This Week in AI | Week 20
Let’s get right into the most recent developments in artificial intelligence throughout the week. Here is what’s new, including OpenAI updates and Google’s new model.
ChatGPT for Mac is Now Widely Available
OpenAI has finally released its ChatGPT app for Mac users, following a gradual rollout that began with Plus subscribers. The app, which is completely integrated into macOS, allows users to access ChatGPT quickly using a simple keyboard shortcut.
This integration lets Mac users easily explore information using a chat interface on their desktop.
Key Features:
- Easy Access: The app can be activated using the Option + Space keyboard shortcut, similar to how Spotlight Search is accessed with Command + Space.
- Advanced Capabilities: Beyond text, users can upload files, take screenshots, and even converse directly with ChatGPT.
- Broader Integration: This rollout builds on OpenAI’s partnership with Apple, which will soon bring ChatGPT features to iOS and iPadOS devices through various applications including Siri and Writing Tools.
OpenAI Introduces CriticGPT for Improved Code Quality
OpenAI recently announced a new AI model, CriticGPT, designed to enhance the quality of code generated by GPT-4 by identifying and fixing errors more effectively. Using a technique known as reinforcement learning from human feedback (RLHF), CriticGPT has shown a 63% improvement in error detection over ChatGPT.
Model Insights:
- Enhanced Error Detection: CriticGPT can better identify errors in code, a crucial feature for developers relying on AI-assisted programming.
- Still in Development: While promising, CriticGPT is not yet available for public use as it undergoes further testing, particularly to address its limitations with longer code sequences and multiple error scenarios.
Gemma 2 Released to Developers and Researchers
Google DeepMind has released Gemma 2, an upgrade to their Gemma family of models known for their efficiency and high performance. Available in 9 billion and 27 billion parameter versions, Gemma 2 is designed to deliver robust AI capabilities more affordably and efficiently.
Advancements in Gemma 2:
- Performance and Efficiency: Gemma 2 has major upgrades, making it competitive with much larger models while still being optimised to run on less expensive hardware.
- Accessibility and Integration: The model supports a variety of AI frameworks and can be integrated into different development environments, making it widely accessible to developers and researchers.
Read Past weeks AI Developments
Conclusion
That’s it for this week’s AI news. These updates are making technology even more useful in our daily lives. As AI advancing, we are seeing powerful tools becoming more effective and accessible for everyone.
Stay tuned for more updates next week!
