• Work With AI
  • Posts
  • Microsoft Will Introduce AI Office Products Soon

Microsoft Will Introduce AI Office Products Soon

02-13-2023

⚡ Today’s Highlights

  • 📰 News: Microsoft will demo AI Office products soon and AIs are hallucinating

  • 💰 Funding: Automating business tasks and funding to improve critical care at University of Florida

  • 🦾 Tools: AI for crypto, Amazon Marketplace, and classifying waste

  • 🌎 Content: Linkin Park’s AI music video

  • 📅 Events: Jasper AI’s Gen AI Conference is tomorrow and online registration closes TONIGHT

📰 Today's Top Stories

(4 min read) (Source: The Verge)

TLDR: Microsoft is preparing to demonstrate its expansion plans for its core productivity applications, such as Word, PowerPoint, and Outlook. According to an inside source, the company is planning an announcement for March to further showcase its revolutionary updates to its search and productivity apps through its partnership in OpenAI.

The Big Picture: The integration of OpenAI tools into the Office suite of products will probably be one of the first ways that AI is widely introduced into the workplace. Data management and information processing is about to get a whole lot fancier.

(8 min read) (Source: WIRED)

TLDR: Microsoft has started testing a new chatbot interface for Bing that can sometimes provide access to news websites' content, bypassing their paywalls. This has raised concerns from media companies claiming that tech companies like Microsoft have not been paying enough for the news content used on search engines like Bing and Google. Microsoft President Brad Smith had previously testified in a US congressional hearing, saying that tech companies should do more to support journalism and that Microsoft was committed to continuing "healthy revenue-sharing" with news publishers.

However, the new Bing chatbot, based on technology from OpenAI's ChatGPT, provides answers that draw on media content without paying for it. This has prompted publishers to consider what kind of actions need to be taken against tech giants, like Google and Microsoft, that are positioned to dominate not just the search engine market, but also where information is coming from. Trade groups are also concerned about the potential for the proliferation of misinformation through search chatbots and the influence on the accountability role of journalism and media in democracies.

The Big Picture: The biases that have been very evident in the mainstream chatbots being used currently may soon play a much more serious role in the dissemination of information as they become the popular tool for searching. It is quickly becoming more important for legislation to be implemented to facilitate how these search engine chatbots are providing answers to the public’s questions, particularly those related to sensitive subjects.

(2 min read) (Source: Reuters)

TLDR: A few days after the Google release of its own AI powered chatbot fell flat on its face due to a factual error, Prabhakar Raghavan head of Google Search, warned customers of AI-powered chatbots like Bard and ChatGPT “hallucinating”.

While this may sound like a creepy phenomenon, you’ve probably already encountered examples of this everywhere. The term ‘hallucination’ in this context is a confident response by AI-tools that does not seem to be justified by the tool’s training data. There are many reasons for an AI to hallucinate. For example, errors in encoding and decoding between text and representations can cause hallucinations. Training these tools to produce diverse responses can also lead to hallucination.

The Big Picture: This is different from a factual gap in the data or an incomplete/outdated dataset. ChatGPT is trained only up to data from 2021. So a few days ago, when Lebron James became the leading scorer in the NBA, it took ChatGPT some time to learn that the answer it was providing (Kareem Abdul-Jabbar) had just recently been outdated. Either way, everyone developing these tools needs to do their best to minimize these hallucinations. A hallucination can lead to misinformation and in high-stakes, real-world scenarios, a fatal error in judgment.

(7 min read) (Source: NASA)

TLDR: NASA is using AI-designed spacecraft and mission hardware that look like alien exoskeletons but are lighter, stronger, and designed faster when compared to those designed by humans. These new designs can save two-thirds of the weight compared to regular components and can be easily produced by commercial mills.

(8 min read) (Source: MIT News)

TLDR: Artificial Intelligence researchers have recently learned more about a phenomenon called "in-context learning" in which large language models can perform new tasks without being explicitly retrained for that task. This is possible because these models, like GPT-3, are trained on a vast amount of data which includes text from billions of websites. When a new task is presented to the model, it recognizes patterns it has seen before during its training process and is able to perform the new task.

Scientists believe that this is because these large language models contain smaller machine-learning models within them that they can use to perform new tasks. The researchers have also theoretically proven that these models can write a new, smaller model within its many layers of programming. In this way, the model can simulate and train a smaller version of itself to perform the new task.

The Big Picture: In summary, in-context learning allows large language models to perform new tasks without explicit retraining, because they have the ability to recognize patterns they have seen before and train smaller models within themselves to complete the task.

