


A few days ago, Musk, Yoshua Bengio and others jointly signed an open letter calling on all AI laboratories to immediately suspend the training of AI models more powerful than GPT-4. Now, someone wants to stop the released GPT-4.
This time it was the non-profit organization Center for Artificial Intelligence and Digital Policy (CAIDP) that stopped GPT-4. CAIDP requested the U.S. Federal Trade Commission (FTC) to investigate OpenAI and prohibit the company from further releasing GPT-4.
File address: https://cdn.arstechnica.net/wp-content/uploads/2023/03/CAIDP-FTC -Complaint-OpenAI-GPT-033023.pdf
CAIDP filed this application with the FTC because they believed that "the consumer product GPT-4 released by OpenAI is biased, Deceptive and posing a risk to privacy and public safety. The model's output cannot be proven or reproduced and is not independently evaluated before deployment."
CAIDP calls for independent oversight and evaluation of all commercial artificial intelligence products in the United States and ensures necessary "safeguards" are in place to protect consumers, businesses, and the commercial market.
Previously, the FTC has announced an artificial intelligence standard that "while promoting accountability, requires the use of artificial intelligence to be transparent, explainable, fair, and "Empirically reasonable", CAIDP stated that "OpenAI's GPT-4 does not meet any of these requirements."
It’s only been two weeks since the release of GPT-4, and people are already deeply divided over this type of powerful AI model. On the one hand, those who want to stop powerful models such as GPT-4 believe that these models will pose greater risks to information security and even human society; on the other hand, some people believe that this is a good time for artificial intelligence to flourish. Accelerated technological progress should be promoted.
Interestingly, OpenAI CEO Sam Altman posted a new tweet: "Stay calm in the center of the storm." This may be his response to the recent series of "suspended GPT-4, etc." A response to the call for “model research”.
In terms of risk, OpenAI said at the time of the release that it had asked external experts to assess the potential risks posed by GPT-4. However, CAIDP made it clear in its filing to the FTC which acts GPT-4 violated.
CAIDP believes that GPT-4 has seriously affected commercial fairness, saying: "The commercial release of GPT-4 violates Section 5 of the FTC Act and the FTC's regulations on the use and advertising of artificial intelligence products. Standards set by enterprises, as well as emerging norms for artificial intelligence governance, etc."
In addition, OpenAI did not disclose any technical details of GPT-4. This is one of the reasons why CAIDP filed an application with the FTC.
CAIDP said: "OpenAI did not disclose details about the architecture, model size, hardware, computing resources, training techniques, dataset construction or training methods, and it is common practice in the research community to document Training data and training technology for large language models, but OpenAI chose not to do these things for GPT-4. In particular, generative artificial intelligence models are not ordinary consumer products, because they may exhibit some abnormal behaviors during use, These behaviors may not have been discovered by the issuing company before."
Complaint details
Specifically, CAIDP targeted GPT-4 and A related model, ChatGPT, points out a range of potential risks.
For example, OpenAI has made it clear in the article "GPT-4 System Card" that GPT-4 may reinforce and reproduce specific biases and worldviews, including against certain marginalized groups. of stereotypes and derogatory associations. CAIDP also cited a blog post from the company OpenAI, which said that a similar large model, ChatGPT, sometimes responded to harmful instructions or exhibited biased behavior.
In the complaint submitted to the FTC, CAIDP stated that "OpenAI released GPT-4 to the public for commercial use despite fully understanding the risks." The complaint also alleges that "the GPT-4 System Card does not provide details about the security checks OpenAI conducted during its testing, nor does it detail any steps OpenAI takes to protect children, which also raises questions about the use of GPT-4 on children." Concerns."
