当前位置:首页|资讯|OpenAI

The history of OpenAI from early days to cutting-edge AI researc

作者:少侠李寻欢0_0发布时间:2023-05-02

In December 2015, a research institute was founded in Silicon Valley with the goal of providing humanity with a general artificial intelligence, this institute is called Open AI. And in this video we will be discussing what exactly it is, what kind of research Open AI is doing, and what criticism has been leveled at the company.

2015年12月,一家研究机构在硅谷成立,目标是为人类提供通用人工智能。这个研究所被称为Open AI。而在这段视频中,我们将讨论它到底是什么,Open AI正在做什么样的研究,以及对该公司有哪些批评意见。


It was founded on December 2015, in the United States by a veritable who's who of tech industry figures. The most well known founder of the Institute is, of course, Elon Must. Entrepreneur, billionaire and currently a source of Twitter drama.

它成立于2015年12月,在美国由一个名副其实的科技行业人物组成。该研究所最知名的创始人当然是埃隆-穆斯。企业家,亿万富翁,目前是推特上的一个源头。


Also among the founders is Sam Altman, the former president of Y Combinator, one of the most well known startup incubators in the United States. Additionally, Greg Brockman, the former chief technology officer of stripe, is among the founders as well as several researchers including Ilya Sutzkever, the researcher in the field of deep learning who has worked at Facebook, Google, and Nvidiam, and John Schulman, a researcher in the field of reinforcement learning.

创始人中还有美国最知名的创业公司孵化器之一Y Combinator的前总裁Sam Altman。此外,Stripe的前首席技术官格雷格-布罗克曼(Greg Brockman)也是创始人之一,还有几位研究人员,包括曾在Facebook、谷歌和Nvidiam工作的深度学习领域的研究人员伊利亚-苏茨克维尔(Ilya Sutzkever),以及强化学习领域的研究人员约翰-舒尔曼(John Schulman)。


At the time of its founding, open AI was provided with a hefty $1 billion in startup capital, which came from the founders themselves, as well as other individuals and institutions, such as Peter Thiel and Amazon Web Services. Open AI's goal is to be one of the leading research institutions in the field. of artificial intelligence, and to use its research to benefit humanity without being influenced by financial considerations. This is a lofty goal that has held true to its core, although there have been some financial issues that have arisen in recent years. Open AI primarily communicates its research findings through its own blog, although individual researchers also published scientific papers. However, let's be honest, the average person is not going to read those papers. On the blog, you can find well written, accessible articles that break down the latest research findings into a few paragraphs.

在成立之初,Open AI获得了10亿美元的巨额启动资金,这些资金来自创始人本身,以及其他个人和机构,如彼得-泰尔和亚马逊网络服务。Open AI的目标是成为人工智能领域的领先研究机构之一,并利用其研究为人类造福,不受财务因素的影响。这是一个崇高的目标,尽管近年来出现了一些财务问题,但它一直坚守着自己的核心。开放人工智能主要通过自己的博客来交流其研究成果,尽管个别研究人员也发表了科学论文。然而,说实话,一般人是不会去读这些论文的。在博客上,你可以找到写得很好、易于理解的文章,将最新的研究成果分解成几个段落。


In 2016, Open AI released a sort of toolkit for developers called Open AI Gym, which was designed to improve the development of reinforcement learning algorithms. The project is still ongoing and has been renamed Open AI API.

2016年,Open AI为开发者发布了一种名为Open AI Gym的工具包,旨在改善强化学习算法的开发。该项目仍在进行中,并已改名为Open AI API。


One of open AI's most well known projects is GPT2, a machine learning model that is able to generate human like text. It was first released in 2019, and has since been improved upon and expanded.

Open AI 最知名的项目之一是GPT2,这是一个能够生成类似人类文本的机器学习模型。它于2019年首次发布,此后又进行了改进和扩展。


Another major project for the company is Dactyl,  an artificial intelligence that is able to learn to solve a Rubik's cube using a human like robotic hand. This was seen as a major milestone in the field of robotics and artificial intelligence. 

该公司的另一个主要项目是Dactyl,这是一个人工智能,能够学习使用像人一样的机器人手解决魔方。这被看作是机器人和人工智能领域的一个重要里程碑。


In addition to his research, open AI has also focused on the ethical and societal implications of artificial intelligence, and his work to educate the public about the potential risks and benefits of AI. However, the company has faced some criticism for its close ties to powerful tech industry figures, as well as its decision to not release the full version of GPT to over concerns that it could be used to spread misinformation. Overall, Open AI is a leading player in the field of artificial intelligence research, and its work has had a significant impact on the development of AI technologies around the world. 

