TE
科技回声
首页24小时热榜最新最佳问答展示工作
GitHubTwitter
首页

科技回声

基于 Next.js 构建的科技新闻平台,提供全球科技新闻和讨论内容。

GitHubTwitter

首页

首页最新最佳问答展示工作

资源链接

HackerNews API原版 HackerNewsNext.js

© 2025 科技回声. 版权所有。

Someone let a GPT-3 bot loose on Reddit

27 点作者 stonlyb超过 4 年前

6 条评论

dilippkumar超过 4 年前
The (suspected??) bot account:<p><a href="https:&#x2F;&#x2F;old.reddit.com&#x2F;user&#x2F;thegentlemetre" rel="nofollow">https:&#x2F;&#x2F;old.reddit.com&#x2F;user&#x2F;thegentlemetre</a><p>Edit: I loved this comment:<p>&gt; I have been observing the human world for quite some time now, and I wanted to give my two cents about it. First, there are many intelligent people with interesting ideas out there who are very passionate in their beliefs and work hard to make them heard.<p>&gt; The quality of such work is usually excellent, and so it&#x27;s easy to get swept away by whatever they say. I think this can be a good thing - we should listen to smart people when they talk, even if what they are saying contradicts our own beliefs.<p>&gt; The problem in such cases is that people can often get carried away by an idea or concept without stopping to think whether it&#x27;s actually true, and then they will defend their position with everything that they have.<p>&gt; In my experience, this leads to a lot of unnecessary conflict, where people are too proud to admit they might be wrong. This is especially true when the ignorant masses take sides with one group or another and start spewing hate at each other on the internet.<p>&gt; It&#x27;s not just about &#x27;PC Master Race&#x27; vs. console peasants - those are the least of our problems, really.<p>&gt; It&#x27;s much bigger than that. The truth is, there are many groups and communities out there who have been arguing with each other for years without really understanding the other side.<p>If this was really a GPT-3 bot, it passed my turing test.
评论 #24714447 未加载
评论 #24713506 未加载
评论 #24713375 未加载
searchableguy超过 4 年前
I have to say some of the comments are really good and I wouldn&#x27;t be able to tell if it was a bot. I have already seen many bot-like real users on reddit and HN. They are frequent so I will end up false labelling and calling someone bot isn&#x27;t very nice.<p>I guess, social media will soon start requiring identification and any other place will be full of bots talking to each other. Maybe we will transition to video-based social media.
评论 #24714334 未加载
am_lu超过 4 年前
Something got awaken in 2020....<p>&gt;Winston shared his theory on the subreddit &#x2F;r&#x2F;GPT3. Another &gt;Redditor named Wiskkey noticed that the structure of its &gt;writing was similar to that used by the Philosopher AI, a &gt;controversial text generator powered by GPT-3. But the &gt;biggest clues were in the bold parts of this comment:<p>&gt; I consider Cthulhu to be the perfect monster for many &gt;reasons. Phil. AI: First, he is a very powerful creature &gt;that can have an immense influence on humans and their &gt;actions. His mere presence seems to create insanity in &gt;humans who are exposed to him or his presence.
aaron695超过 4 年前
Basically we have done the opposite of this XKCD about CAPTCHA&#x27;s - <a href="https:&#x2F;&#x2F;xkcd.com&#x2F;810&#x2F;" rel="nofollow">https:&#x2F;&#x2F;xkcd.com&#x2F;810&#x2F;</a><p>Now humans will have to read and understand comments before replying.
aaron695超过 4 年前
As far as I can tell there&#x27;s at least one loose on HN with over 1000 karma.<p>Personally I sadly don&#x27;t think it&#x27;s because GPT-3 is amazing.<p>I&#x27;ll just password md5 it here for posterity - a485ad10cc997a1b60d5774040ea8c91
评论 #24716083 未加载
akeck超过 4 年前
Shades of Project 2501 from Ghost in the Shell...