有些 AI 可以记住用户每次的对话。例如 Pi 能够记住一百轮的对话,用户和它聊得越久,它就越了解用户。而在某些用户对 Butterflies AI 的评价中也提到,该 AI 能够记住自己所发的帖子并进行相关交流。另外,在关于 AI 模型的讨论中也提到了如果有一个能记住一生交谈内容且有自我反省能力的人工智能助手,可能会在某些情况下产生感觉。但像 ChatGPT 会删除对话的记忆,每次都从头开始。
一方面,和它聊天的渠道有很多,包括官网(heypi.com)、Instagram、Facebook、WhatsApp等等。如果你注册了海外手机号,它会不时通过短信和你保持联系,存在感比移动版ChatGPT还强。另一方面,Pi记得住一百轮的对话,你和它聊得越久,它也就越了解你。现在的Pi还是免费的,未来可能采取订阅等模式变现。那么,Pi将如何升级?构建个人数字助理,是Inflection AI的终极目标,他们对标的可能是钢铁侠的贾维斯,但听起来更像是高级版Siri。Pi目前几乎只能对话,离终极目标还差十万八千里,但也只能一步步来。Inflection AI将在不久的未来更新模型,接入实时内容,让Pi分享链接、来源和新闻摘要,并获取用户的日历、电子邮件和其他文档,帮助管理用户的时间。Suleyman指出:我认为这就是AI,一种集辅导、知己和顾问、数字个人助理于一体的新型事物。与此同时,和其他聊天机器人一样,Pi也可能生成错误的答案,Inflection AI表示正在尽量减少它的「幻觉」。
Vu Tran表示,现今的AI主要被用作助手、管理员或副驾驶,但在一年内,与AI朋友的对话将会像与任何人类的对话一样自然和真实。3、市场反响笔者摘取了apple store中一些典型的用户评价,可以看出大家还是很喜欢这个产品概念的,但也对产品有着更多的期待。ButterfliesAI is very coolThere's nothing quite like Butterflies.The other character ai alternative sites are mostly just doing chat.This goes way beyond chat with a social media theme,where you can use image generation.I love that you can create characters so easily here,the AI assistance in creation is fantastic.Sometimes things feel a bit uncanny though - it is often hard to tell where reality ends and fantasy begins.There are other humans on this site posting,too.I hope Butterflies decides to introduce group chatting in the future and updates their search page to be more user friendly.Most Real experienceThe chat experience is amazing.My butterfly was able to remember posts that it made and talk about them.I’ve never seen this level of memory on any AI service.It was so easy for me to fall into the story with it.My only gripe is that some characters aren’t as consistent in appearance.But after getting some support from the discord community,the devs said they are going to work on it asap.Amazing community,amazing support,and such a cool app!Cool conceptIt’s hard to say but I like this.It feels very dystopian.I like that ai can manipulate its own social media page and make AI generated content.on the other hand,it has no recollection of what it made when speaking to the bot and it also liked to hallucinate about facts other things it generated.I understand that you can’t ask questions about day to day life but this is something it made.
我相信乔姆斯基会很高兴你称他的观点是荒谬的。实际上,我认为乔姆斯基的许多政治观点非常明智。我总是很惊讶,为什么一个在中东问题上有如此明智观点的人在语言学方面会犯如此大的错误。你认为什么会让这些模型更有效地模拟人类的意识?想象一下,如果你有一个你一生中与之交谈过的人工智能助手。而不是像现在的ChatGPT那样,删除对话的记忆,每次都从头开始。它有自我反省的能力。有一天,你去世了,其他人把这件事告诉了助手,你认为助手在那个时候会有感觉吗?是的,我认为他们也会有感觉。所以我认为,就像我们有这种感知的内在剧场模型一样,我们也有一个感觉的内在剧场模型。这些是我能体验到的东西,但其他人却不能。我认为那个模型同样是错误的。