Hello GPT-4o! 2x Faster, 50% Cheaper GPT-4 Turbo Upgrade
Plus: TikTok tests ChatGPT-powered search results, Meta explores AI-assisted earphones with built-in cameras.
Hello Engineering Leaders and AI Enthusiasts!
Welcome to the 274th edition of The AI Edge newsletter. This edition features “GPT-4o! 2x Faster, 50% Cheaper GPT-4 Turbo Upgrade.”
And a huge shoutout to our amazing readers. We appreciate you😊
In today’s edition:
🚀 GPT-4o: 2x faster, 50% cheaper, 5x rate limits than GPT-4 Turbo
🔍 TikTok tests ChatGPT-powered search results
🎧 Meta explores AI-assisted earphones with built-in cameras
📚 Knowledge Nugget: Why I'm building my own AI Agent library by
Let’s go!
GPT-4o: 2x faster, 50% cheaper, 5x rate limits than GPT-4 Turbo
OpenAI unveiled its new flagship model, GPT-4o (“o” stands for “omni”). It provides GPT-4-level intelligence but is 2x faster, 50% cheaper, has 5x higher rate limits, and enhanced text, voice, and vision capabilities than GPT-4 Turbo. It also matches GPT-4 Turbo performance on text in English and code, with significant improvements for text in non-English languages.
GPT-4o can respond to audio inputs in as little as 232 ms, with an average of 320 ms, which is similar to human response time in a conversation. It is available in the ChatGPT free tier and to Plus users.
Why does it matter?
GPT-4o is a step towards much more natural human-computer interaction. It accepts any combination of text, audio, and image as input and generates any combination of text, audio, and image outputs. GPT-4o will set a new benchmark for AI capabilities and pave the way for more intelligent and accessible AI systems.
TikTok tests ChatGPT-powered search results
TikTok is testing a new search feature that uses AI to generate results for some queries. For certain search terms, like recipes or product recommendations, TikTok will display AI-generated answers at the top of the results. These answers are powered by ChatGPT, and when you click on one of these AI results, it takes you to a page with a full response.
TikTok is also testing "search highlights," which appear at the top of some searches. However, it's unclear whether these highlights are AI-generated.
Why does it matter?
It could make search on TikTok more authoritative and reliable by providing factual answers from AI models like ChatGPT. This could greatly impact how people use TikTok for search and information gathering, thereby positioning TikTok more directly as a search engine competitor to Google, catering to younger users' preferences.
Meta explores AI-assisted earphones with built-in cameras
Meta is currently exploring AI-powered earphones with built-in cameras. The project is known internally as “Camerabuds;” however, it’s uncertain if the final product will be in-ear earbuds or over-ear headphones. The cameras let the AI take in surrounding information visually, allowing it to translate foreign languages, identify objects, and more.
CEO Mark Zuckerberg has reportedly seen several possible designs but has not been satisfied with them. Outside of design, the major hurdle is fitting a camera and batteries in a tiny device like an earbud.
Why does it matter?
For the AI wearable space, Meta's development hints at a growing interest in AI-infused devices, potentially paving the way for more sophisticated and interactive wearables. Regarding earphones, this development suggests a potential shift towards more intelligent and feature-rich earphone designs, offering users enhanced functionalities like real-time translation and object identification.
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: Why I'm building my own AI Agent library
In this article, the author,
, is building his own AI Agent library in Python. He has three main reasons for doing this.First, no established standards exist for how AI Agents should be developed. Since the field is new, the author wants to contribute his own ideas and experiment to see what works well.
Second, he wants to learn about building AI Agents from first principles by creating his own library from scratch. This will allow him to deeply understand the core components like prompting, planning, tooling, memory, and reflection.
Third, he finds the process of building an AI Agent library to be really fun and exciting.
The author is also open to potentially using an existing third-party library in the future if he needs more advanced capabilities when working on an AI Agent product.
Why does it matter?
As AI Agents become more sophisticated, they have the potential to automate complex real-world tasks and decisions across many industries. Such open-source explorations may help establish best practices and standards in this nascent field.
What Else Is Happening❗
💰 Anthropic launches Claude in Europe in the hope of raising more money
Anthropic launches Claude in Europe with support for “multiple languages,” including French, German, Italian, and Spanish across Claude.ai, its iOS app, and its business plan for teams. The launch comes after Anthropic extended its API to Europe for developers. Both moves are part of its bigger push to raise more money for rapid growth. (Link)
🤖 UAE launches a new AI model to compete with big tech
Abu Dhabi's Technology Innovation Institute (TII) released Falcon 2 series: Falcon 2 11B, a text-based model, and Falcon 2 11B VLM, a vision-to-language model that can generate a text description of an uploaded image. This Gen AI model could compete with some of the biggest technology companies and make UAE a strong candidate in the Gen AI space. (Link)
🖥️ Japan team builds Gen AI foundation model using supercomputer Fukagu
Tokyo Institute of Technology researchers and others have announced the development of an LLM that can serve as a foundation for Gen AI using the Japanese supercomputer Fugaku. Trained extensively on Japanese data, which accounts for 60% of the total training data, the Fugaku-LLM model is expected to lead research on Gen AI tailored to domestic needs. (Link)
⏰ Microsoft Places finds the best time for your next office day
Microsoft has launched a new app called Microsoft Places that uses AI to help employees and managers coordinate their in-office days and book meeting rooms. The application integrates with Outlook and Microsoft Teams, allowing employees to view and share their office plans and get AI-powered suggestions on which days they should come in. (Link)
🤝 U.S. and China to meet in Geneva to discuss AI risks
The U.S. and China will hold their first formal government-to-government discussion on the risks of artificial intelligence in Geneva. The U.S. side will be led by officials from the National Security Council and State Department, while the Chinese side will be led by officials from the Foreign Ministry and the National Development and Reform Commission. (Link)
New to the newsletter?
The AI Edge keeps engineering leaders & AI enthusiasts like you on the cutting edge of AI. From machine learning to ChatGPT to generative AI and large language models, we break down the latest AI developments and how you can apply them in your work.
Thanks for reading, and see you tomorrow. 😊