Sponsored by

Today:

Learn how to make AI work for you.

How do you stay up-to-date with the insane pace of AI? Join The Rundown – the world’s fastest-growing AI newsletter with over 500,000+ readers learning how to become more productive using AI every morning.

1. Our team spends all day researching and talking with industry experts.

2. We send you updates on the latest AI news and how to apply it in 5 minutes a day.

3. You learn how to become 2x more productive by leveraging AI.

Subscribe with one click.

AI Unicorn Anthropic Releases Claude 3, A Model It Claims Can Beat OpenAI’s Best

Anthropic, a tech company, claims its latest AI model, Claude 3, beats rivals like GPT-4 and Google's Gemini 1.0 Ultra. With three versions varying in performance and cost, Claude 3 aims at business solutions, excelling in tasks like data analysis and biomedical research. 

Opus, the top-tier model, outshines competitors but comes at a higher price. Sonnet, the mid-tier, suits most tasks, while Haiku, the cheapest, works for live interactions and content moderation. Anthropic plans to enhance features like code interpretation and search functions. Despite concerns, the company focuses on business clients, with early users reporting significant improvements in productivity.

FORBES

OpenAI signs open letter to build AI responsibly just days after Elon Musk sued the company for putting profit ahead of people

Tech giants like OpenAI and Salesforce have pledged to develop artificial intelligence (AI) responsibly in response to growing concerns about its societal impact. Led by venture capitalist Ron Conway, they emphasize the need to maximize AI's benefits while mitigating risks. The move comes after Elon Musk sued OpenAI, alleging it prioritized profit over humanity. 

Despite the optimism, specific actions remain unspecified. The letter highlights AI's potential akin to historical innovations but stresses human responsibility in shaping its impact. This commitment reflects a broader industry dialogue on AI's ethical implications, spurred by initiatives like OpenAI's ChatGPT.

FORTUNE

Microsoft derides ‘doomsday futurology’ of New York Times’ AI lawsuit

Microsoft has accused The New York Times of engaging in 'doomsday futurology' by predicting that ChatGPT could ruin the news business in a lawsuit that claims copyright infringement by OpenAI, ChatGPT's creator. Microsoft likened The New York Times to Hollywood studios that tried to stop the introduction of the VCR in the 1980s, claiming that the media organization is holding back a 'groundbreaking new technology'.

The Times filed a lawsuit in December against OpenAI and Microsoft, accusing them of copying millions of articles without permission to build their AI chatbots. In response, Microsoft argued that copyright law is not an obstacle to large language models and that the content used to train tools like ChatGPT does not supplant the market for the works.

FINANCIAL TIMES

Microsoft’s Mistral deal beefs up Azure without spurning OpenAI

Microsoft's investment of €2 billion ($2.1 billion) into French AI company Mistral indicates a long-term commitment to the AI sector. The deal focuses on integrating Mistral's latest AI model, Mistral Large, into Microsoft's Azure platform, expanding its model offerings. 

While Microsoft remains heavily invested in OpenAI, the move signals a strategic push to establish Azure as a prominent AI ecosystem. By partnering with Mistral, popular among developers, Microsoft aims to showcase the diversity of choices available on Azure. Additionally, the investment allows Microsoft to tap into the European AI market and navigate impending regulatory challenges posed by the EU's AI Act.

THE VERGE

Wix’s new AI chatbot builds websites in seconds based on prompts

Wix’s new AI chatbot swiftly constructs websites with just prompts, making site creation a breeze. Free to start, but premium plans offer advanced features like payment acceptance and personalized domains, ranging from $17 to $159 monthly. To begin, click "Create with AI," answer a few questions, and within seconds, voila! 

The site's basic, but editable with conventional tools. AI assists with image and text suggestions. Despite its ease, customization beyond basics requires deeper web knowledge. Nevertheless, for a quick start, Wix's AI chatbot proves handy, bridging the gap between novice website creators and advanced developers.