CAIDP also pointed out concerns raised by the European consumer organization BEUC: "If ChatGPT is used for consumer credit or insurance scores, is it possible that it will produce inappropriate results? Fair and biased outcome?" This tweet was also cited in CAIDP's complaint application.
In addition, in terms of network security, ChatGPT can be used for phishing, creating fake texts, or generating malicious code. On the privacy front, CAIDP said there were reports this month that OpenAI displayed private chats to other users.
In another case, an AI researcher described how ChatGPT could take over someone's account, view chat history, and access their bill without them realizing it information. However, OpenAI fixed the vulnerability.
CAIDP also stated that GPT-4 can provide text responses from image inputs, a feature that has a huge impact on personal privacy and personal autonomy, and allows users to convert personal images to linked to detailed personal data. It is understood that OpenAI has suspended the image-to-text function, but it is difficult to say what the actual situation is.
CAIDP believes that the FTC should prohibit OpenAI from further commercial deployment of GPT, require independent evaluation of GPT products before deployment and throughout the GPT AI life cycle, require OpenAI to comply with FTC AI standards, and Establish a publicly accessible incident reporting mechanism for GPT-4, similar to the FTC's mechanism for reporting consumer fraud.
CAIDP also urged the FTC to further publish standard specifications to serve as "basic standards for products in the generative artificial intelligence market area."
AI academic community, a debate
Thousands of people signed a petition in the past two days to suspend the development of GPT-4 follow-up AI large models. Today, CAIDP asked the FTC to investigate OpenAI’s ban Release a new commercial version of GPT-4. In just one or two days, various discussions have exploded, and AI bosses and various experts have come out to respond publicly. There are those who oppose it and those who support it.
In terms of suspending the research and development of GPT-4 follow-up AI large models, Turing Award winner Yoshua Bengio, Tesla CEO (OpenAI co-founder) Elon Musk, New York University Emeritus Professor Gary Marcus and UC Berkeley Professor Stuart Russell are all in favor. They have signed the open letter to suspend giant AI experiments. Notably, Marc Rotenberg, president and founder of CAIDP, also signed the open letter.
Open letter address: https://futureoflife.org/open-letter/pause-giant-ai-experiments/
#However, Yann LeCun, who has been criticizing ChatGPT, publicly stated that he would not sign the open letter and disagreed with the content of the open letter.
Thomas G. Dietterich, professor emeritus at Oregon State University, said on Twitter, "I didn't sign it either. This letter is filled with a bunch of scary stuff. Rhetoric and ineffective or non-existent policies. There are important technical and policy issues that are being worked on." LeCun publicly stated "I agree."
Ng Enda also issued an article publicly opposing the thousands of people who signed the petition, and said: "GPT-4 has many new applications in education, health care, food, etc., and will help many people. Unless the government intervenes, the suspension will be implemented And preventing all teams from scaling LLM is unrealistic. Furthermore, asking governments to suspend emerging technologies they don’t understand is anti-competitive, sets a bad precedent, and is bad policy.”
Tian Yuandong later also supported Ng Enda’s view, “stated that he would not sign a moratorium. And said that once this kind of thing starts, there is no way to stop or reverse this trend. This It is the inevitable necessity of evolution. We should look forward from a different perspective, better understand LLM, adapt to it and utilize its power, and feel the heat."
# #Yi Tay (formerly a senior researcher at Google Brain), who just announced his departure from Google Brain, said: "If people who discuss LLM randomly on the Internet are banned for 6 months, I will sign."
The above is the detailed content of GPT-4 was asked to be banned due to complaints: OpenAI does not meet any of the artificial intelligence standards issued by the FTC. For more information, please follow other related articles on the PHP Chinese website!

