_{Welcome back for the Latest AI Drops!}

Today’s Drops:

• Bill Gates on AI's Global Impact

• Turn your ideas into songs with Suno on Microsoft Copilot

• AI Determines Location from Photos

• Trending AI Startup

• Trending on X

• AI News “Research reveals that ChatGPT struggles with interpreting SEC filings, while TomTom and Microsoft team up to develop an AI vehicle assistant, Stability AI rolls out a subscription model, and U.S. Senators propose strict penalties to mitigate AI misuse in financial crimes."

• Trending GitHub Projects

• AI Tools

Read Time: 4.5 minutes

Bill Gates on AI's Global Impact

In his annual end-of-year letter, Bill Gates discussed the significant potential for AI to revolutionize various aspects of our lives starting in 2024. He emphasized that the work done in the next year will pave the way for a technology boom later in the decade.  

Gates considers AI a significant force driving global health improvements and innovation in developed and developing countries. He predicts a rapid acceleration in discoveries due to AI. This technology could assist in managing antibiotic-resistant diseases, aiding non-English speakers with high-risk pregnancies, and improving access to medical information. However, challenges remain in scaling these technologies and maintaining their quality and functionality over time. Gates suggests that smart investments in AI could make the world more equitable. 

Key Points:

1. AI Adoption Timeline: Gates anticipates widespread AI adoption in the U.S. within 18–24 months and in Africa within three years, marking a significant reduction in the digital divide.

2. Personal Experience: Despite planning to use AI for the Gates Foundation's strategy reviews, Gates admits to sticking to traditional preparation methods.

3. AI for Global Health: Gates highlights AI's potential in global health, especially in addressing antibiotic resistance and managing high-risk pregnancies. The Gates Foundation is developing tools like AI copilots, home-based ultrasound equipment, and voice-enabled apps for maternal health workers.

4. Educational Innovations: Gates is excited about AI tools like Khanmigo and MATHia, which can offer personalized tutoring to students.

Gates' Vision for AI:Gates sees AI as a crucial driver in solving global challenges in education and health. He emphasizes the need for AI tools to be tailored to their users, particularly in developing countries. Gates is optimistic about AI's role in speeding up drug discovery and addressing critical health issues like AIDS, tuberculosis, and malaria.

Challenges Ahead:Despite the promise of AI, Gates acknowledges significant hurdles, including the need to prevent AI biases and hallucinations (false claims made by AI systems). He stresses the importance of designing AI products that are accessible and relevant to their users.

Source

Turn your ideas into songs with Suno on Microsoft Copilot

Microsoft Copilot, an AI-powered chatbot by Microsoft, has expanded its capabilities to include music composition by integrating with the GenAI music app Suno. This feature allows users to generate complete songs from simple text prompts, including lyrics, instrumentals, and singing voices. For instance, users can prompt for a pop song about adventures with family, and Suno, accessed via a plugin on Microsoft Edge, transforms this input into a full musical piece.

To get started creating your music, follow these steps:

• Open Microsoft Edge, visit copilot.microsoft.com and ensure you're signed in with your Microsoft Account

• Enable the Suno plugin or click on the Suno logo that says, "Make music with Suno."

• Ask Copilot to create a song for you, such as, "Create a pop song about adventures with your family."

• Jam along to your new tune

• Share on social or with your friends and colleagues

The Microsoft Copilot and Suno integration are notable advancements in making AI-driven music creation more accessible, fun and widespread. Yet, it's surrounded by ethical and legal issues, primarily because AI might use existing music without the original artists' consent or fair compensation. The legal clarity around AI-created music is still evolving, with potential changes on the horizon due to proposed legislation protecting artists' digital rights, including their unique musical styles.

Source

AI Determines Location from Photos

Stanford graduate students have developed an AI model named PIGEON, which can accurately determine specific locations by analyzing Google Street View images. This model boasts a 92% success rate in identifying the correct country and can pinpoint a place within 25 kilometres in over 40% of cases. PIGEON's abilities were highlighted when it ranked in the top 0.01% of players in GeoGuessr, a game that involves guessing locations from Google Street View photos, and even defeated one of the world's best professional GeoGuessr players, Trevor Rainbolt.

PIGEON utilizes CLIP, a neural network by OpenAI, trained on visual category names. It was developed using 100,000 locations from GeoGuessr, each with four images to create a panoramic view, totalling 400,000 images. This training dataset is relatively small compared to other AI models like OpenAI's DALL-E 2, which is trained on hundreds of millions of images.

A related model named PIGEOTTO was created using over four million photos from Flickr and Wikipedia. PIGEOTTO has shown impressive results in image geolocalization, surpassing previous city and country accuracy benchmarks.

Despite the positive applications of these models, like aiding in autonomous driving and visual investigations, the developers are cautious about privacy concerns. They have opted not to release the model weights publicly, limiting access to the code for academic validation only.

Source

Trending on X

AI News

Research indicates that large language models like ChatGPT often struggle to accurately answer questions based on Securities and Exchange Commission filings, underscoring the challenges AI faces in industries like finance that require high precision and reliability. Link

TomTom, a digital mapping specialist, has collaborated with Microsoft to develop an AI-powered conversational assistant for vehicles. This assistant, leveraging Microsoft's Azure OpenAI Service, will enable natural voice interactions with vehicle systems, such as infotainment and location search. TomTom, known for competing with platforms like Google Maps, utilized various Microsoft services to create this voice assistant, which can be integrated into other automotive systems. Link

Stability AI has introduced a subscription service for commercial use of its AI models, including Stable Diffusion, with three tiers: a free tier for personal and research use, a $20 per month tier for smaller entities, and an enterprise plan. This move aims to balance profitability with openness while funding future AI research and development. The paid tiers allow commercial use of new AI models, reflecting a shift in how Stability AI handles commercial rights. Link

U.S. Senators Mark Warner and John Kennedy have introduced a bill proposing strict penalties for using AI in illegal financial activities like market manipulation and securities fraud. The legislation aims to mitigate AI-related financial risks, empowering the SEC to impose triple fines and encouraging regulatory bodies to develop new AI governance methods in finance. This move follows warnings from financial authorities about AI's potential risks to market stability. Link

Welcome to the innovative world of GitHub projects!

1. geekan/MetaGPT: MetaGPT is a multi-agent framework that can generate Product Requirement Documents (PRD), designs, tasks, and repositories from a single line of requirement​​.

2. open-mmlab/Amphion: Amphion is a toolkit for audio, music, and speech generation, aimed at supporting reproducible research and assisting junior researchers and engineers in these fields​​.

3. gpt-engineer-org/gpt-engineer: This project involves an AI that builds specified projects, asking for clarifications as needed during the process​​.

4. microsoft/Mastering-GitHub-Copilot-for-Paired-Programming: A six-lesson course designed to teach everything about utilizing GitHub Copilot for AI-assisted paired programming​​.

5. bytedance/ImageDream: This repository contains the code for ImageDream (https://image-dream.github.io/)