THE VERGE

OpenAI adds ‘Read Aloud’ voiceover to ChatGPT, allowing it to speak its outputs

OpenAI updates ChatGPT with a "Read Aloud" feature, allowing users to hear responses on mobile and web. Despite rumors of GPT-5, it's a minor enhancement amid competition from Anthropic and Google. Activated by holding ChatGPT's response, the feature aids hands-free usage and comprehension. 

OpenAI aims to keep pace with advancements while facing lawsuits and criticism. The update targets convenience, especially for mobile users, in a competitive landscape of evolving AI models.

VENTUREBEAT

Access to the most powerful Anthropic AI models begins today on Amazon Bedrock

Amazon Bedrock now offers Anthropic's Claude 3 AI models, boasting top-tier accuracy, speed, and cost-effectiveness. These models, like Claude 3 Opus and Claude 3 Sonnet, promise near-human intelligence and enhanced vision capabilities. Anthropic claims superiority over competitors like OpenAI's GPT-4, especially in tasks like coding and multilingual understanding. 

With Claude 3, customers can automate tasks, develop user-facing applications, and expedite research across various sectors. Already embraced by global giants like Pfizer and the PGA TOUR, these models are set to revolutionize AI applications. This move strengthens AWS's commitment to democratizing AI and fostering innovation across industries.

AMAZON

🧠RESEARCH

VisionLLaMA: A Unified LLaMA Interface for Vision Tasks

VisionLLaMA is a vision transformer model inspired by LLaMA, initially designed for processing text. VisionLLaMA adapts the transformer architecture for 2D image tasks, offering a unified framework for various vision tasks. It demonstrates superior performance in image generation and understanding, potentially becoming a new standard model.

Learning and Leveraging World Models in Visual Representation Learning

Image World Models (IWM) is an extension of Joint-Embedding Predictive Architecture (JEPA), which leverages world models for self-supervised learning. IWM predicts global photometric transformations, offering a broader approach than masked image modeling. Finetuning IWM yields competitive results across diverse tasks, enabling control over representation abstraction levels.

Resonance RoPE: Improving Context Length Generalization of Large Language Models

Resonance RoPE is addressing the train-short-test-long (TSTL) challenge in Large Language Models (LLMs) equipped with Rotary Position Embedding (RoPE). It refines RoPE interpolation for out-of-distribution positions in longer sequences, significantly improving model performance without additional computational costs. Experiments demonstrate enhanced recognition of out-of-distribution positions and superior performance in language modeling tasks.

AtP*: An efficient and scalable method for localizing LLM behaviour to components

Attribution Patching (AtP*) as an efficient method for attributing behavior to components in Large Language Models (LLMs). It addresses limitations of previous methods, improving scalability and reducing false negatives. AtP* outperforms alternative approaches, providing significant enhancements in computational efficiency and accuracy. Additionally, it offers a method to bound the probability of false negatives in estimates.

RealCustom: Narrowing Real Text Word for Real-Time Open-Domain Text-to-Image Customization

RealCustom is a method for real-time text-to-image customization that disentangles subject similarity from text controllability. It gradually narrows the influence of real text words to relevant parts only, enhancing both subject similarity and text control. RealCustom's "train-inference" framework achieves superior real-time customization performance, achieving unprecedented balance between subject similarity and text controllability.

🛠️TOP TOOLS

Ubique - enhancing sales outreach with AI-powered video messages

Klap - Turn videos into viral shorts

Text Humanizer - Free Advanced AI Text Content Humanizer Tool

Localllm - Run LLMs locally on Cloud Workstations.

Fondo -  file your Delaware Franchise Tax for you

Frame - AI-powered glasses that offer visual analysis, live web search, and real-time translation, enhancing how users interact with their environment.

