DeepSeek Again Appears Way Ahead of Rivals

Also, anticipated AI boom is driving electricians to data center sites as boom towns pop up.

⚡️ Headlines

🤖 AI

AI Is Killing Coding - The emergence of AI-powered development environments like Cursor, a fork of Visual Studio Code, is streamlining coding processes, potentially reducing the need for traditional coding skills. [Andrew Zuo].

Most Office Workers Don’t Pay for AI Tools, Tech Brew Survey Finds - A recent survey reveals that 78% of office workers use free AI tools, with only 22% paying for such services, indicating widespread adoption of cost-free AI solutions in the workplace. [Emerging Tech Brew].

Microsoft Is Forcing Its AI Assistant on People—and Making Them Pay - Microsoft is integrating its AI assistant, Copilot, into its 365 suite, requiring users to adopt the feature and pay an additional fee, sparking discussions on user autonomy and cost. [The Wall Street Journal].

We Need to Think Again About What the 'A' in AI Signifies - The article suggests redefining AI from "artificial" to terms like "augmented" or "accelerated" intelligence, emphasizing collaboration rather than replacement, amidst growing public hostility due to concerns over data security and job displacement. [Financial Times].

🦾 Emerging Tech

9 Robotics Startups to Watch - An overview of nine innovative robotics startups making significant strides in the tech industry, highlighting emerging players to monitor. [The Information].

Bitcoin Flirts with $100K as Crypto Industry Rallies for Trump - Following President-elect Trump's victory, Bitcoin's value has surged, reflecting the crypto industry's anticipation of favorable federal policies under the new administration. [The Hill].

🤳 Social Media

The Evolution of X: How Musk Transformed the Social Media Giant in 2024 - Elon Musk's acquisition of Twitter, now rebranded as X, has led to significant changes in the platform's dynamics, influencing political discussions and user interactions throughout 2024. [BBC News].

How Social Media Is Influencing Our Interactions with Public Lands - Social media platforms are shaping public perceptions and behaviors towards natural landscapes, affecting conservation efforts and visitor experiences. [West Virginia Watch].

🔬 Research

Dartmouth Study Shows AI Could Be ‘Double-Edged Sword’ in Medical Research - A Dartmouth study highlights that while AI can enhance medical imaging research, it also poses risks, emphasizing the need for cautious implementation in healthcare. [New Hampshire Public Radio].

🎱 Random

Predictions for AI in 2025: Collaborative Agents, AI Skepticism, and New Risks - Stanford experts forecast the rise of collaborative AI systems, increased skepticism, and emerging risks in the AI landscape by 2025. [Stanford HAI].

The Internet Is Rife with Fake Reviews. Will AI Make It Worse? - The proliferation of AI tools has exacerbated the challenge of fake online reviews, creating a burden for merchants, service providers, and consumers, particularly during peak shopping periods like the holiday season. [Associated Press].

🔌 Plug-Into-This

DeepSeek, a Chinese AI startup, has unveiled DeepSeek-V3, an ultra-large open-source model with 671 billion parameters. Utilizing a mixture-of-experts (MoE) architecture, it activates 37 billion parameters per token, enhancing efficiency and performance. Benchmark tests indicate that DeepSeek-V3 surpasses leading open-source models like Meta's Llama 3.1-405B and rivals closed-source models from Anthropic and OpenAI.

  • Innovative Architecture: DeepSeek-V3 employs multi-head latent attention (MLA) and the proprietary DeepSeekMoE framework, enabling specialized experts to process tasks efficiently by activating only necessary parameters.

  • Enhanced Training Efficiency: The model introduces multi-token prediction (MTP), allowing simultaneous prediction of multiple tokens, resulting in a threefold increase in training speed and the ability to generate 60 tokens per second.

  • Balanced Resource Utilization: An auxiliary loss-free load-balancing strategy dynamically adjusts the workload among experts, ensuring optimal resource use without compromising performance.

  • Extensive Training Data: DeepSeek-V3 was trained on a diverse dataset comprising 14.8 trillion tokens, contributing to its advanced capabilities in various tasks.

  • Open-Source Accessibility: The model is available under DeepSeek's license agreement via platforms like Hugging Face, promoting transparency and collaboration within the AI community.

🚀 The release of DeepSeek-V3 signifies a pivotal advancement in open-source AI, narrowing the performance gap with proprietary models and fostering a more collaborative and innovative AI ecosystem.

The rapid expansion of artificial intelligence is driving a construction boom in data centers across central Washington, drawing electricians from across the U.S. These projects leverage the region's abundant hydropower, reshaping the local economy while straining housing and infrastructure resources.

  • Central Washington’s Appeal: The region's ample hydropower and lower costs have made it a hub for data center construction, with facilities from companies like Microsoft proliferating.

  • Economic Transformation: These data centers are significantly boosting local property values and tax revenues, funding public projects like new schools but also driving up housing costs.

  • Challenges for Communities: The influx of well-paid workers has led to housing shortages and increased living costs, leaving many long-time residents struggling to keep pace.

  • Union Labor and Training: The International Brotherhood of Electrical Workers (IBEW) is expanding apprenticeship programs to meet the soaring demand for skilled labor, focusing on diversifying its workforce.

  • Sustainability Concerns: Questions loom about the long-term sustainability of this growth as power grids near capacity, with potential solutions like nuclear fusion being explored.

💡 This regional shift underscores the intricate balance between technological progress and its socioeconomic ripple effects, suggesting a broader need for planning in high-growth tech zones.

In 2024, three highly anticipated tech products—the Rabbit R1, Apple's Vision Pro, and Humane's Ai Pin—failed to meet expectations, facing criticism for functionality, practicality, and security issues.

  • Rabbit R1: Introduced as a pocket-sized AI companion, the R1 garnered attention at CES 2024 but was later criticized for limited functionality and significant security vulnerabilities.

  • Apple's Vision Pro: This $3,499 mixed-reality headset showcased advanced technology but was deemed too expensive and impractical for widespread consumer adoption.

  • Humane's Ai Pin: Designed to be worn on a lapel, this AI assistant faced critical malfunctions and design flaws, leading to low sales and high return rates.

👎 Despite these setbacks, all three companies continue to update their products, striving to address the issues and improve user experience. Notably, Rabbit seems intent on pivoting to a soft “AI for kids” angle, launching some goofy gamified features into their R1 devices.

 🆕 Updates

📽️ Daily Demo

🗣️ Discourse