• Rabbit is the most discussed AI thing presented at CES so far – it’s a cute standalone device that runs Rabbit OS, a personal system operated through natural language. The heart of it is a Large Action Model that trains AI agents (“rabbits”) to perform actions across different apps for you. Some people on Twitter can’t wait to get their hands on this new Pokedex-like gadget, while others remain skeptical and ask “shouldn’t this just be an app?” - Rabbit
• OpenAI released a statement condemning the New York Times lawsuit, with four main points:
–”We collaborate with news organizations and are creating new opportunities
–Training is fair use, but we provide an opt-out because it’s the right thing to do
–’Regurgitation’ is a rare bug that we are working to drive to zero
–The New York Times is not telling the full story”
- OpenAI
• In what seems like a Hollywood-inspired twist of fate, Transformers are powering the next generation of robotics – large foundation models are used to help robots understand real-world settings, in preparation for a future in which “these models and systems can be integrated to create robots.” - DeepMind
• An unusual but noteworthy example of AIs impact – Microsoft is adding a Copilot button to keyboards, the first new button since the Windows button in the 1990s. - AP News
• Interesting AI-related predictions for 2024:
–Relational will break free of SQL
–Vector databases will become the most sought-after technology
–Fishing for LLM gold in enterprise data lakes
–Companies without sophisticated enough automation to power AI will feel the burn
–Cloud FinOps teams will optimize their data pipelines
–Intent data will become a must-have for go-to-market teams
–Data and business teams will lock horns over onboarding AI products
–Enterprises will get a double whammy from real-time and AI
–Knowledge graphs will help users eliminate data silos
–AI will change the current approach to data management
–The role of Chief Data Officer will become a prerequisite for CIO hopefuls
- Venture Beat
• We’re still early in the AI revolution, and yet it’s already too easy to build a system that can identify your location from a photo with 95% accuracy – that’s what PIGEON, built by three computer science students from Stanford, has achieved and while it can be used for many positive cases, it also creates serious privacy concerns. - NPR
• Qualcomm CEO confirms that AI applications will shift from cloud to mobile devices in 2024, enhancing user experience and efficiency – he sees genAI as a driver for a new upgrade cycle in smartphones, requiring more powerful chips. - Financial Times
• Intel forms Articul8 AI, an independent company focusing on AI software – led by former Intel executive Arun Subramaniyan, it aims to address privacy and security concerns of customers hesitant to use large cloud computing services, and manage potential cost issues. - Reuters
• "On-chip governance mechanisms" for AI chips are a potential method to implement policy controls directly on hardware – this would secure AI systems against misuse (like cyberattacks or mass surveillance) while maintaining the competitiveness of US firms. Useful for export control enforcement and future international AI regulations, it can leverage technologies already present in chips from major companies like AMD, Apple, Intel, and NVIDIA. - CNAS
• One to think about – with over 100 billion Arm Cortex M chips idling in the world, embedded devices hold the power to execute greater computational operations than the world’s active GPUs and TPUs. What if we used this “Dark Compute” to power AI systems? - Pete Warden
