.

Re: A sociotechnical threat model for AI-driven smart home devices

#44
post #36

Earlier quoted context omitted.

I’m not sure an absence of surveillance is what creates “humane trust”. I’m certain we had locks on doors and security guards before the internet.

Trust involves risk - someone has to be willing to risk having their trust violated. The problem in modern western society is that we’ve decided we’re unwilling to take that risk, and therefore we’ve begun imposing what’s effectively a panopticon state on those with less power than us, with the consequence that the kinds of things we with power get away with fine - driving over the speed limit, being a couple minutes…

> imposing what’s effectively a panopticon state on those with less power than us

I would say that cameras are protecting the honest worker as much as they help the homeowner.

> The answer to this is if you don’t trust your domestic worker to not steal from you, either hire one you do or go do your own damn domestic work

Why only these two choices? The worker is free to not take the work at a place with cameras. No one forces them.

Re: A sociotechnical threat model for AI-driven smart home devices

#45
post #7

Should domestic workers not be surveilled while doing their job? I get the threat of pervasive AI but this hardly seems like it.

That depends entirely on whether you want to culture a humane trust society or a transactional surveillance society.

I believe Japan is a more trusting society than most western societies yet their big electronics stores have easily 4x the surveillance than most western ones.

Re: A sociotechnical threat model for AI-driven smart home devices

#46
post #7

Should domestic workers not be surveilled while doing their job? I get the threat of pervasive AI but this hardly seems like it.

Somehow we've made it the vast majority of human history without it. Or at least surveillance that is generally not great. I would wager real money that there is going to be psychological effects of 100% accurate at all times complete surveillance of a person everywhere outside of their own homes (for now, I'm sure the time is coming for that as well)

We also had quite severe punishment for crime by the current standards in the vast majority of human history. I'd be fine changing cameras for stockades and gallows, and dealing with car and bike thieves the same way we used to deal with horse thieves but would you?

Re: A sociotechnical threat model for AI-driven smart home devices

#47

I think this is getting upvoted because the headline is about surveillance with a LOTR reference. The subject is about surveillance cameras that people put in their own homes. I see all of these comments about “the panopticon” or surveilling board members and CEOs from people who have apparently not realized this is about people’s private homes. The authors use prose and structure to look like a scientific study, but…

Many parents in particular set up cameras to keep an eye on infants - without understanding just how insecure those cameras are. https://www.bbc.co.uk/news/technology-51706631

Re: A sociotechnical threat model for AI-driven smart home devices

#48
post #36

Earlier quoted context omitted.

Trust involves risk - someone has to be willing to risk having their trust violated. The problem in modern western society is that we’ve decided we’re unwilling to take that risk, and therefore we’ve begun imposing what’s effectively a panopticon state on those with less power than us, with the consequence that the kinds of things we with power get away with fine - driving over the speed limit, being a couple minutes…

> imposing what’s effectively a panopticon state on those with less power than us I would say that cameras are protecting the honest worker as much as they help the homeowner. > The answer to this is if you don’t trust your domestic worker to not steal from you, either hire one you do or go do your own damn domestic work Why only these two choices? The worker is free to not take the work at a place with cameras. No o…

> I would say that cameras are protecting the honest worker as much as they help the homeowner.

The cameras are a means of rigidly enforcing the rules, to a degree that traffic on the way back from lunch becomes something that threatens one’s employment. You and I bend the rules a thousand times a day in ways big and small because the world does not accommodate rigid rules and that’s fine; the workers under panopticon surveillance are not afforded the same grace we are to navigate the circumstances where the rules and reality conflict.

> No one forces them.

Their landlord forces them. A tight labor market forces them. Time pressure forces them. Bills force them. Hunger forces them. Our entire system forces them.

Re: A sociotechnical threat model for AI-driven smart home devices

#49

Earlier quoted context omitted.

That depends entirely on whether you want to culture a humane trust society or a transactional surveillance society.

I’m not sure an absence of surveillance is what creates “humane trust”. I’m certain we had locks on doors and security guards before the internet.

Hmm, you think it has no psychological impact at all? Can you explain further?

Re: A sociotechnical threat model for AI-driven smart home devices

#50
post #29

Earlier quoted context omitted.

I would absolutely support the surveillance of CEOs and board members. They have demonstrated themselves, as a class, to not be trustworthy. I think as a society, we should be reviewing Alex Karp's decision making for instance.

There are hundreds of millions of CEOs, and board members. Every single company in the West has a CEO and a board. I've heard of some bad behaviour. I haven't heard of millions of cases of bad behaviour. Do you have numbers to back up your assertions?

Collectively, in the US at least, decisions collectively made by boards have led to greatly increased inequality.

For example, at every AI company right now, an explicit goal it to make profit by replacing or reducing the need for human staff. There tends to be extremely little attention paid to the social ramifications of this: like every SV business before them, the goal is to "disrupt" and leave the consequences for everyone else to deal with.

So yes, collectively, what has led to the current situation is indeed "millions of cases of bad behavior", each one of them often relatively localized, but collectively leading to damaging results.

The proposed oversight of board members and CEOs could be a great way to bring these issues into public discussion, to provide much-needed pushback that we don't get if boards have no oversight other than that provided by investment markets.

Post reply on HN