Welcome back to Daily Zaps, your regularly-scheduled dose of AI news ⚡️ 

Here’s what we got for ya today:

• 🍎 Honest review of Apple Intelligence

• 🤖 Power-hungry AI data centers

• ⛓️ Anthropic launches tool to connect AI to datasets

• 🎖️ Experts find Meta’s Defense Llama advise flawed

Let’s get right into it!

Sign Up | Advertise | Tools

REVIEWS

Honest review of Apple Intelligence

Marques Brownlee reviewed Apple Intelligence, highlighting its promise but noting its underwhelming delivery so far. Announced at WWDC 2024, most features are still in development, with a full rollout expected by March 2025. Existing tools like writing assistants, notification summaries, and generative emojis work but feel niche or repetitive.

The background eraser in Photos and visual intelligence tools show promise, but innovations like AI-enhanced Siri or call summaries fall short of expectations. MKBHD praised Apple for running features on-device and increasing base RAM across devices, but overall, Apple Intelligence feels overhyped compared to its actual impact. He’s hopeful updates will improve the experience.

ENVIRONMENT

Power-hungry AI data centers

The rapid growth of AI and cloud computing is driving massive energy demands, with data centers now consuming as much electricity as entire cities or states. Developers are struggling to find enough power and land to accommodate these expanding facilities, some requiring gigawatts of energy—enough to power hundreds of thousands of homes.

While companies prefer renewable energy, the scale and reliability needed are pushing reliance on natural gas and even nuclear power. This expansion is challenging sustainability goals, with industry leaders emphasizing partnerships with utilities to mitigate community impacts and exploring innovations to improve efficiency and reduce emissions.

STARTUPS

Anthropic launches tool to connect AI to datasets

Anthropic has introduced the Model Context Protocol (MCP), a new open-source tool that makes it easier for AI systems to connect with different data sources. Unlike OpenAI’s recent feature for specific coding tools, MCP is designed to work with any AI system and data.

Developers can set it up once and use it to access multiple data sources without creating custom connections for each one. Tools like Replit and Codeium are already using MCP to build AI agents that can handle tasks across various platforms. This makes AI integration simpler and more efficient as the technology continues to grow.

BIG TECH

Experts find Meta’s Defense Llama" advise flawed

Meta's AI tool, "Defense Llama," developed in partnership with Scale AI, aims to support military planning using generative AI. However, its marketing, featuring hypothetical airstrike scenarios, has drawn criticism from experts who find the tool's advice flawed, oversimplified, and potentially dangerous.

While Scale AI claims the model adheres to ethical and military guidelines, critics argue that its advertised outputs fail to account for critical nuances in targeting, munitions selection, and legal obligations. The tool's marketing highlights concerns about relying on AI for complex military decisions, raising questions about its practicality and ethical implications in real-world defense scenarios.