Goto

Collaborating Authors

 new bing


Multi-step Jailbreaking Privacy Attacks on ChatGPT

Li, Haoran, Guo, Dadi, Fan, Wei, Xu, Mingshi, Huang, Jie, Meng, Fanpu, Song, Yangqiu

arXiv.org Artificial Intelligence

With the rapid progress of large language models (LLMs), many downstream NLP tasks can be well solved given appropriate prompts. Though model developers and researchers work hard on dialog safety to avoid generating harmful content from LLMs, it is still challenging to steer AI-generated content (AIGC) for the human good. As powerful LLMs are devouring existing text data from various domains (e.g., GPT-3 is trained on 45TB texts), it is natural to doubt whether the private information is included in the training data and what privacy threats can these LLMs and their downstream applications bring. In this paper, we study the privacy threats from OpenAI's ChatGPT and the New Bing enhanced by ChatGPT and show that application-integrated LLMs may cause new privacy threats. To this end, we conduct extensive experiments to support our claims and discuss LLMs' privacy implications.


Search-in-the-Chain: Towards Accurate, Credible and Traceable Large Language Models for Knowledge-intensive Tasks

Xu, Shicheng, Pang, Liang, Shen, Huawei, Cheng, Xueqi, Chua, Tat-Seng

arXiv.org Artificial Intelligence

Making the contents generated by Large Language Model (LLM) such as ChatGPT, accurate, credible and traceable is crucial, especially in complex knowledge-intensive tasks that require multi-step reasoning and each of which needs knowledge to solve. Introducing Information Retrieval (IR) to provide LLM with external knowledge is good potential to solve this problem. However, where and how to introduce IR into LLM is a big challenge. Previous work has the disadvantage that the wrong knowledge retrieved by IR misleads the LLM or breaks the reasoning chain of LLM. In this paper, we propose a novel framework called Search-in-the-Chain (SearChain) for the interaction between LLM and IR to solve the challenges. First, LLM generates the global reasoning chain called Chain-of-Query (CoQ) where each node consists of an IR-oriented query and the answer to the query. Second, IR verifies the answer of each node of CoQ, it corrects the answer that is not consistent with the retrieved information when IR gives high confidence, which improves the credibility. Third, LLM can mark its missing knowledge in CoQ and IR can provide this knowledge to LLM. These three operations improve the accuracy of LLM for complex knowledge-intensive tasks in terms of reasoning ability and knowledge. Finally, SearChain generates the reasoning process and marks references to supporting documents for each reasoning step, which improves traceability. SearChain transforms the topology of reasoning from chain to tree, which can modify the reasoning direction. Experiment shows that SearChain outperforms baselines on complex knowledge-intensive tasks including multi-hop question-answering, slot filling, fact checking, and long-form question-answering.


Easily access the new AI-powered Bing across your favorite mobile apps

#artificialintelligence

Bing recently hit 100M daily users (and 100M chats)! Today, we're excited to share new AI-powered experiences that extend these capabilities to millions of additional people across devices and around the globe! In recent weeks, we've added a variety of new ways to access and interact with the new Bing. Today, we are announcing yet another, with powerful updates to SwiftKey that put the Bing AI experience one touch away across any iOS or Android mobile experience that supports a third-party keyboard. An updated SwiftKey represents a growing set of access points and improvements to Bing experiences, including new updates to existing app integrations spanning Bing, Skype, Microsoft Start, and Microsoft Edge apps.


ChatGPT creator launches even more powerful GPT-4

#artificialintelligence

OpenAI, the creator of the wildly popular ChatGPT, has launched a new and even more powerful AI bot, GPT-4 -- and admits it's so advanced it could be'harmful'. The bot can now accept inputs in the form of images as well as text, but still outputs its answers in text, meaning it can offer detailed descriptions of images. If asked, 'What's funny about this image?' Among other things, it can instantly calculate people's tax liability and can pass the Uniform Bar Exam. OpenAI said in a blog post: 'We've created GPT-4, the latest milestone in OpenAI's effort in scaling up deep learning.


