- Future Blueprint
- Posts
- New Chinese AI model eclipses GPT-4o 🌒
In partnership with
📖 TODAY’S ISSUE
Howdy, humans!
🧵 Here are some useful AI updates and tools we gathered today:
AI lie detectors are better than humans at spotting lies
Latest in tech and AI + dev resources
Cool AI tools + see who raised funds
Prompt of the day and more
PRESENTED BY AI TOOL REPORT
We scour 100+ sources daily
Read by CEOs, scientists, business owners and more
3.5 million subscribers
Sponsored Ad
🗞️ HIGHLIGHT OF THE DAY
TL;DR
SenseTime unveiled SenseNova 5o, China's first real-time multimodal AI model to rival GPT-4o, at the World Artificial Intelligence Conference. It processes audio, text, image, and video data, enabling interactive conversations. SenseTime also introduced SenseNova 5.5, which boasts significant performance improvements.
Summary
SenseNova 5o: First real-time multimodal AI model in China.
Capabilities: Processes audio, text, image, and video data.
SenseNova 5.5: 30% performance increase, enhanced reasoning and language skills.
Edge Models: SenseChat Lite-5.5 offers faster, cost-effective inference.
Vimi AI: Generates one-minute clips from a single photo with advanced controls.
What we think
SenseTime's innovations with SenseNova 5o and 5.5 mark a significant advancement in AI capabilities, particularly in multimodal interaction and real-time processing. This progress enhances China's position in the global AI landscape. SenseTime's focus on edge-based solutions also underscores the growing importance of efficient, scalable AI applications.
⚡ LATEST IN TECH AND AI
OpenAI's DALL-E 2 is revolutionizing art creation with its ability to generate realistic images and art from text descriptions. This generative model allows users to create a vast array of imagery, from detailed paintings to photorealistic scenes, simply by describing what they envision. The technology represents a significant leap in AI's creative capabilities, blending deep learning with artistic expression to produce visually stunning and conceptually rich outputs. DALL-E 2's versatility and ease of use open up new possibilities for artists, designers, and anyone interested in exploring the fusion of AI and creativity.
Groq has introduced a high-speed LLM inference engine powered by its innovative Language Processing Units (LPUs), designed to offer faster and more cost-effective AI language model processing than Nvidia GPUs. With over 270,000 users already, Groq's solution focuses on privacy by not logging chat queries and aims to become the leading infrastructure choice for startups by the end of 2024. This advancement positions Groq as a formidable competitor in the AI hardware market.
Talks between Oracle and Elon Musk's AI startup xAI for a $10 billion server deal have fallen through. The potential agreement aimed to leverage Oracle's cloud infrastructure for xAI's advanced AI projects but did not reach fruition. This development may prompt xAI to explore other cloud service providers for their infrastructure needs.
💻 DEV RESOURCES
A developer discovered a way to access the internet for free while clicking through an airline's Wi-Fi payment menu. This article discusses how they exploited this discovery to develop a tool that allowed them to access the full internet for free, albeit at a very slow pace. The exploit involves editing data on the developer's airmiles account to send and receive HTTP requests and responses.
Maelstrom is a suite of tools that can run tests in hermetic microcontainers locally or distributed across arbitrarily large clusters. It currently has test runners for Rust and Python, with more on the way. Maelstrom is reliable, easy to use, fast, and scalable. The project is currently Linux-only and relies on namespaces to implement containers.
TicketMaster issues tickets via a system called Mobile Entry, or SafeTix. The system displays a form of rotating barcode that refreshes every few seconds. It prevents people from printing or screenshotting the tickets they purchased. This article explains how the system works - it's possible to duplicate these barcodes in custom apps and even resell tickets outside of TicketMaster's closed marketplace.
🤖 COOL AI TOOLS
Discover mind-blowing AI tools
The first voice AI that responds empathically, built to align technology with human well-being.
An AI assistant that can research, book, coordinate, and pay vendors in order to ease the process of moving to a new city.
A low-latency, conversational AI chatbot that can think and talk at the same time.
💸 WHO RAISED?
Discover startups who just raised funds
💪 POST OF THE DAY
The uncanny valley has been crossed
Westworld is getting closer
— Barsee 🐶 (@heyBarsee)
1:04 PM • Jul 2, 2024
✨ AI IMAGE CHALLENGE
Leche flan
Image 1
Which image do you think is real? |
Image 2
✅ PROMPT OF THE DAY
Jobs-to-be-done
Prompt: Assume I'm a Product Manager for a [insert product here, ex: grocery delivery mobile app]. It will be serving the [insert region here, example: Toronto GTA region] only. I want to conduct some consumer research to figure out my consumer's jobs-to-be-done.
What personas should I target and then what questions should I be asking them to elicit their job-to-be-done. Take inspiration from the book "Mom Test".You can adapt the prompt to your specific needs.
🔥 LIKING THE STUFF?
Share The Edge with your fellow founders and get awesome stuff from our team as a thank you!
⭐️ RATE THIS
How did you like it?
Rate this newsletter!Let us know what you like! |
If you have any comments or feedback, just respond to this email!
Thanks for reading,
Sam & The Edge team
Reply