Microsoft Pushes Copilot Ahead of Super Bowl
Plus: DeepMind's 'self-discover' framework for LLMs, YouTube's plans to empower creativity with AI.
Hello Engineering Leaders and AI Enthusiasts!
Welcome to the 206th edition of The AI Edge newsletter. This edition brings you Microsoft’s plans and updates for Copilot before the Super Bowl.
And a huge shoutout to our amazing readers. We appreciate you😊
In today’s edition:
🦾 Microsoft pushes Copilot ahead of the Super Bowl
🧠 Deepmind presents ‘self-discover’ framework for LLMs improvement
🎥 YouTube reveals plans to use AI tools to empower human creativity
💳 Knowledge Nugget: Measuring the "readability" of texts with Large Language Models by
Let’s go!
Microsoft pushes Copilot ahead of the Super Bowl
In their latest blogs and Super Bowl commercial, Microsoft announced their intention to showcase the capabilities of Copilot exactly one year after their entry into the AI space with Bing Chat. They have announced updates to their Android and iOS applications to make the user interface more sleek and user-friendly, along with a carousel for follow-up prompts.
Microsoft also introduced new features to Designer in Copilot to take image generation a step further with the option to edit generated images using follow-up prompts. The customizations can be anything from highlighting the image subject to enhancing colors and modifying the background. For Copilot Pro users, additional features such as resizing the images and changing the aspect ratio are also available.
Why does this matter?
Copilot unifies the AI experience for users on all major platforms by enhancing the experience on mobile platforms and combining text and image generative abilities. Adding additional features to the image generation model greatly enhances the usability and accuracy of the final output for users.
Google Deepmind presents GPT-4 performance improvement using ‘self-discover’ framework
Google Deepmind, with the University of Southern California, has proposed a ‘self-discover’ prompting framework to enhance the performance of LLMs. Models such as GPT-4 and Google’s Palm 2 have witnessed a performance improvement on challenging reasoning benchmarks by 32% compared to the Chain of Thought (CoT) framework.
The framework works by identifying the reasoning technique intrinsic to the task and then proceeds to solve the task with the discovered technique ideal for the task. This framework also works with 10 to 40 times less inference computation, which means that the output will be generated faster using the same computational resources.
Why does this matter?
Improving the reasoning accuracy of an LLM is largely beneficial to users as they can achieve the desired output with fewer prompts and with greater accuracy. Moreover, reducing the inference directly translates to lower computational resource consumption, leading to lower operating costs for enterprises.
YouTube reveals plans to use AI tools to empower human creativity
YouTube CEO Neal Mohan revealed 4 new bets they have placed for 2024, with the first bet being on AI tools to empower human creativity on the platform. These AI tools include:
Dream Screen, which lets content creators generate custom backgrounds through AI with simple prompts of an idea.
Dream Track will allow content creators to generate custom music by just typing in the music theme and the artist they want to feature.
These new tools are mainly aimed to be used in YouTube Shorts and highlight a priority to move towards short-form content.
Why does this matter?
The democratization of AI tools for content creators allows them to offer better quality content to their viewers, which collectively boosts the quality of engagement on the platform. This also lowers the bar to entry for many aspiring artists and lets them create quality content without the added difficulty of generating custom video assets.
Enjoying the daily updates?
Refer your pals to subscribe to our daily newsletter and get exclusive access to 400+ game-changing AI tools.
When you use the referral link above or the “Share” button on any post, you'll get the credit for any new subscribers. All you need to do is send the link via text or email or share it on social media with friends.
Knowledge nugget: Measuring the "readability" of texts with Large Language Models
Is a LLM truly better at evaluating the readability of text than human evaluators? This is what
aimed to figure out with his article, where he put an LLM against other readability formulas and human readability ratings to find out the answer. He used GPT-4 Turbo to rate the readability of different content pieces on a scale of 1-100 and compared them against the CLEAR corpus of readability to see how accurate the ratings were.In his findings, he states that not only were the results of GPT-4 Turbo highly correlated with the human ratings, but it outperformed all the other predictors he tested. Furthermore, Sean states that GPT-4 Turbo wasn’t performing to its fullest potential, as the results could have been more accurate if the model was fine-tuned with the relevant data.
Why does this matter?
Readability plays a vital role in the editorial process of any documentation and has a direct impact on how easily readers can grasp the subject matter being discussed. However, traditional readability formulae and human interpretations have the chance of being subjective. This is why an AI model that can accurately gauge the readability of content can be pivotal in making documentation easier.
What else is happening❗
🧑🤝🧑 OpenAI forms a new team for child safety research.
OpenAI revealed the existence of a child safety team through their careers page, where they had open positions for a child safety enforcement specialist. The team will study and review AI-generated content for “sensitive content” to ensure that the generated content aligns with their platform policy. This is to prevent the misuse of OpenAI’s AI tools by underage users. (Link)
📜 Elon Musk to financially support efforts to use AI to decipher Roman scrolls.
Elon Musk shared on X that the Musk Foundation will fund the effort to decipher the scrolls charred by the volcanic eruption of Mt.Vesuvius. The project run by Nat Freidman (former CEO of GitHub) states that the next stage of the effort will cost approximately $2 million, after which they should be able to read entire scrolls. The total cost to decipher all the discovered scrolls is estimated to be around $10 million. (Link)
🤖 Microsoft’s Satya Nadella urges India to capitalize on the opportunity of AI.
The CEO of Microsoft, Satya Nadella, at the Taj Mahal Hotel in Mumbai, expressed how India has an unprecedented opportunity to capitalize on the AI wave owing to the 5 million+ programmers in the country. He also stated that Microsoft will help train over 2 million employees in India with the skills required for AI development. (Link)
🔒 OpenAI introduces the creation of endpoint-specific API keys for better security.
The OpenAI Developers account on X announced their latest feature for developers to create endpoint-specific API keys. These special API keys allow for granular access and better security as they will only let specific registered endpoints access the API. (Link)
🛋️ Ikea introduces a new ChatGPT-powered AI assistant for interior design.
On the OpenAI GPT store, Ikea launched its AI assistant, which helps users envision and draw inspiration to design their interior spaces using Ikea products. The AI assistant helps users input specific dimensions, budgets, preferences, and requirements for personalized furniture recommendations through a familiar ChatGPT-style window. (Link)
New to the newsletter?
The AI Edge keeps engineering leaders & AI enthusiasts like you on the cutting edge of AI. From ML to ChatGPT to generative AI and LLMs, We break down the latest AI developments and how you can apply them in your work.
Thanks for reading, and see you tomorrow. 😊
Thanks for discussing the readability article!