This has big implications for how efficiently large language models like ChatGPT can operate and learn which not only improves performance but will reduce energy expenditure. The faster they can learn, the smarter they can get, and the more quickly they can adapt to changes to their models that need to be implemented immediately.

💰 Funding Alerts

  1. Rezolve.ai, a company offering an AI-based service desk for automating business tasks and processes related to employee support, IT, and HR, has secured $11 million in a Series A funding round. The funding round was led by SIG Venture Capital and included participation from Exfinity Venture Partners, 9Unicorns, and Tri Valley Ventures. The company plans to utilize the funds to drive its growth and enhance the capabilities of its platform to provide an unparalleled experience to employees.

  2. The National Institutes of Health's Bridge2AI program has awarded the University of Florida a sum of $3.6 million as part of a larger grant of $23.5 million to multiple institutions. The aim of the program is to establish an infrastructure for artificial intelligence in critical care and to enhance the application of AI in a manner that boosts patients' chances of recovery from severe illnesses.

🦾 Trending Tools

  • ChainGPT is an AI model that utilizes algorithms and high-performance computing to enable users to generate smart contracts, perform market analysis, manage risk, or create non-fungible tokens (NFTs).

  • Sellesta utilizes artificial intelligence to increase your sales on Amazon Marketplace with its GPT-similar tool. It provides comprehensive keyword research, quick listing optimizations, review analysis, and competition analysis.

  • Deep Waste, powered by machine learning, ensures that you are managing your waste in the most environmentally friendly manner possible. By simply taking a picture of the waste using their mobile device, users can quickly classify it into categories such as paper, plastic, glass, metal, and more. The app provides real-time waste identification, making it easier for you to manage your waste sustainably.

🌎 Popular Content

1. The music video for Linkin Park’s new song “Lost” was created using AI. They used original footage of the band along with art created by pplpleasr and Maciej Kuciara depicting the band as anime characters. The art and footage was then blended using Kaiber, an AI generation engine, to create the full anime music video. This comes a few months after Kaiber secured $6.9m in funding from the likes Andreessen Horowitz, Joe Tsai, Kevin Durant, and Paris Hilton. Enjoy the nostaliga. (YouTube)

    Mike Shinoda explains here. (Twitter)

    2. This AI somehow blended Dwayne Wade, Michael Jordan, Lebron James, Shaq, Bismack Biyombo, Shannon Sharpe, and Ronnie Coleman into one person. (Twitter)

    3. Say your prayers ChatGPT, once we hit Maruary it’s all over. (Twitter)

    4. This is what an educational AI tool should look like. It should be a research assistant, not a factory for boring essays that are full of made up “facts”. (Reddit)

    👀 More Reading

    🧠 Resources

    Understand the basics of reinforcement learning, a powerful modern machine learning technique, in this DeepMind lecture series by a pioneer in the field; David Silver.

     📅 Upcoming Events

    1. The Gen AI Conference (TOMORROW February 14, 2023. San Francisco, CA) hosted by Jasper AI, is the first-ever generative AI conference. Attendees can learn about the many recent developments in the field of AI from experts and network with like-minded individuals in AI, business, and marketing. The event’s sold out, but check here for more news!

    2. RE•WORK AI Summit West, Deep Learning Summit (February 15-16, 2023. San Francisco, CA), a chance to hear the latest technology advancements, practical examples of how to apply AI to solve challenges in cross-industry settings, business and society, and delve deeper into the work of leading AI experts in a series of presentations, panel discussions, interviews and fireside chats.

    3. Data Science Salon Austin (February 21-22, 2023. Austin, TX + Virtual) is a two-day 500-person conference focused on AI and machine learning applications in the enterprise. The intimate event curates data science sessions to bring industry leaders and specialists face-to-face to educate each other on innovative new solutions in artificial intelligence, machine learning, predictive analytics and acceptance around best practices.

     💼 Jobs

    That's all for todays email! If you want more please follow us at the social channels linked below, or check out our website!

    How'd you like today's email?

    Login or Subscribe to participate in polls.

    Share our newsletter: If you like our work please share/forward this email with your friends, colleagues, and family. It's the best way to support us!

    If this email was forwarded to you please sign up here to continue receiving them.

    Want your content, product, jobs, or event featured in our newsletter? Reply to this email with the details, and our team will reach out to you.

    Do you use AI for work? Tell us how, and you could be featured in our newsletter!

    Check out our website for more resources, including a list of AI investors, products, events, and twitter follows.

    For an archive of all our posts, click here.

    We'd love to hear from you! You can always leave us comments or feedback by replying to this email!

    Powered by AI. Curated and edited by Humans.