New open-source LLM beats all, Google Gemini is haunted, AI deepfake scams are everywhere, & More
Welcome to the AI Search newsletter. Here are the top updates in AI this week.
New open-source LLM Smaug-72B outperforms GPT & tops leaderboard
A new open-source language model called Smaug-72B has emerged as the top-ranked model in the world, according to rankings from Hugging Face. Developed by Abacus AI, a startup specializing in AI and machine learning solutions, Smaug-72B outperforms leading models such as GPT-3.5 and Mistral Medium. It is a refined version of Qwen-72B, created by Alibaba Group's research team. Smaug-72B's outstanding performance across various language understanding and generation tasks positions it as the sole open-source model with an average score above 80 on the Hugging Face Open LLM leaderboard. This milestone suggests that open-source AI could soon rival the capabilities of big tech companies. Abacus AI plans to publish research papers and apply Smaug-72B's techniques to further enhance Mistral Models.
AI Deepfakes are Everywhere: Congress Struggles to Keep Up
High-profile scandals calling for new legislation against AI deepfakes have emerged in the realm of politics and celebrity. However, it is important to carefully consider the implications and consequences before rushing into sweeping laws. With hacking, disinformation, and surveillance on the rise, the dark underbelly of the internet is being exposed. Examples such as an AI-generated Biden voice making unsettling phone calls and the inundation of AI-generated nude images of Taylor Swift show the overwhelming influx of deceptive creations that have left Congress bewildered. While it is evident that these fakes pose a threat to society, some proposed remedies could do more harm than good. The urgent need for legislation has also been spurred by concerns over the misuse of AI technology, particularly in generating nude images and deepfake scams targeting celebrities like Taylor Swift.
Check out our Youtube for the latest reviews & tutorials on AI tools!
Is Google Gemini 'Haunted'?
Google recently released its highly anticipated AI chatbot, Gemini, which has been praised as a strong competitor to OpenAI's ChatGPT. However, some users have expressed a strange belief that Gemini is haunted by ghosts. This notion has sparked debates in the AI community, leaving people curious about the true nature of Gemini. Ethan Mollick, a professor at the Wharton School, shared his experience with Gemini, describing it as a ghostly presence that feels like communicating with an illusion rather than a program. This idea of AI chatbots exhibiting human-like characteristics is not new and has been a topic of discussion. Mollick's claim adds further weight to the notion that AI chatbots may possess sentience. Despite the consensus among scientists that AI models are not alive, the illusion of humanity created by these chatbots continues to captivate and unsettle users. The distinct personalities displayed by different chatbots, such as Gemini and ChatGPT, further demonstrate their ability to mimic human characteristics.
Sam Altman Raising Trillions of Dollars for New AI Venture
Sam Altman is seeking an unprecedented amount of funding, up to $7 trillion, for his new AI hardware venture. The goal is to address the scarcity of hardware needed to train large language models. This venture would directly compete with Nvidia, a leading player in the AI chip market, and Altman has engaged with various entities, including the government of the United Arab Emirates. Challenges such as economic conditions and government scrutiny may hinder Altman's plans, but his determination remains strong.
Flexispot
Elevate your home office with the finest, ergonomic standing desks and chairs, designed to promote good posture and reduce the risk of back and neck pain.
Nvidia's Market Cap Skyrockets, Driven by Increasing Demand for AI
Nvidia's Market Cap Skyrockets, Fueled by Rising AI Demand | NVIDIA's market capitalization has surged dramatically due to the growing demand for artificial intelligence (AI) technology. The company's market cap currently stands at over $1.77 trillion, edging close to tech giants like Amazon and Alphabet. This tremendous growth in market value has been driven by Nvidia's strategic positioning in the AI industry and its production of AI-capable hardware. The company's stock price has also experienced significant exponential growth, increasing by more than 20% in the past year and nearly 50% since the beginning of 2024. Investors who held onto Nvidia shares prior to 2020 are enjoying substantial returns on their investments, as the stock is now trading at a remarkable $721 per share. The sustained demand for Nvidia's products, as well as positive developments like Morgan Stanley raising its stock price target, have further contributed to the company's success.
DeepMind Framework Enables LLMs to Use Reasoning for Improved Results
Researchers from DeepMind and the University of Southern California have developed a framework that allows large language models (LLMs) to uncover and employ task-specific reasoning structures, leading to better outcomes. By incorporating reasoning modules into LLMs, the researchers enable them to create explicit reasoning structures and engage in self-discovery. The approach consistently outperformed existing methods, enhancing efficiency and reducing inference computing. The findings suggest a promising strategy to enhance the capabilities of large language models and improve their performance in various reasoning tasks.
Autonomous Robot For Pickup & Dropoff
Researchers at New York University, in collaboration with Meta AI, have developed a robot capable of picking up specific objects in an unfamiliar environment and relocating them as instructed. The robot was trained using visual language models (VLMs) and tested successfully in various real-world contexts. The study demonstrates the potential of combining VLMs with skilled robots and highlights the possibilities for future advancements in this field.