4 ms·

> you can't just change definitions like this Of course you can, don’t be obtuse. Words can literally mean anything the speaker wants them to, language evolves

• danShumway 3y ago
  > language evolves Language evolves organically through public usage, not via soda FAQs. If you asked pretty much anyone on the street in any context "an AI is going to scan a video and locate on object, is it 'watching' the video" -- they would say yes. This is not a niche interpretation of what it means to watch a video. The visual data from the video output is being piped into an image recognition engine; the vast majority of the general public would call that "watching". Maybe in the future that will change, but unless Mountain Dew has a time machine hidden in their office, "language evolves" is not an excuse. > if they’re not storing it the details about the stream then their target audience would agree that they’re not watching and storing data. They are storing details about the stream, they have to or else the system doesn't work. They have to store information about which streams they've checked, haven't checked, which streams have opted in, which streams have opted out, when they last messaged the stream or spammed in chat. The FAQ also implies that they are storing frequency information about how often the logo appears, so it's not even accurate to say they're not storing data that relates to the actual contents of the stream. That's not nitpicking, it's not a technicality, we're not debating whether or not a proxy service counts as storage or talking about where data is at rest. They are just straight up lying -- there is stream and content metadata being stored on their servers and being used for the project. > and it’s doubly bad when the target audience neither knows nor cares about technical details. "Are you storing information about me" "None that you would understand" is a terrible answer for a FAQ and belittles the reader. The streams are intended to be public, they could just say, "yes we are scanning public streams and collecting metadata." People know what metadata is, it's not that complicated. There are so many simple answers they could give to this question that would be honest to what they're doing without risking confusion. "The public would be confused" is not an excuse to outright lie about what information a company is and isn't storing. Obviously it's not an excuse, this is Facebook levels of weirdness to suggest, to say "the public doesn't understand what we're storing so we don't have to inform them about it."
  • c0pium 3y ago
    > This is not a niche interpretation of what it means to watch a video You should be aware that is very much up for debate. I have seen lots of market research on words around AI, and there is a large group of people who think that it is impossible to use human verbs like “watch” in the context of AI. Computer programs can’t watch things any more than you can be overclocked; the word simply doesn’t apply. Your examples of storing data about streams are similarly confused about how people use language. That’s not data on streams, it’s data on channels. There’s no reason for them to store data on streams. Whether the can appeared has nothing to do with the content of the stream. The content is whatever activity is happening, not some irrelevant prop detail that has nothing to do with the content. Save your outrage for something that’s worth it, this is a nothing burger.
    • danShumway 3y ago
      > and there is a large group of people who think that it is impossible to use human verbs like “watch” in the context of AI Oh come on, you're reaching here. If the intention was to make a commentary on the personhood of AI, the FAQ would not have said, "the AI isn't watching your video", it would have said, "the AI lacks the ability to perceive reality in human terms." I think we both know that an advertiser was not sitting down to write this thinking, "wait, holy crud the AI isn't watching the video... because it can't watch anything". Keep in mind the context that this is a corporate FAQ about an advertising campaign, not a philosophical essay about AI personhood. > That’s not data on streams, it’s data on channels. [...] Whether the can appeared has nothing to do with the content of the stream. What on earth are you talking about? The visible content in the video stream is not the content in the stream because it's not important enough to the overall content of the stream, so that means this is all data on channels, even though the data is specifically relevant to individual streams and not to channels and even though all of the communication between the bot and the streamer is happening through the individual stream and not through the channel, and even though all of the policies are per-stream and not per-channel. Do I understand that correctly? This is absurd. > Save your outrage I guess speaking of ambiguity of language on HN I should clarify that I'm not outraged at Mountain Dew, I'm mocking the company. I agree this is not a big issue largely because the advertising campaign itself is pathetic and meaningless and is probably going to be largely ignored by actual streamers. This is a campaign that in the absence of compensation is offering streamers the ability to "supe" their followers via randomly selected 10-minute clips on the Mountain Dew Twitch channel. So agreed, this is not worth taking seriously. But while it's not worth getting outraged over the Mountain Dew company debasing itself, attempts to excuse deceptive or outright inaccurate messaging about corporate data usage based on technicalities and strained excuses about what people "understand" about privacy is all too common in the current tech landscape and it's generally good to mock the companies that try to pull that crap and to call them out as scummy.