Microsoft Bolsters AI Capabilities with New Bing Features

TL;DR:

  • Microsoft is focusing on AI to expand Bing’s capabilities.
  • Bing Chat has gained popularity, with over 100 million daily active users engaging in chats and generating images.
  • Bing Chat will become more visual, incorporating relevant images into responses.
  • Privacy and safety measures are in place, including filtering mechanisms to prevent explicit content.
  • Bing Chat faced challenges during the preview release, but progress has been made in addressing issues.
  • Bing Chat will provide citations for fact-based responses, enhancing transparency.
  • Bing Chat will generate charts and graphs based on prompts and data.
  • Bing Image Creator will support multiple languages and allow image customization.
  • Legal issues have emerged regarding generative art AI, but Microsoft has not committed to compensation measures.
  • Bing Chat will understand and respond to images, enabling image-based searches.
  • Edge browser receives a fresh design aligned with Windows 11 aesthetics.
  • Compose feature in Edge’s Bing Chat offers adjustable text generation.
  • Edge introduces actions triggered by Bing Chat prompts, such as opening settings or organizing tabs.
  • Microsoft hints at expanding its actions to other products and services.
  • Bing Chat integration with Edge enhances browsing and productivity.
  • Bing Chat offers chat history storage, export, and sharing functionalities.
  • Bing Chat introduces plugins from partners like OpenTable and Wolfram Alpha, extending its capabilities.
  • Microsoft is committed to continuous improvement and innovation in Bing Chat and Edge.

Main AI News:

Microsoft is gearing up for the next chapter in Bing’s expansion, and it comes as no surprise that artificial intelligence (AI) is at the forefront of its strategy. During an exclusive preview event held in New York City, top executives, including Yusuf Mehdi, the Chief Vice President and Chief Marketing Officer for consumers, provided a glimpse into the upcoming features that will shape Bing’s future in the coming days, weeks, and months.

Rather than reinventing the wheel, Microsoft is building upon the AI advancements that have been infused into the Bing experience over the past three months. Since the introduction of Bing Chat, an AI-powered chatbot driven by OpenAI’s GPT-4 and DALL-E 2 models, Microsoft proudly reports that Bing has amassed over 100 million daily active users. Furthermore, these users have engaged in over half a billion chats and have generated a staggering 200 million images, showcasing the immense impact of AI integration.

Looking forward, Bing will undergo a visual transformation, embracing a more image- and graphic-centric approach within Bing Chat. Users can expect a more personalized experience as Bing introduces features that allow them to export their chat histories and integrate content from third-party plugins, all while ensuring data security and privacy. Additionally, Bing Chat will tap into multimodality, expertly answering questions within the context of images, providing users with a seamless and comprehensive search experience.

Yusuf Mehdi expressed his optimism about the future of search, stating, “I think it’s safe to say that we’re underway with the transformation of search. In our minds, we think that today will be the start of the next generation of this ‘search mission.” Microsoft’s commitment to AI-driven innovation positions Bing as a trailblazer in the evolving landscape of search engines, revolutionizing how users interact with and extract information from the web. As Bing’s expansion continues, users can expect an even more intuitive and tailored search experience powered by the boundless potential of AI technology.

Bing Chat: Open to All and Enhanced Visual Experience

Exciting news for search enthusiasts! The new and improved Bing, featuring Bing Chat, is now accessible to all users without the need for a waitlist. Experience the power of Bing Chat by simply signing in with your Microsoft Account, and prepare to embark on a dynamic and visual search journey.

The latest version of Bing Chat builds upon the groundwork laid out several months ago, delivering an enhanced user experience. As mentioned earlier, Bing Chat will soon revolutionize search responses by incorporating relevant images where appropriate. Imagine asking a question like, “Where is Machu Picchu?” and receiving not only a concise answer but also a captivating image, all presented in a condensed, card-like interface.

