_{Welcome back for the Latest AI Drops!}

Today’s Drops:

• Tesla Releases A New Bot "Optimus-Gen 2"

• Meta's AI For Ray-Ban Smart Glasses

• Microsoft just released Phi-2, Small Language Models (SLM)

• Trending on X “Channel 1’s AI Generated News Anchors”

• Trending Startup “Korus”

• AI Headlines: Ashley's AI campaign tool sparks disinformation concerns. Meta faces lawsuit for training AI with pirated books. Snapchat Plus introduces AI-powered image creation. Essential AI secures $56.5M for "Enterprise Brain," backed by tech giants.

• Trending GitHub Projects

• AI Tools

Read Time: 4.5 minutes

Tesla Releases A New Bot "Optimus-Gen 2"

Tesla has introduced "Optimus Gen 2," a new version of its humanoid robot to automate repetitive tasks. Initially, many were skeptical of Optimus, seeing it as an underdeveloped concept with unimpressive demonstrations, including a basic prototype that could barely walk and wave.

However, Tesla's vision for the robot is to use AI, batteries, and electric motor expertise to create a humanoid robot capable of replacing human labour in specific tasks. The company's progress in this area was evident at their 2023 shareholders meeting, where they showcased more advanced prototypes performing valuable tasks.

Recent updates indicate significant advancements in Optimus:

• Enhanced Mobility: The robot walks 30% faster than the previous version.

• Weight Reduction: Optimus Gen 2 has slimmed down by 22 lbs, making it more efficient.

• Smoother Movement: Improved limb movements for more lifelike precision.

• Advanced Components: It features Tesla-designed actuators and sensors, including a 2-Dof actuated neck, integrated electronics in actuators, foot force/torque sensing, articulated toe sections, and human foot geometry.

• Improved Hands: The new hands are faster, with 11 degrees of freedom and tactile sensing on all fingers, capable of handling delicate objects like eggs.

Tesla plans to use Optimus in its manufacturing operations before selling it commercially. Elon Musk, Tesla's CEO, has high expectations for Optimus, believing it could become a significant part of Tesla's value in the long term.

However, the success of Optimus hinges on Tesla's Full Self-Driving (FSD) technology, which is the basis for the robot's AI. As FSD's effectiveness is still uncertain, the viability of Optimus remains to be seen. This uncertainty has led to optimism and skepticism about the robot's future capabilities and applications.

Check out the launch video.

Meta's AI For Ray-Ban Smart Glasses

Meta has introduced innovative AI features for its Ray-Ban smart glasses, enhancing their capabilities significantly. Users can activate a virtual assistant by saying, "Hey, Meta." This assistant utilizes the glasses' camera and microphones to observe and listen to the user's surroundings, enabling various functionalities:

1. Object Identification

2. Language Translation

3. Image Captioning

4. Summarization

Meta is rolling out these features as part of an early access test, initially limited to a small group of opt-in participants in the United States. Meta's CTO, Andrew Bosworth shared the announcement and details about the test phase, and the capabilities of the AI were showcased in various demonstrations, including an Instagram reel by Mark Zuckerberg.

These advancements position Meta's Ray-Ban smart glasses as a multi-functional tool, integrating AI into everyday wearables to provide both practical and innovative uses.

Source

Microsoft just released Phi-2 another small language models (SLM)

Microsoft has unveiled Phi-2, a small language model tailored for laptops and mobile devices. Despite its relatively small size with 2.7 billion parameters, Phi-2 showcases outstanding capabilities in reasoning and language comprehension, often outperforming significantly larger models across various benchmarks. Its achievements can be attributed to two main factors: high-quality training data emphasizing "textbook-quality" content and innovative scaling techniques that draw from Phi-1.5's knowledge to boost performance. Notably, Phi-2 exhibits superior behaviour concerning toxicity and bias and excels in multi-step reasoning tasks such as coding and mathematics. Researchers can access Phi-2 for non-commercial research purposes through a custom Microsoft Research License.

As seen above,Phi-2 surpasses other models, including Mistral and Llama-2, on various benchmarks, especially in multi-step reasoning tasks like coding and math. Its performance is also compared favorably with Google Gemini Nano 2.

Source

Trending on X “Channel 1’s AI Generated News Anchors”

Trending Startup “Korus”

Korus, a startup co-founded by DJ Deadmau5, offers an AI-powered music creation platform that lets users remix and monetize music using AI and licensed stems from artists and labels. Its key features include the "Sound Mosaic" composition algorithm, interactive visuals, and video recording tools. The platform also introduces AI companions for music generation and a soft currency, "NOIZ," for in-platform transactions. Korus aims to blend music with other entertainment sectors, as evidenced by its partnership with Netflix's "Black Mirror" for a web3 experience. Since its launch, Korus has seen over 15,000 songs created and minted by users.

Source

More AI News

Ashley, an AI campaign caller developed for Shamaine Daniels' campaign, offers personalized voter engagement in multiple languages but raises concerns about disinformation and election integrity, prompting ethical and legal considerations. Link

Meta faces a lawsuit for using pirated books to train its AI language model, Llama, despite legal risks highlighted by its own lawyers. Link

Snapchat Plus now offers subscribers AI-generated image creation tools, including the ability to send AI-crafted snaps and selfies with friends.Snapchat Plus subscribers can now also ‘extend’ a photo by using AI to fill in the surrounding environment. Link 

Essential AI has raised $56.5 million in Series A funding to develop the “Enterprise Brain”.This funding round, includes notable investors like AMD, Google, NVIDIA, and others.Based in San Francisco, CA and founded in 2023, Essential AI’s mission is to deepen the partnership between humans and computers, unlocking collaborative capabilities that far exceed what could be achieved by humans alone. Link

Welcome to the innovative world of GitHub projects!

Patch Fusion is an End-to-End Tile-Based Framework for High-Resolution Monocular Metric Depth Estimation

SkyPilot is a framework for running LLMs, AI, and batch jobs on any cloud, offering maximum cost savings, highest GPU availability, and managed execution.

LinyiLYi/voice-assistant is a simple Python script for conversing with a large language model using voice input.

EfficientSAM provides a method for efficient image segmentation using masked image pre-training.

Papermage is a library designed to support NLP (Natural Language Processing) and CV (Computer Vision) research on scientific papers. It offers tools for processing, representing, and manipulating documents, especially in PDF format.