Weekly Magic AI - Google’s Gemma, X and Midjourney, and Opinions about OpenAI’s Sora
Top AI news of the week, Magic AI tool of the week, and article of the week
Hi AI Enthusiasts,
Welcome to our weekly Magic AI, where we bring you the latest updates on Artificial Intelligence (AI) in an accessible way.
This week’s Magic AI tool is a graphic design tool for creating hand-drawn diagrams. We think that’s a hidden gem, so be sure to check it out! 😀
Top AI news of the week
💬 Google introduces AI model Gemma
Google has unveiled Gemma, a family of open-source AI models based on the same research and technology as the Gemini models. Google releases the model in two sizes: Gemma 2B and Gemma 7B. Both are available in pre-trained and instruction-tuned variants. In addition, Gemma outperforms Meta’s Llama-2 model in key benchmarks.
The Gemma models are not as powerful as the Gemini models but the models provide speed and cost efficiencies, allowing them to run directly on a laptop.
Our thoughts
Google goes open-source! That’s good news. While the Gemma models may not be as powerful as the commercial Gemini variant, they are directly competing with Meta’s Llama models.
Google’s Gemma release could also put pressure on OpenAI, as OpenAI is one of the few AI companies without an open-source model on the market. Very irritating, because OpenAI was founded on the idea of openness.
More information
- Gemma: Introducing new state-of-the-art open models - Google Blog
🎇 Partnership between X and Midjourney?
Musk said in a live Space that X is looking for a partnership with image generator Midjourney. X could integrate the image generator into its Grok chatbot. In addition, some X users are hinting that X could soon be labeling AI images. This could also indicate a deal with Midjourney. It is still unclear whether the partnership will actually take place.
Our thoughts
Integrating Midjourney into Grok could take the chatbot to a whole new level. Midjourney is one of the best image generators on the market. It could also allow users to create and share unique AI-generated images. This could significantly improve the user experience on the platform and increase interaction. The combination of Grok’s ability to produce creative texts and Midjourney’s image generation could enhance X’s offering.
More information
- A conversation with Elon Musk - Space on X
🤔 Opinions about OpenAI’s Sora
The AI model Sora is primarily perceived as a text and video-to-video model. However, the research goal of OpenAI is a world simulator. In the opinion of Meta’s AI boss Yann LeCun, OpenAI’s Sora is not suitable as a world simulator and therefore no progress towards Artificial General Intelligence.
He argues that generative models for sensory inputs will fail due to their complexity. Generative AI works well for texts because text is discrete and has a finite number of symbols. However, sensory inputs generate a higher level of complexity.
Instead, LeCun presents his model, the Video Joint Embedding Predictive Architecture (V-JEPA), which aims to predict and interpret complex interactions. V-JEPA operates in a broader conceptual space, similar to human cognitive image processing. It can adapt to different tasks without requiring complete retraining. Meta’s AI team plans to enhance V-JEPA’s capabilities and develop comprehensive world models for autonomous AI systems.
Elon Musk has also commented on the Sora hype. He says that Tesla video generation outperforms OpenAI’s model by predicting “extremely accurate physics” for its self-driving functions.
Our thoughts
World models are essential for the development of autonomous systems like Tesla’s FSD. Everyone takes a different approach to creating good world models. We’ll see which ones work best.
More information
Magic AI tool of the week
Have you ever looked for a tool that lets you create hand-drawn diagrams with ease? Then, Excalidraw is right for you! This open-source, privacy-first, and user-friendly sketching tool is perfect for crafting diagrams, wireframes, and concept maps.
In addition, it offers a range of features like collaborative drawing and diagramming, enabling multiple users to work together on the same drawing in real-time. And the best thing is that you can even use a prompt to create diagrams (OpenAI API key required). How cool is that?
Example:
We use Excalidraw very often to create diagrams for presentations. The AI feature is a useful addition.
Article of the week
On the Internet, you often find websites titled “The 100 Best Prompts to Boost Your Productivity.” These prompts are mostly for tools like ChatGPT, Grok or Microsoft Copilot to optimize simple repetitive tasks.
But most of the time, you still need to customize these prompts to your use case. That’s why you should learn to write effective prompts yourself, especially as a software developer. That’s exactly what our article covers. Check it out to learn more!
💡 Do you enjoy our content and want to read super-detailed articles about AI? If so, subscribe to our blog and get our popular data science cheat sheets for FREE.
Thanks for reading, and see you next time.
- Tinz Twins
P.S. Have a nice weekend! 😉😉
Leave a comment