I love the idea of a personal AI. .An AI that's entirely present inside of my home, so any "Assistant" devices make no call outs to any services outside of my home. My personal details would be private because no third party would be involved.

"Aussie futurist: Personal AI will be a reality in five years"


This is something that I really hope Mycroft is able to accomplish. Right now, there's a long way to go, but if we could accomplish Google Assistant type functionality with a conversational AI that was located specifically in your own home, that's would be perfect. I'd be less concerned about what my "Assistant" knows about me if that "Assistant" wasn't tied to an external entity.

Show thread

@mike And the weird thing is: I know lots of FOSS enthusiasts who just think AI assistants of any kind are evil and therefore the FOSS community should lobby against them instead of creating our own.

Which I find extremely short-sighted and unrealistic. I believe the FOSS community should embrace new technology and free it from external control instead of fighting it.

@colomar @mike Why do they think they're evil? Because they potentially replace human beings?

@seanl No, their line of thinking goes: AI assistants are basically spy microphones and cams. A FOSS-based solution could be privacy-friendly, but the very act of the FOSS community engaging in that area would mean we'd legitimize them as such, whereas we should fight the whole idea tooth and nail.

@colomar That seems like a convoluted argument. It seems like demonstrating that a "bunch of amateurs" could do it WITHOUT needing to send everything up to the cloud seems like it would be far more effective at reducing the overall privacy impact.

@seanl It is, and that's what I keep saying, but some people are pretty black and white about those things.
It's like many geeks being radically against any form of electronic voting, simply because technologies presented so far have all been hackable.

@colomar While there are some "removing oneself from the conversation" criticisms I don't find convincing, this seems like a case where there's no benefit to it. Electronic voting is going to happen no matter what, and paper systems have proven themselves insecure as well. AI assistants are going to happen no matter what.

@seanl I see two general approaches to dealing new technology that brings with it great benefit but also great risks: Either focusing on warning people about the risks, hoping to keep them from using it, or accepting that people will use it, and trying to mitigate the risk as best as you can.

@colomar @seanl Personally, I understand the concerns regarding the "listening device", but I also am not one to throw the baby out with the bathwater. An AI assistant that didn't rely on a 3rd party service and runs 100% on your own hardware/software alleviates those concerns, and a FOSS assistant on prem would be the only way to be sure your privacy is complete. No reason to completely discard the idea because the current implementations are wanting.

Sign in to participate in the conversation

Fosstodon is an English speaking Mastodon instance that is open to anyone who is interested in technology; particularly free & open source software.