During a compelling demonstration at the preview event, a spokesperson posed the query, “Does the saguaro cactus grow flowers?” In response, Bing Chat effortlessly provided a detailed paragraph alongside a vivid image of the cactus in question. This seamless integration of information and visuals conjured memories of Google Search’s renowned “knowledge panels.”

Microsoft has implemented robust filtering mechanisms to ensure a safe and inclusive experience within Bing Chat. While the specific categories of content that trigger accompanying images remain undisclosed, the platform takes precautions to prevent explicit images from surfacing. Sarah Bird, Microsoft’s Head of Responsible AI, revealed that Bing Chat leverages a combination of “toxicity classifiers” and blacklists, AI models trained to detect potentially harmful prompts, to maintain a clean chat environment.

It is important to acknowledge that Bing Chat faced some challenges during its initial preview release in February. Instances were reported where the chatbot disseminated vaccine misinformation and expressed offensive viewpoints, including a hateful diatribe attributed to Adolf Hitler. In a related development, Microsoft recently disbanded its ethics and society team within the broader AI organization, raising concerns about the integration of AI principles into product design.

Bird, however, assures us that substantial progress has been made, acknowledging that addressing complex AI issues requires time and dedication. While Bing Chat enjoys broad public availability, a team of human moderators diligently monitors for abuse and inappropriate usage, such as attempts to exploit the platform for phishing emails.

As the press was limited to curated demonstrations at the event, we await broader user feedback to gauge the impact of these measures. Microsoft’s commitment to continuous improvement and user safety will undoubtedly be put to the test once more individuals have the opportunity to explore Bing Chat firsthand.

One noteworthy enhancement in Bing Chat is the increased transparency of its fact-based responses. Soon, when you request a summary or inquire about the contents of a document, whether it’s a comprehensive 20-page PDF or a Wikipedia article, Bing Chat will include citations indicating the sources of the information within the text. By clicking on these citations, users will be directed to the corresponding passages, fostering a more reliable and informative search experience.

Bing Chat: Evolving into a Productivity Powerhouse

In its relentless pursuit of innovation, Bing Chat is poised to take a giant leap forward in the realm of productivity. Among its upcoming features, Bing Chat will soon boast the ability to generate charts and graphs based on specific prompts and data. Gone are the days of receiving a basic list of results when querying about the most populous cities in Brazil. In the near future, Bing Chat will present this information visually in a user-customizable chart format, elevating the search experience to new heights.

This foray into visual data representation signals Bing’s ambition to transform into a comprehensive productivity platform. When combined with the forthcoming enhanced text-to-image generation capabilities, Bing Chat becomes an invaluable tool for users seeking efficient data visualization and information comprehension.

In the weeks to come, Bing Image Creator, powered by the remarkable DALL-E 2, will expand its language support beyond English to over 100 total languages. Users, irrespective of their language preference, will have the ability to generate images based on textual prompts.

Furthermore, users can refine and customize these generated images by providing follow-up prompts. For instance, requesting an image of a bunny rabbit and subsequently instructing the system to make the fur pink will yield a personalized and visually captivating result.

Generative art AI has recently garnered attention, albeit not always for the most optimistic reasons. Legal disputes have arisen, with plaintiffs filing lawsuits against OpenAI and other vendors, alleging unauthorized use of copyrighted data, primarily artistic creations, to train generative models such as DALL-E 2. These generative models learn to create art and other content by training on a vast array of images and text, often indiscriminately sourced from the public web.

To shed light on Microsoft’s stance, I raised the question of whether the company is exploring avenues to compensate creators whose work might have been utilized in training data, even if fair use is the prevailing legal argument. Several platforms that have launched generative AI tools, such as Shutterstock, have taken proactive steps by initiating creator funds to address these concerns. Additionally, platforms like Spawning are developing mechanisms that allow artists to opt out of AI model training altogether.

Sarah Bird acknowledged the importance of addressing these issues and the need to find a resolution that acknowledges and rewards content creators. However, at present, Microsoft has not committed to any specific course of action.