除了研究之外,Open AI 还关注人工智能的伦理和社会影响,以及他在教育公众了解人工智能的潜在风险和好处方面的工作。然而,该公司也面临一些批评,因为它与科技行业的强势人物关系密切,以及由于担心GPT可能被用来传播错误信息而决定不发布完整版本。总的来说,Open AI 是人工智能研究领域的领先者,其工作对全世界的人工智能技术发展产生了重大影响。


In a blog post was written about a charter and the key points, but this is not publicly available. It states that any influence of Open AI on general artificial intelligence should be used for the benefit of all, and should not cause harm under any circumstances. It also addresses the safety of research regarding LTI Open AI. The company is concerned that a late stage in the race for general artificial intelligence could become a competitive race without proper safety measures, if another company loses control of general artificial intelligence. The charter mentions technical leadership and a cooperative orientation.

该公司担心,如果另一家公司失去了对通用人工智能的控制,通用人工智能的后期阶段可能会成为一场没有适当安全措施的竞争。章程中提到了技术领导和合作方向。在一篇博文中写到了一个宪章和关键点,但这并没有公开。它指出,Open AI 对一般人工智能的任何影响都应该是为了所有人的利益,而且在任何情况下都不应该造成伤害。它还涉及有关LTI Open AI 的研究安全问题。


Open AI must always be up to date on the capabilities of artificial intelligence in order to effectively address the influence of general artificial intelligence. The institute also wants to cooperate with other research labs and work together to find solutions. In addition to this charter, something else of media interests happened. Elon Musk left the leadership of Open AI after just two years, it is reported that there were some disagreements and he needed to focus more on Tesla and SpaceX. This seems familiar. It's just that there was competition for hiring talented researchers. In June 2017, Tesla lost a well known AI researcher named Andrej Karpathy, a new donors were announced and the leadership was adjusted.

Open AI 必须始终掌握人工智能的能力,以有效解决通用人工智能的影响。该研究所还希望与其他研究实验室合作,共同寻找解决方案。除了这个章程之外,还发生了一些媒体感兴趣的事情。埃隆-马斯克在短短两年后就离开了开放人工智能的领导层,据说是有一些分歧,他需要更多关注特斯拉和SpaceX。这似乎很熟悉。只是,在雇用有才华的研究人员方面存在竞争。 2017年6月,特斯拉失去了一位知名的人工智能研究员安德烈-卡尔帕西,新的捐助者被宣布,领导层被调整。


Let's move on to 2019 and, in our opinion, the most controversial change at Open AI, the restructuring into a limited partnership. This is where the open and open AI is not quite as open as it was announced to be in 2015. The research lab now wants to work with a limited scope for profit. So a nonprofit organization is now a pre profit company. The goal is to attract more investors to finance the research and give investors an incentive that their money is not completely wasted. We are talking about a so called profit model, where investor profits are half at 100 times their investment. There are no exact numbers for the cost of open AI but to get a reference point, we can look at the research lab DeepMind owned by alphabet, which is Google's parent company. 

让我们进入2019年,在我们看来,Open AI 最具有争议性的变化,即重组为有限合伙企业。这时,开放的人工智能并不像在2015年宣布的那样开放。该研究实验室现在想在有限的范围内进行盈利工作。因此,一个非营利性组织现在是一个预营利公司。目标是吸引更多的投资者为研究提供资金,并给投资者一个激励,使他们的钱不会完全被浪费。我们正在谈论一个所谓的盈利模式,投资者的利润是其投资的100倍的一半。Open AI 的成本没有确切的数字,但为了得到一个参考点,我们可以看看由字母表(即谷歌的母公司)拥有的研究实验室DeepMind。


In 2016, they had a pre tax loss of 164 million US dollars, and in 2017, it was even 368 million US dollars. Research in artificial intelligence is very expensive, especially with the model of Open AI, which tries to solve problems with a lot of computing resources. However, this newly founded Limited Partnership is regulated by a parent organization, which is again nonprofit. This is supposed to make the management of the company free from the influence of capital, at least that is the plan. 

