Is Manus All Hype? What’s Really Going On?

⚡️ Headlines

🤖 AI

Signal President Meredith Whittaker Warns of Security and Privacy Risks in Agentic AI - Meredith Whittaker, President of Signal, cautions that agentic AI systems pose significant threats to user privacy and security. [TechCrunch].

Beijing Introduces AI Courses for Students to Boost Sector Growth - Beijing schools are set to implement AI courses for primary and secondary students, aligning with China's ambitions to lead in the AI industry. [Bloomberg].

Reflection AI Aims to Build Superintelligent Autonomous Systems - Reflection AI outlines its mission to develop superintelligent autonomous systems, focusing initially on creating an autonomous coding platform. [Reflection AI].

Sudowrite Launches Muse: AI Model Tailored for Fiction Writers - Sudowrite introduces Muse, an AI model specifically designed to assist fiction authors in crafting narratives. [Sudowrite].

🦾 Emerging Tech

Meta Recommends Developers Use Unity and Unreal's Built-In OpenXR Support - Meta advises developers to utilize the native OpenXR support in Unity and Unreal engines for better cross-platform compatibility. [UploadVR].

🤳 Social Media

YouTube Introduces Premium Lite for Ad-Free Viewing at $7.99/Month - YouTube expands its Premium Lite pilot to the US, offering ad-free viewing of most videos for $7.99 per month. [YouTube Blog].

🔬 Research

Chain of Draft Method Could Reduce AI Costs by 90% - Researchers introduce "chain of draft," a technique that significantly cuts AI token usage, potentially lowering operational costs by 90%. [VentureBeat].

Naturally Occurring Molecule Rivals Ozempic in Weight Loss - Stanford Medicine researchers identify a molecule that suppresses appetite and reduces body weight in animals, similar to semaglutide (Ozempic), but without some side effects. [Stanford Medicine].

⚖ Legal

Undocumented Commands Found in Bluetooth Chip Used by a Billion Devices - Researchers discover undocumented commands in the ESP32 microchip, potentially allowing unauthorized access to numerous devices. [BleepingComputer].

AI Claims to Crack CIA's Kryptos Sculpture Code, Artist Disagrees - AI attempts to decipher the enigmatic Kryptos sculpture at CIA headquarters, but the artist refutes these claims. [Wired].

🔌 Plug-Into-This

Manus probably isn't China's second 'DeepSeek moment'

Manus, a Chinese-developed autonomous AI agent, has recently gained significant attention. However, its current performance is hindered by various limitations and bugs, suggesting it may not replicate the groundbreaking impact of DeepSeek.

• Manus is designed to independently analyze, plan, and execute complex tasks without direct human intervention.

• Despite its ambitious design, users have reported several functional issues affecting its reliability and effectiveness.

• The AI community has expressed skepticism regarding Manus's ability to meet its high expectations.

• In contrast, DeepSeek's R1 model demonstrated superior performance at a fraction of the cost, significantly impacting the AI industry and global markets.

• The rapid success of DeepSeek's R1 model led to a substantial sell-off in global equities markets, highlighting its disruptive influence.

🧩 It’s still being hyped on X like it was the “biggest game changer ever” (isn’t everything, though?) but the tone feels a bit strange compared to DeepSeek. It seems only influencers and specific people were given access, so there are a few loud voices saying it’s amazing, but more staying quiet. On the other hand, significant skepticism has been expressed, so we’ll keep an eye on this for now.

Thousands of Cat-Eared Robots Are Waiting Tables in Japan

In response to a significant labor shortage, Japan's restaurants have increasingly adopted cat-eared service robots to assist with waiting tables. This shift aims to maintain operational efficiency and customer satisfaction amid workforce challenges.

• The service robot market in Japan is experiencing rapid growth as businesses seek technological solutions to labor shortages.

• These robots are designed to navigate dining areas, deliver food to patrons, and return used dishes to the kitchen, reducing the workload on human staff.

• Equipped with interactive features, such as expressive cat-like ears and the ability to respond to customer interactions, the robots enhance the dining experience.

• Some establishments have reported increased efficiency and customer engagement since integrating these robotic servers.

• The adoption of service robots aligns with Japan's broader strategy to incorporate automation and robotics into various sectors to address demographic challenges.

🤖 Once a novelty, as Forbes explores, the service robot could very much become the norm in aging populations like Japan as time goes on and (a) people get more used to robots (b) their functionality improves.

Microsoft develops in-house AI models to rival OpenAI's

Microsoft is developing proprietary artificial intelligence (AI) models, aiming to reduce reliance on OpenAI and enhance its own AI capabilities. These in-house models are intended to match or surpass the performance of OpenAI's offerings.

• The initiative is led by Mustafa Suleyman, CEO of Microsoft AI, who envisions conversational AI as transformative as the web browser.

• Microsoft has tested models from xAI, Meta, and DeepSeek as potential alternatives to OpenAI's technology in its Microsoft 365 Copilot.

• The company's AI division has trained a family of models called MAI, performing comparably to leading models from OpenAI and Anthropic.

• Microsoft plans to integrate these “MAI” models into Copilot and may release them as an API for external developers by year-end.

• This diversification seeks to lower costs and dependence on OpenAI's GPT-4, initially used in 365 Copilot.

🧠 Microsoft's development of in-house AI models signifies a strategic move to assert greater control over its AI infrastructure as the underlying schism with OpenAI widens. Will we see a future where every big tech firm has it’s own in-house AI, or will one provider fuel them all eventually? For now, it looks like new firms such as OpenAI and Anthropic simply got themselves a seat at the table, rather than buying the restaurant.

 🆕 Updates

📽️ Daily Demo

🗣️ Discourse