- Horizon AI
- Posts
- ChatGPT-4 No Longer the Best? 🔥
Sponsored by
Welcome to another edition of Horizon AI,
For the first time since its release, GPT-4 has been pushed to second place on the Chatbot Arena leaderboard. Taking the first spot is Anthropic's recently released Claude 3 Opus.
Let's jump right in!
Read Time: 3.5 min
Here's what's new today in the Horizon AI
A Chart to Kickstart Your Day 📈: How people are using GenAI
Anthropic's Claude 3 Dethrones GPT-4 as Users' Favorite AI Chatbot 🔥
AI Research: Google Unveils 'VideoPrism', a Model for General-Purpose Video Understanding
AI Tutorial: How to access Google’s new DJ mode
AI Tools to check out
The Latest in AI and Tech 💡
A chart to kickstart your day
How people are using GenAI
Researchers mined web forums like Quora and Reddit, filtering through tens of thousands of posts to identify 100 different use-case categories, which they then organized into six themes.
AI News
ANTHROPIC
Anthropic's Claude 3 Dethrones GPT-4 as Users' Favorite AI Chatbot 🔥
Source: HuggingFace
Anthropic's Claude 3 Opus has knocked OpenAI's GPT-4 off its pedestal as the highest user-rated large language model on the Chatbot Arena leaderboard.
Details:
In contrast to other forms of benchmarking for AI models, the LMSYS Chatbot Arena relies on human votes. People blind-rank the output of two different models to the same prompt.
This is the first time since GPT-4's release around a year ago that another model has outperformed it in user ratings.
Impressively, Anthropic's much cheaper Haiku model also nearly matched GPT-4's ratings.
These are great news for Anthropic. However, OpenAI is expected to counter with a new flagship model rumored to be released soon. OpenAI CEO Sam Altman has confirmed that his company intends to launch an "amazing" AI model this year.
TOGETHER WITH CLUMIO
With rising costs for Amazon S3 storage and potentially devastating business consequences from data loss, you need a holistic approach to cutting unnecessary spending and guarding against risks. Lawrence Miller, a consultant to multinational corporations who holds numerous networking certifications, has authored a concise volume that lays out the path to success in managing backup and compliance for S3 data lakes.
AI RESEARCH
Google Unveils 'VideoPrism', a Model for General-Purpose Video Understanding
Source: Google Research
Google Research has introduced VideoPrism, a new visual video encoder suitable for a wide range of video understanding tasks.
Details:
The model excels at identifying objects and activities in videos, as well as providing accurate answers to questions about them.
It achieved top scores on 30 out of 33 video understanding benchmarks
Showed strong performance for niche scientific applications like animal behavior analysis and ecology.
The research team aims for VideoPrism to drive breakthroughs in AI and video analytics, unlocking the potential of video models in areas like education, and healthcare.
AI Tutorial
How to access Google’s new DJ mode
Google has released a new DJ mode for MusicFX, a experimental tool that allows you to generate music from a series of text prompts. To access it:
Go to Google Labs, find the MusicFX section, and select “Try it now”
*Note: You might need to wait a bit before accessing it. If it's not available in your region, you can try using a VPN.
Turn on DJ Mode
At the top of MusicFX, you will find a DJ Mode button, just select it.
This will allow you to add the sounds and moods you want with a prompt.
You will be given some default prompts which you can delete and you can also add your own prompt.
Start experimenting
Keep adding more of your own prompts or select the suggestions at the bottom. You can adjust each level to highlight what you want.
The music will update in real time according to the changes you make.
If you are not convinced about a specific one, you can delete or temporarily mute it.
AI Tools to check out
🦾 Speck: AI assistant that turns browser recordings into automations.
🗣 Fluently: AI-powered speaking coach for calls. It helps non-native professionals improve their English with instant feedback after each call.
👗 Kopia: AI-powered virtual try-on tool for clothing brands.
💹 Persuva: Create persuasive ad copy with AI.
🖼 Lummi: The ultimate source for free AI-generated stock photos and royalty-free images.
The latest in AI and Tech
Apple has announced that its Worldwide Developers Conference (WWDC) 2024 will take place on June 10th. It is expected that significant AI-related announcements will be made at this year's event.
Elon Musk has announced that Grok will now be available to all premium subscribers of the social media platform. In a post on X, Musk announced this change without providing further details. Previously, the Grok chatbot was exclusive to Premium+ subscribers.
Researchers in Belgium are leveraging AI to enhance the taste of beer. While the results were mostly positive, the scientists emphasize the continued importance of brewers' skills in the process.
Google has added its Gemini 1.0 Pro chatbot to the Google Messages beta for RCS Messaging. To access the Gemini AI integration, users need to have RCS Messaging enabled, a Google Account, and a newer phone like the Pixel 6 or a recent Samsung Galaxy model.
SK Hynix, an Nvidia supplier, is set to allocate approximately $4 billion toward constructing an advanced chip-packaging facility in West Lafayette, Indiana. An investment that is expected to generate up to 1,000 job opportunities.
That’s a wrap!
We'd love to hear your thoughts on today's email!Your feedback helps us improve our content |
Not subscribed yet? Sign up here and send it to a colleague or friend!
See you in our next edition!
Gina 👩🏻💻