Bing now features an AI image generator -- here's how to use it

#artificialintelligence

Following on from the integration of ChatGPT into the Bing search engine, Microsoft have now followed up by integrating another of OpenAI's products: the AI image generator DALL-E 2. It's fair to say that rolling out the "new Bing" with its ChatGPT-powered AI chat functionality was a whopping success for Microsoft. Now, says Microsoft Corporate VP Yusuf Mehdi in a blog post (opens in new tab), the tech giant are "taking the chat experience to the next level by making the new Bing more visual." What that boils down to is utilizing the OpenAI's AI image generator, DALL-E 2, to form the Bing Image Creator. Essentially, instead of using DALL-E 2 to generate images on its own website, you can type prompts into the Bing search engine to receive AI generated imagery from there using the same engine. So how do you use the Bing Image Creator?


Microsoft begins making Bing Chat AI searches available to everyone

Engadget

Microsoft launched its ChatGPT-powered version of Bing last month in a limited beta, and it promptly brought a bunch of new viewers and some respect to the beleaguered search engine. Now, it appears that Microsoft has opened up the new Bing to nearly everyone who wants to use it, as Windows Central has noticed. While the signup page still says "join the waiting list," all you have to do is sign in to get instant access -- a trick that worked for myself and a colleague. Microsoft has yet to confirm the change, but we may learn more at an event it's holding today called "Reinventing productivity with AI," as spotted by TechCrunch. The company is supposed to be introducing AI-powered tools for its Microsoft 365 suite and SalesForce rival Dynamic 365, but it may announce Bing changes as well.


Bing said to remove waitlist for its GPT-4 powered chat

#artificialintelligence

Microsoft's Bing is enjoying the spotlight for the first time in a decade after it released a GPT-powered interface last month. But the tech giant has so far been cautious about the pace at which it is making the new Bing offering -- powered by OpenAI's GPT-4 tech -- available to users. But it appears, Bing is bringing those walls down. Microsoft, a major investor in OpenAI, appears to have lifted the waitlist from the new Bing, ostensibly allowing anyone to gain instant access to the new experience. Windows Central, which first spotted this change, said users don't have to wait to try out the new Bing anymore.


Microsoft's new Bing was using GPT-4 all along

#artificialintelligence

When Microsoft announced the new AI-enabled Bing, which it built on top of OpenAI's GPT models, neither company would confirm which version of GPT was being used beyond saying it was a next-gen version of the model that powered ChatGPT. Today, OpenAI announced GPT-4, a significant update to GPT-3.5. As it turns out, Bing was using it all along. "We are happy to confirm that the new Bing is running on GPT-4, which we've customized for search," Microsoft's Yusuf Mehdi, the company's corporate VP and consumer chief marketing officer, wrote in today's announcement. "If you've used the new Bing preview at any time in the last five weeks, you've already experienced an early version of this powerful model."


ChatGPT 2.0: Creator of AI bot that took world by storm launches even more powerful version

Daily Mail - Science & tech

Open AI, the creator of the wildly popular ChatGPT, has launched a new and even more powerful AI bot, GPT-4 -- and admits it's so advanced it could be'harmful'. The bot can now accept inputs in the form of images as well as text, but still outputs its answers in text, meaning it can offer detailed descriptions of images. If asked, 'What's funny about this image?' it will reply. Among other things, it can instantly calculate people's tax liability and can pass the Uniform Bar Exam. OpenAI said in a blog post: 'We've created GPT-4, the latest milestone in OpenAI's effort in scaling up deep learning.


Microsoft says Bing has crossed 100 million daily active users

Engadget

Bing has crossed 100 million daily active users a month after the launch of its chatbot AI, according to Yusuf Mehdi, Microsoft's VP for Modern Life, Search and Devices. He said the company is fully aware that it's still just "a small, low, single digit share player," but hey, there was a time when Bing wasn't even a part of the conversation. Now, after the tech giant released its next-gen version, even those who haven't used it in the past are relying on it for their searches: Mehdi noted that one-third of Bing's daily active users are new to the search engine. "We see this appeal of the new Bing as a validation of our view that search is due for a reinvention and of the unique value proposition of combining Search Answers Chat Creation in one experience," the VP said. In addition to seeing a boost in numbers, Microsoft is also apparently enjoying a growth in engagement, with more people conducting more searches. The company credits two factors for that particular victory, the first being Edge's growth in usage, most likely aided by the addition of Bing's chat AI as a new feature.