3 ms·
I feel like you are complaining about the first car. Yes, it sucks and, god, I'm not sure it's good for the world. But zoom out a bit and it seems fairly clear
by solidasparagus 2y ago
I feel like you are complaining about the first car. Yes, it sucks and, god, I'm not sure it's good for the world. But zoom out a bit and it seems fairly clear to me that this is what the future is going to look like and this is why Microsoft feels compelled to act.
This is the first step to creating an assistant that automatically does tasks for you on your computer driven by natural language. The context being collected by Recall is essential for this technology to work effectively as people frame natural language instructions with the assumption that the assistant shares your context. And they probably want to collect a training set, although it's excellent (and necessary for trust) that this data is local-only, which might mean this doesn't generate any useful training data. I'm super happy about that.
Despite occasional snarky comments from tech savvy people, people absolutely do want this - normal people like computing to be simpler.
I do agree that this feels a bit early, but I don't know enough to say yet. Rewind AI (the startup) handles privacy by not capturing incognito windows, which I feel is a good start and we need to keep iterating on techniques like that to balance trust against usefulness (of course with the ability to disable the whole thing if you are willing to lose the features). Hopefully Microsoft spent enough time thinking about this, but time will tell.
Fundamentally the need for AI to understand your context to correctly assist you as well as the feedback techniques essential to training effective AI are going to run up against privacy concerns and we're going to have to figure them out as an industry - preferably in public, preferably with a company that takes trust seriously. This is fundamental to the technology IMO. Is Microsoft that company? I don't know, but I wouldn't underestimate how valuable this data would be if it was on Microsoft's server, so I personally am quite happy that they made the right choice there.
- wkat4242 2y ago> people absolutely do want this - normal people like computing to be simpler. Of course I want this. I just want to have it myself. Like for the assistant and its data to be truly mine. I don't want Microsoft to run it for me. I don't mind paying for someone to build it but not to operate it. Especially not Microsoft. They're as bad as Google. I work with Microsoft people daily and they're so brainwashed. They think their company is beyond any reproach, that the products are the best in the market, that using them is a win for us etc. I hate them and their company with a passion but unfortunately our leadership is easily impressed and not too smart. They just hobble along with whatever Gartner is paid to tell them.
- solidasparagus 2y agoWhat do you mean for the assistant to be yours? There is a core model somewhere that needs to be trained and the cost/technical requirements are prohibitive for that to be done by each individual. If the model is local, is that enough? The data of course should be yours - and until we solve information leaks in AI (never?), it will have to be or the product has huge viability risks. Agree on data, which is one of the core issues. The expectation of privacy on your device is very high. But that can change, especially with the rise of web app tools that blur the line between your private space, the company's space, and public.
- asadotzler 2y agoThe first car wasn't automatically rolled out to 100% of horse and buggy users without their permission and no way to return to their buggy. Why wasn't it? Because it wasn't fit for purpose for most horse and buggy users, and it would be decades, almost half a century before it was.
- solidasparagus 2y agoYou can disable Recall.