🧠 AI is entering the classroom in multiple ways

🎨 AI Image of The Week

🧠 AI and Machine Learning

New A.I. Chatbot Tutors Could Upend Student Learning (5 minute read)
In a lively classroom at Khan Lab School, Palo Alto, sixth-grade students huddled around their laptops, working on math lessons. Instead of immediately seeking their teacher's help, they utilized Khanmigo, an experimental chatbot tutor powered by artificial intelligence. With clear instructions and a congratulatory approach, Khanmigo guided students like Zaya through problem-solving, making it an engaging and personalized learning experience. These students are among the first in the US to explore the potential of conversational chatbots designed for educational purposes, offering one-on-one tutoring simulation in a user-friendly manner.

MosaicML’s latest models outperform GPT-3 with just 30B parameters (3 minute read)
MosaicML, the open-source LLM provider, has announced the release of their latest models, the MPT-30B Base, Instruct, and Chat. These advanced models have been trained on the MosaicML Platform using NVIDIA's latest H100 accelerators, offering high-quality generative AI while prioritizing data privacy and security. With over 3.3 million downloads since their May 2023 launch, the MPT-7B models have gained popularity. The newly released MPT-30B models provide even higher quality, with improved possibilities for various applications. Notably, MPT-30B achieves superior quality to GPT-3 while using fewer parameters, making it more accessible and cost-effective for local hardware usage and deployment.

In tech, everything is labeled "AI" now (3 minute read)
The current artificial intelligence frenzy in the tech world has led to an indiscriminate labeling of products and services as "AI." This distorted understanding of the technology hampers thoughtful application and increases the potential for harm. Genuine advancements in machine learning have sparked a tech industry investment bubble, encouraging companies to loosely define AI and include any statistically-driven computer program. Striking a balance between recognizing the risks and avoiding overgeneralization is essential in navigating this fast-paced and hype-driven AI landscape.

💼 Business

Meta announces Voicebox, a generative model for multiple voice synthesis tasks (3 minute read)
Meta Platforms' AI research arm recently unveiled Voicebox, an impressive machine learning model that can convert text into speech. What makes Voicebox stand out is its unique capability to perform various tasks it hasn't been explicitly trained for, such as editing, noise removal, and style transfer. Developed using a special method by Meta researchers, Voicebox shows great promise, although Meta has withheld its release due to ethical concerns. Nevertheless, the model's potential is vast and could pave the way for exciting applications in the future. Voicebox acts as a generative model, synthesizing speech in six languages including English, French, Spanish, German, Polish, and Portuguese. Unlike traditional large language models, Voicebox goes beyond understanding word and text sequences and instead focuses on learning the patterns that link voice audio samples to their corresponding transcripts.

AWS launches $100M program to fund generative AI initiatives (2 minute read)
Amazon is launching a new initiative aimed at supporting customers and partners focused on generative AI. The AWS Generative AI Innovation Center will allocate a generous $100 million to connect AWS-affiliated data scientists, strategists, engineers, and solutions architects with customers and partners. Through a variety of means including workshops, engagements, and training, Amazon aims to accelerate enterprise innovation and success in the realm of generative AI. By leveraging best practices and industry expertise, AWS will assist customers in envisioning and shaping the most valuable use cases for their businesses.

❓ Misc

Harvard's New Computer Science Teacher Is a Chatbot (2 minute read)
Harvard University is embracing the power of generative AI by officially incorporating it as a learning tool in its renowned coding course, Computer Science 50: Introduction to Computer Science (CS50). Starting this fall, students enrolled in CS50 will have the opportunity to utilize AI to aid them in debugging code, offering feedback on their designs, and providing individualized assistance with error messages and unfamiliar lines of code. Professor David J. Malan, the instructor of CS50, envisions a future where AI can approximate a 1:1 teacher-to-student ratio, supporting students' learning at their own pace and in their preferred style. This significant step by Harvard represents a swift turnaround from the previous academic year when the university did not have an AI policy by the end of the fall semester in 2022.

Wimbledon to feature AI commentary and draw analysis (3 minute read)
IBM and the All England Lawn Tennis Club have AI-powered enhancements in store for the Wimbledon digital fan experience at this year's Championships. Tennis enthusiasts can look forward to generative AI commentary and AI draw analysis, designed to elevate engagement and provide insightful information. Developed in collaboration with IBM iX, the AI commentary feature allows fans watching match highlights videos to receive audio commentary and captions for key moments, adding an extra layer of excitement. Additionally, a groundbreaking feature introduced this year is AI draw analysis, a first in the world of tennis. Using AI, this feature assesses the favorability of the path to the final for each player in the singles draw. Various factors, including potential match-ups with future opponents and a player's position compared to competitors, contribute to the draw favorability rating.

🐥 Best of Twitter

Thanks for reading, if you enjoyed, tell your friends!