机器学习是一个不断发展的学科,一直在创造新的想法和技术。本文罗列了2023年机器学习的十大概念和技术。 本文罗列了2023年机器学习的十大概念和技术。2023年机器学习的十大概念和技术是一个教计算机从数据中学习的过程,无需明确的编程。机器学习是一个不断发展的学科,一直在创造新的想法和技术。为了保持领先,数据科学家应该关注其中一些网站,以跟上最新的发展。这将有助于了解机器学习中的技术如何在实践中使用,并为自己的业务或工作领域中的可能应用提供想法。2023年机器学习的十大概念和技术:1. 深度神经网

实现自我完善的过程是“机器学习”。机器学习是人工智能核心,是使计算机具有智能的根本途径;它使计算机能模拟人的学习行为,自动地通过学习来获取知识和技能,不断改善性能,实现自我完善。机器学习主要研究三方面问题:1、学习机理,人类获取知识、技能和抽象概念的天赋能力;2、学习方法,对生物学习机理进行简化的基础上,用计算的方法进行再现;3、学习系统,能够在一定程度上实现机器学习的系统。

本文将详细介绍用来提高机器学习效果的最常见的超参数优化方法。 译者 | 朱先忠审校 | 孙淑娟简介通常,在尝试改进机器学习模型时,人们首先想到的解决方案是添加更多的训练数据。额外的数据通常是有帮助(在某些情况下除外)的,但生成高质量的数据可能非常昂贵。通过使用现有数据获得最佳模型性能,超参数优化可以节省我们的时间和资源。顾名思义,超参数优化是为机器学习模型确定最佳超参数组合以满足优化函数(即,给定研究中的数据集,最大化模型的性能)的过程。换句话说,每个模型都会提供多个有关选项的调整“按钮

截至3月20日的数据显示,自微软2月7日推出其人工智能版本以来,必应搜索引擎的页面访问量增加了15.8%,而Alphabet旗下的谷歌搜索引擎则下降了近1%。 3月23日消息,外媒报道称,分析公司Similarweb的数据显示,在整合了OpenAI的技术后,微软旗下的必应在页面访问量方面实现了更多的增长。截至3月20日的数据显示,自微软2月7日推出其人工智能版本以来,必应搜索引擎的页面访问量增加了15.8%,而Alphabet旗下的谷歌搜索引擎则下降了近1%。这些数据是微软在与谷歌争夺生

荣耀的人工智能助手叫“YOYO”,也即悠悠;YOYO除了能够实现语音操控等基本功能之外,还拥有智慧视觉、智慧识屏、情景智能、智慧搜索等功能,可以在系统设置页面中的智慧助手里进行相关的设置。

人工智能在教育领域的应用主要有个性化学习、虚拟导师、教育机器人和场景式教育。人工智能在教育领域的应用目前还处于早期探索阶段,但是潜力却是巨大的。

阅读论文可以说是我们的日常工作之一,论文的数量太多,我们如何快速阅读归纳呢?自从ChatGPT出现以后,有很多阅读论文的服务可以使用。其实使用ChatGPT API非常简单,我们只用30行python代码就可以在本地搭建一个自己的应用。 阅读论文可以说是我们的日常工作之一,论文的数量太多,我们如何快速阅读归纳呢?自从ChatGPT出现以后,有很多阅读论文的服务可以使用。其实使用ChatGPT API非常简单,我们只用30行python代码就可以在本地搭建一个自己的应用。使用 Python 和 C

人工智能在生活中的应用有:1、虚拟个人助理,使用者可通过声控、文字输入的方式,来完成一些日常生活的小事;2、语音评测,利用云计算技术,将自动口语评测服务放在云端,并开放API接口供客户远程使用;3、无人汽车,主要依靠车内的以计算机系统为主的智能驾驶仪来实现无人驾驶的目标;4、天气预测,通过手机GPRS系统,定位到用户所处的位置,在利用算法,对覆盖全国的雷达图进行数据分析并预测。


Hot AI Tools

Undresser.AI Undress
AI-powered app for creating realistic nude photos

AI Clothes Remover
Online AI tool for removing clothes from photos.

Undress AI Tool
Undress images for free

Clothoff.io
AI clothes remover

AI Hentai Generator
Generate AI Hentai for free.

Hot Article

Hot Tools

ZendStudio 13.5.1 Mac
Powerful PHP integrated development environment

SAP NetWeaver Server Adapter for Eclipse
Integrate Eclipse with SAP NetWeaver application server.

EditPlus Chinese cracked version
Small size, syntax highlighting, does not support code prompt function

DVWA
Damn Vulnerable Web App (DVWA) is a PHP/MySQL web application that is very vulnerable. Its main goals are to be an aid for security professionals to test their skills and tools in a legal environment, to help web developers better understand the process of securing web applications, and to help teachers/students teach/learn in a classroom environment Web application security. The goal of DVWA is to practice some of the most common web vulnerabilities through a simple and straightforward interface, with varying degrees of difficulty. Please note that this software

Atom editor mac version download
The most popular open source editor
