被ChatGPT逼到墙角的谷歌，终于要正面反击了：
谷歌版ChatGPT，已确定发布。
谷歌CEO皮查伊亲自撰文宣布：将推出对话人工智能服务Bard。
就在今天，Bard将向测试人员开放，并在未来几周内与公众见面。
并且还明确，新的人工智能技术将会被首先整合到搜索服务中。
消息一出，直接在社交媒体上掀起狂澜。
一种“打起来打起来”的热闹氛围，已经在网友中蔓延开来。
更有华生发现了盲点，直指谷歌此番火药味十足：之所以给产品起名“Bard”（吟游诗人），大概是因为ChatGPT以莎士比亚风格的诗作闻名。而莎翁就被称作“吟游诗人”。
有意思的是，据CNBC消息，在一份内部备忘录中，皮查伊还要求员工“All in Bard”：全员加入这一AI聊天机器人的测试。
而且据硅星人消息，Bard大部分是由华人团队开发。
另外，量子位获悉，Bard的正式推出，应该就在明晚。
话不多说，一起来看看劈柴哥具体透露了哪些有关这个“谷歌版ChatGPT”的消息。
谷歌版ChatGPT定档
“先进的生成式人工智能和大型语言模型正在抓住世界各地人们的想象力”，皮查伊在公开信中如是说。
显然，谷歌并不打算在这个风口上拱手让位。
两年前，谷歌已经发布对话模型LaMDA，并在2022年开放了Demo体验。而根据皮查伊透露的信息，Bard依然是在LaMDA的基础上打造。
皮查伊在信中写道：
Bard的追求是将世界知识的广度与庞大语言模型的力量、智慧和创造力结合起来。
它利用来自网络的信息提供新鲜、高质量的回应。
他将Bard比喻为“创造力的发泄口”“好奇心的发射台”，因为它可以深入浅出地解释一些听起来很复杂的话题。
比如，能够帮助人们向一个9岁的孩子解释NASA韦伯望远镜的最新发现。
Bard真的有这样的力量吗？
为它提供技术支持的是轻量级版本的模型LaMDA。
LaMDA是一集基于Transformer架构的对话模型，它能够在“听懂”人类指令的基础上，对答如流，且保证事实无误、逻辑正确。
去年年初，LaMDA训练已经用到了1370亿参数。
值得一提的是，这一模型此前也引起了一波“恐慌”，有谷歌研究员认为其已具备人格。而他公布的与LaMDA之间的聊天记录，也让网友直呼“san值狂掉”。
而这回提到的轻量级，则意味着所需算力的显著降低，这利于服务更多的用户，也利于收集更多的反馈。
没错，Bard也会根据来自用户的外部反馈，加上内部测试，进行不断调整完善，目的是让Bard的回答“符合高质量标准”，并且安全、真实。
具体到发布的时间线上，皮查伊表示：
今天起，Bard向部分人员开放测试，并将在未来几周内向公众开放。
很高兴有这个阶段的测试，（它）能帮助我们继续提高Bard的质量和速度。
另外，谷歌从下个月开始，还将为Bard招募外部开发人员、企业，让他们能通过API加入到Bard的应用开发中。
尽管也有网友吐槽：谷歌这只是“宣布”，而非“发布”。
但更多人还是觉得：事情正在变得越来越有意思。
去年11月OpenAI“随手”发布的一个AI聊天机器人，现在迫使世界上最大的科技公司转向。
这真是一个激动人心的时刻。
而综合多方信息来看，“发布”也已定档。
明晚（2月8日）7点半，可以期待一波了~
翻译：
Google, backed into a corner by ChatGPT, is finally fighting back head-on:
Google’s version of ChatGPT has been confirmed for release.
Google CEO Pichai himself wrote to announce the launch of the conversational AI service Bard.
Just today, Bard will open its doors to testers and meet the public in the coming weeks.
It is also clear that new AI technologies will be integrated into search services first.
The news went viral on social media.
A “fight fight fight” lively atmosphere, has been spread among netizens.
What’s more, Watson sees a blind spot, pointing to Google’s explosive nature: the product was named “Bard” presumably because ChatGPT was known for his Shakespearean poetry. Shakespeare is called a Bard.
Interestingly, in an internal memo, Pichai also asked employees to be “All in Bard” : fully engaged in testing the AI chatbot, according to CNBC.
And according to Silicon Man sources, Bard was developed largely by Chinese teams.
In addition, Qubits have learned that the official launch of Bard should be tomorrow night.
Without further ado, let’s take a look at what ChatGPT Google has revealed.
Google version of ChatGPT
“Advanced generative artificial intelligence and large-scale language models are capturing the imagination of people around the world,” Pichai said in his letter.
Clearly, Google is not about to give up on this issue.
Google released LaMDA, a conversational model, two years ago, and opened up the Demo experience in 2022. According to Pichai, Bard is still based on LaMDA.
Pichai wrote in his letter:
Bard’s quest was to combine the breadth of the world’s knowledge with the power, wisdom, and creativity of vast linguistic models.
It uses information from the Web to provide fresh, high-quality responses.
He likens Bard to “an outlet for creativity” and “a launching pad for curiosity” because it can explain some seemingly complicated topics in a simple way.
For example, it helps people explain the latest discoveries from NASA’s Webb Telescope to a 9-year-old.
Does Bard really have such power?
It is powered by a lightweight version of the LaMDA model.
LaMDA is a dialogue model based on the Transformer architecture, which is able to “understand” human commands and answer them fluently, with correct facts and logic.
Early last year, LaMDA training had used 137 billion parameters.
It’s worth noting that this model has also caused a wave of “panic” in the past, with some Google researchers claiming that it already has a personality. And he posted chat logs with LaMDA, the Internet also called “san value crazy.”
In this case, lightweight means significantly less computing power, which allows for more users to be served and more feedback to be collected.
Yes, Bard also adjusts and refines its responses based on external feedback from users, as well as internal testing, in order to ensure that Bard’s responses “conform to high quality standards” and are safe and truthful.
In terms of the timeline for the launch, Pichai said:
Bard is open to a select group of people for beta testing starting today, and will be available to the general public in the coming weeks.
We are glad to have this phase of testing, which will help us continue to improve the quality and speed of the Bard.
In addition, starting next month, Google will also be recruiting outside developers and enterprises for Bard to participate in the development of Bard applications through the API.
Although some netizens quipped that Google was “announcing” rather than “releasing.”
But more people still feel that things are getting more interesting.
An AI chatbot that OpenAI “casually” released last November is now forcing the world’s biggest tech companies to turn around.
It’s really an exciting time.
And comprehensive information, “release” has also been set.
Tomorrow night (February 8) at 7:30, you can expect a wave ~
本文由数字化转型网（www.szhzxw.cn）转载而成，来源：量子位；编辑/翻译：数字化转型网宁檬树。
免责声明: 本网站(http://www.szhzxw.cn/)内容主要来自原创、合作媒体供稿和第三方投稿，凡在本网站出现的信息，均仅供参考。本网站将尽力确保所提供信息的准确性及可靠性，但不保证有关资料的准确性及可靠性，读者在使用前请进一步核实，并对任何自主决定的行为负责。本网站对有关资料所引致的错误、不确或遗漏，概不负任何法律责任。
本网站刊载的所有内容(包括但不仅限文字、图片、LOGO、音频、视频、软件、程序等) 版权归原作者所有。任何单位或个人认为本网站中的内容可能涉嫌侵犯其知识产权或存在不实内容时，请及时通知本站，予以删除。