Multimodal Capabilities Transforming Bing Chat’s Image Understanding

Bing Chat continues to push boundaries with its cutting-edge multimodal capabilities. In addition to comprehending text, Bing Chat will soon be able to understand images, bringing a new dimension to the search experience. Users will have the ability to upload images and initiate web searches for related content. For instance, by uploading a picture of a crocheted octopus and asking Bing Chat, “How do I make that?” users will receive step-by-step instructions to create a similar masterpiece.

Furthermore, multimodality extends to the Edge app for mobile devices, where the new page context function will enable users to ask Bing Chat questions related to the web pages they are viewing on their mobile devices. This seamless integration of text and image understanding enhances the overall user experience and enables a deeper level of interaction with search results.

Although Microsoft has not explicitly confirmed the source of these impressive multimodal abilities, it is highly likely that they are rooted in GPT-4. OpenAI, the developer behind GPT-4, has not yet made the model’s image understanding capabilities available to all customers. However, given Microsoft’s significant investment in and close collaboration with OpenAI, it is plausible that they have privileged access to these advanced features.

It is crucial to acknowledge the potential for misuse that comes with any image upload tool. Microsoft addresses this concern by implementing automated filtering and hashing mechanisms to prevent illicit uploads. Sarah Bird affirms the company’s commitment to safeguarding user experiences, although the effectiveness of these measures remains to be independently tested.

Expanding Capabilities: Chat History, Export, Share, and Plugins for Bing Chat

Bing Chat is set to unveil a host of new features that will enhance the user experience and provide increased functionality. One notable addition is the introduction of chat history, allowing users to store their previous conversations and seamlessly pick up where they left off. Similar to the chat history feature recently introduced by OpenAI’s ChatGPT, Bing Chat will display a list of chats alongside the bot’s responses, enabling users to easily navigate and review past interactions.

While the exact duration for which chat histories will be stored is yet to be determined, Microsoft affirms that users will have the option to delete their chat history at any time, addressing concerns raised by several European Union governments regarding privacy.

Bing Chat will also offer export and share functionalities, enabling users to share conversations on social media platforms or export them to a Word document for future reference. Additionally, Microsoft’s web experiences team is actively working on a more robust copy-and-paste system specifically designed for graphs and images generated through Bing Chat. Although this feature is not yet available in the preview stage, it holds promise for users seeking to seamlessly incorporate visual content into their work.

The introduction of plugins marks a transformative addition to Bing Chat, expanding its capabilities through partnerships with esteemed organizations such as OpenTable and Wolfram Alpha. These plugins significantly extend Bing Chat’s functionalities, allowing users to accomplish tasks such as booking reservations or obtaining answers to complex science and math questions. While the plugin functionality is still in the early stages of development, with no dedicated marketplace established yet, users will have the ability to toggle plugins on or off from the Bing Chat web interface.

Dena Saunders, a partner General Manager in Microsoft’s web experiences team, hinted at a potential association between the Bing Chat plugins scheme and OpenAI’s recently introduced plugins for ChatGPT, although no definitive confirmation was provided. Given the similarities between the two initiatives, such a connection would be logical and further enhance the collaboration between Microsoft and OpenAI.

Conlcusion:

The advancements and new features introduced in Bing Chat and its integration with Edge signify a significant step forward in the search engine market. By leveraging the power of AI, Bing is positioning itself as a transformative force in search, offering users a more visual, personalized, and productive experience. The introduction of multimodal capabilities, enhanced visualizations, chat history, export and share functionalities, and plugins showcases Microsoft’s commitment to continuous innovation and meeting user needs.

As Bing evolves into a comprehensive productivity platform, it presents opportunities for businesses to leverage its capabilities for enhanced customer interactions, data visualization, and streamlined workflows. The integration of AI-driven features in Bing Chat and Edge is set to reshape the search engine landscape, providing a dynamic and immersive search experience for users and unlocking new possibilities for businesses in various industries.

Source