2016年,他们的税前亏损为1.64亿美元,而在2017年,甚至达到3.68亿美元。人工智能的研究是非常昂贵的,尤其是Open AI的模式,它试图用大量的计算资源来解决问题。然而,这个新成立的有限合伙公司是由一个母体组织监管的,而这个母体组织又是非营利性的。这应该是为了使公司的管理不受资本的影响,至少计划是这样的。


A few months after the restructuring into an LP, Microsoft invested 1 billion US dollars in Open AI, which can generate a maximum of 100 billion US dollars. Some products from open AI are already available in Microsoft products, such as GPT 3 Co Pilot, a tool for automatic completion of source code based on the open AI codecs model. This model understands natural language and converts it into code and various programming languages such as Java, PHP, Typescript, Go and many others. It was only announced in 2021 that GPP 3 Co Pilot will be available in Visual Studio and Visual Studio code. With the open AI API, Users can access the company's models through a structured interview. This API has been turned into a commercial product in order to fund further research, rather than being released as open source due to the challenges of operating the models, and the potential benefit only to large companies.

在改组为LP的几个月后,微软向Open AI投资了10亿美元,它最多可以产生1000亿美元。来自 Open AI 的一些产品已经在微软的产品中出现,如GPT 3 Co Pilot,这是一个基于开放人工智能编解码模型的自动完成源代码的工具。这个模型能够理解自然语言,并将其转换为代码和各种编程语言,如Java、PHP、Typescript、Go和其他许多语言。2021年才宣布,GPP 3 Co Pilot将在 Visual Studio和Visual Studio 代码中提供。有了 Open AI API, 用户可以通过结构化的面试来访问公司的模型。这个API已经变成了一个商业产品,以资助进一步的研究,而不是作为开放源码发布,因为操作模型的挑战,以及可能只对大公司有益。


Using the API, it is possible to develop products using GP T 2 and the newer versions. One example of a product is Co Pilot, and the models have various other uses such as summarization, classification, and processing unstructured data. 

使用该API,可以使用GP T 2和较新的版本开发产品。产品的一个例子是Co Pilot,这些模型还有其他各种用途,如总结、分类和处理非结构化数据。


In 2021, Dolly was released, which allows users to generate various images based on a given text. In 2022 GPT a language model was released and is able to hold conversations and respond to commands and follow up questions based on context. We have also tried it and are very surprised how well it works. 

2021年,Dolly被发布,它允许用户根据给定的文本生成各种图像。2022年,GPT一个语言模型被发布,能够进行对话,并根据上下文对命令和后续问题做出反应。我们也试过了,非常惊讶它的效果。


After we have told you a little bit about the history of Open AI with a rather neutral eye, of course slightly subjective and with a slight sarcastic touch. We now want to talk about the general perception of Open AI and also express some criticism. We are not someone who tries to find a flaw everywhere by force, there are actually also good things in this world, and I think that Open AI has good intentions, but the best of them is not perfect. However, I also see that Open AI is a very controversial company. It is not uncommon for the company to face criticism, especially when it comes to the commercialization or the API.

在我们以相当中立的眼光告诉你一些关于 Open AI 的历史之后,当然是略带主观和略带讽刺的。我们现在想谈谈对 Open AI 的普遍看法,同时也表达一些批评意见。我们不是一个想强行到处找缺点的人,这个世界上其实也有好的东西,我认为 Open AI 有好的初衷,但再好的东西也不完美。然而,我也看到,Open AI是一个非常有争议的公司。该公司面临的批评并不少见,尤其是在商业化或API方面。


While Open AI has received a lot of praise for its contributions to the field of AI, it has also faced criticism, particularly in regards to its decision to not open source its models, and its close relationship with Microsoft. It is important to carefully consider these criticisms and ensure that the company is transparent and accountable in its actions. Overall, open AI has made significant contributions to the field of AI and has the potential to continue making a positive impact in the future. 

虽然 Open AI 因其对人工智能领域的贡献而获得了很多赞誉,但它也面临着批评,特别是关于其不开放模型源代码的决定,以及它与微软的密切关系。重要的是要仔细考虑这些批评,并确保该公司在其行动中是透明和负责任的。总的来说,Open AI 已经为人工智能领域做出了重大贡献,并有可能在未来继续产生积极影响。


Thank you for watching our video about open AI, we appreciate your interest in learning about this research institute and its work in the field of artificial intelligence. We hope you found the information we presented to be informative and thought provoking. Please don't forget to hit that subscribe button and give it a like. Your support helps us create even more content for you to enjoy. We appreciate you taking the time to watch and we hope to see you back here soon.

感谢你观看我们关于 Open AI 的视频,我们感谢你有兴趣了解这个研究机构和它在人工智能领域的工作。我们希望你能发现我们所介绍的信息是有价值的,并能引起思考。请不要忘记点击那个订阅按钮,并给它一个赞。你的支持有助于我们为你创造更多的内容,让你享受。我们感谢你花时间观看,我们希望很快在这里看到你。


Copyright © 2024 aigcdaily.cn  北京智识时代科技有限公司  版权所有  京ICP备2023006237号-1