Swizzle - a platform for building web apps, internal tools, and dapps directly in your browser without any setup required

NotesOllama -  allows users to interact with local LLMs directly within Apple Notes, enabling functionalities like summarizing notes, asking questions, and crafting prompts without exiting the ap

📲SOCIAL MEDIA

🗞️MORE NEWS

Google’s morale crisis is about to get worse

Google is facing internal challenges with ongoing layoffs, troubles with Gemini, and anticipated lower raises for employees. The situation is expected to worsen, contributing to a morale crisis within the company. THE VERGE

Why OpenAI’s nonprofit mission to build AGI is under fire — again | The AI Beat

Elon Musk files lawsuit against OpenAI, CEO Sam Altman, and President Greg Brockman, alleging breach of contract and fiduciary duty, claiming they prioritized profit over their nonprofit mission to develop artificial general intelligence (AGI). OpenAI's complex nonprofit and for-profit structure, involving Microsoft, raises concerns about governance and conflicts of interest. Legal experts see Musk's lawsuit as highlighting broader issues with OpenAI's structure and mission. VENTUREBEAT

India reverses AI stance, requires government approval for model launches

India has shifted its stance on AI regulation, requiring significant tech firms to obtain government approval before launching new models. The advisory also mandates products to avoid bias, discrimination, or electoral process threats. Though not legally binding, it signals forthcoming regulation, marking a reversal from India's previous hands-off approach. TECHCRUNCH

JPMorgan’s AI-Aided Cashflow Model Can Cut Manual Work by 90%

JPMorgan's AI-powered cashflow management tool has reduced manual work for corporate clients by nearly 90%. The bank, considering charging for the service, has witnessed significant interest from clients and plans to further develop the solution. About 2,500 clients are currently using the AI tool since its launch a year ago. BLOOMBERG

China offers AI computing ‘vouchers’ to its underpowered start-ups

China is providing AI computing "vouchers" to its fledgling startups lacking computational power. The initiative aims to bolster the country's AI sector, offering support to underpowered startups. Access to AI resources is crucial for their development and competitiveness in the tech landscape. FINANCIAL TIMES

Japan’s Nikkei Tops 40000 for First Time, Driven by AI Optimism

Japan's Nikkei stock index surpasses 40000 for the first time, up 20% this year. Fueled by strong corporate profits and optimism in AI investments, chip-related stocks like Renesas Electronics and Advantest surged. The milestone reflects market confidence in technology's future role. THE WALL STREET JOURNAL

This Startup Wants To Power The Next Big Wave Of AI: Actually Making It Useful

Baseten secures $40 million investment to streamline AI deployment for practical applications. The startup simplifies launching pre-trained AI models into public use, optimizing computing resources based on demand. Focused on inference, it sees significant revenue growth, attracting major clients like Patreon. Competition in AI infrastructure intensifies as investors back similar ventures. FORBES

TurboTax and H&R Block now use AI for tax advice. It’s awful.

TurboTax and H&R Block introduced AI chatbots for tax advice, but in tests, they proved unhelpful or wrong half the time. Questions yielded irrelevant or inaccurate responses, raising concerns about relying on AI for critical financial decisions. Despite disclaimers, users face risks; both companies promise assistance if AI errors lead to audits. THE WASHINGTON POST

Behind big pharma is big intelligence

Consulting firm Lifescience Dynamics aids pharmaceutical companies with AI-driven intelligence to expedite drug development and FDA approval. With tools incorporating AI technologies, they streamline data collection, analysis, and visualization, reducing manual labor and errors. Despite regulatory and data-sharing challenges, AI proves instrumental in enhancing industry efficiency and insight. FORTUNE

Meet Copilot for Finance, Microsoft's latest AI chatbot - here's how to preview it

Microsoft introduces Copilot for Finance, an AI chatbot aimed at streamlining financial tasks like variance analysis and collections. Integrated with Microsoft 365, it pulls data from ERP systems and provides natural-language summaries of financial insights, offering potential time and cost savings for businesses. Public preview available with pricing to be disclosed later. ZDNET