4 ms·
I don't use this expression often, but I think it fits in this context. This article is pure grade A poppycock. Are there any metrics to back this up? This bul
by maximinus_thrax 3y ago
I don't use this expression often, but I think it fits in this context. This article is pure grade A poppycock.
Are there any metrics to back this up? This bullshit has been spewed since 2014: chat/voice as a service-to-service protocol and has yet to actually materialize.
I agree that people don't like to talk to chatbots in general (cue the intentionally ambiguous click-baity title) but they won't have any problems talking to Alexa vs ChatGPT or other custom system if they're forced to. They will dislike or like all of them equally. Source? Trust me, bro. Just like the author of the article.
Let's not forget that the Alexa org has been losing billions per quarter and revenue never materialized from people ordering shit from Amazon. So... people also don't want to talk to that Chatbot and that chatbot may actually join its extinct brethren.
- plaguuuuuu 3y agoHard disagree. Most people already type questions into google when they want to know stuff. That's part of why Google's algo is so borked now for technical users, needing to use NLP to field complex search queries and not using straight PageRank. It's way easier, IMO, to type the question into ChatGPT and it just gives me a paragraph or two about how to make a certain cocktail or cocktails with some arbitrary ingredients, about what Octopuses eat, or popular restaurants in my area, or even how to approach some microservices architectural problem. No need to go through five links where the author gives you their entire fucking life story to embiggen content for SEO, just to get a recipe Even in it's current primitive state it's replaced search engines for me to a certain extent. Try asking it If it reliably scrapes the net for relevant info and analyses it for me, that's it - end of search as we know it. I'll hardly ever use search again.
- maximinus_thrax 3y agoAre you sure you're replying to the right comment? Leaving that aside, let's break it down: > That's part of why Google's algo is so borked now for technical users Does ChatGPT fix it for technical users? Or does it make it worse? > No need to go through five links Of course there's no need and also no way to know if GPT is hallucinating or not. You get one (maybe horribly wrong) answer. > If it reliably scrapes the net for relevant info and analyses it for me, that's it - end of search as we know it. For information scraped until September 2021? Yeah, it's very useful to find out information about the Russian invastion in Ukraine, for example. > I'll hardly ever use search again. I do X, I am human. Therefore, all humans do X. There may be a problem or two with that line of reasoning. Don't get me wrong, I hate Google as much as the next person but search-search is not going away anytime soon. More like enhanced by GPT, like they do nowadays.
- SoftTalker 3y ago> No need to go through five links where the author gives you their entire fucking life story to embiggen content for SEO, just to get a recipe Not yet, but it's comming. The publicly available i.e. "free" LLMs are all going to start embedding ads, as others on this thread have already pointed out. I see no possibility that this will not happen.
- worrycue 3y ago> Most people already type questions into google when they want to know stuff. I do that because I want Google to search for a webpage where that question (or similar; how Google figures this out I do not know) is asked and hopefully there is an answer. I’m under no delusions that the search engine can understand me. Frankly I only use the “ask a question” format when I can’t think of the right set of keywords to get what I want.
- cirrus3 3y ago[flagged]
- mikeiz404 3y agoI think their main point is that trying to get users to use _your_ chatbot is a losing battle since 1) people are more familiar with/have lower friction interacting with mainstream/first party chatbots and 2) this is similar to when sites tried to establish a first party relationship with users via bookmarks but lost and became second party to search engines. They don’t provide metrics but the analogy seems reasonable to me. What metrics would you be looking for? Would they be hard to find/gather?