In partnership with

Intercom for Startups

Join Intercom’s Early Stage Program to receive a 90% discount.

Get a direct line to your customers. Try the only complete AI-first customer service solution.

Apply now

News

Google DeepMind Unveils AI Innovations at ICML 2024

The 2024 International Conference on Machine Learning (ICML) in Vienna, Austria, from July 21-27, will feature significant contributions from Google DeepMind.

This conference is a major event for showcasing advancements, exchanging ideas, and shaping the future of AI research. Google DeepMind will present over 80 research papers and demo new AI models at their booth.

Key Takeaways:

• Defining AGI: Google DeepMind will present a framework to classify AGI capabilities and behaviors, emphasizing the importance of open-ended systems for generating new discoveries.

• Efficient and Responsible AI Scaling: New methods for scaling AI, aligning models with human preferences, and safeguarding privacy will be showcased, including classification techniques for deep reinforcement learning and a game-theory-based approach for fine-tuning large language models (LLMs).

• Generative AI and Multimodality:

  • VideoPoet: A large language model for zero-shot video generation from multimodal inputs.

  • Genie: A model generating playable environments based on text prompts, images, photos, or sketches.

  • MagicLens: An image retrieval system using text instructions to find images with complex relations.

• Supporting AI Diversity: Google DeepMind sponsors initiatives led by Disability in AI, Queer in AI, LatinX in AI, and Women in Machine Learning.

Google DeepMind's extensive participation at ICML 2024 highlights their commitment to advancing AI technology, exploring the potential of AGI, and promoting ethical and responsible AI development.

GPT-4o Mini: Revolutionizing Cost-Efficient AI

OpenAI has launched GPT-4o mini, a highly cost-efficient small model designed to make AI accessible and affordable. Priced at 15 cents per million input tokens and 60 cents per million output tokens, GPT-4o mini is significantly cheaper than previous models. It excels in various tasks with low cost and latency, supporting text and vision inputs, and soon image, video, and audio. With a 128K token context window and up to 16K output tokens per request, it offers superior performance in reasoning, math, and coding.

Key Features:

• Cost-Efficient: Priced much lower than previous models.

• Multimodal Support: Handles text and vision, with future support for more formats.

• High Performance: Scores 82% on MMLU, outperforms competitors in reasoning, math, and coding.

• Large Context Window: Supports 128K tokens, enabling extensive data processing.

• Improved Non-English Handling: Enhanced tokenizer for cost-effective multilingual support.

GPT-4o mini's affordability and efficiency open new avenues for developing a broad range of AI applications, from customer support chatbots to complex data processing tasks, democratizing access to advanced AI capabilities.

Nvidia and Mistral's NeMo: Elevating AI with Advanced Capabilities

Mistral AI announced the release of Mistral NeMo, a cutting-edge 12-billion parameter model developed in collaboration with NVIDIA. Mistral NeMo excels in reasoning, world knowledge, and coding accuracy, featuring a large context window of up to 128k tokens and easy integration into existing systems.

Key Features:

• Large Context Window: Supports up to 128k tokens.

• Multilingual Proficiency: Strong performance in 11 major languages.

• Advanced Tokenizer: Tekken tokenizer offers ~30% more efficient compression.

• Instruction Fine-Tuning: Enhanced for precise instructions and multi-turn conversations.

• Quantisation Awareness: Enables FP8 inference without performance loss.

Mistral NeMo sets a new benchmark for AI models in terms of versatility and efficiency, providing robust support for multilingual applications and advanced coding tasks. Its release under the Apache 2.0 license encourages widespread adoption and innovation in AI research and enterprise solutions.Explore Mistral NeMo’s capabilities on HuggingFace and try it now with mistral-inference and mistral-finetune.

Microsoft Unveils SpreadsheetLLM for AI-Driven Spreadsheet Analysis

Microsoft’s new AI system, SpreadsheetLLM, is set to transform enterprise data management. Designed to understand and work with the complex format of spreadsheets, SpreadsheetLLM combines large language models (LLMs) with structured spreadsheet data. It employs an innovative pipeline to compress and encode spreadsheets, enhancing LLMs' ability to analyze and reason over complex data efficiently.

Key Features:

• Powerful Encoding: Utilizes SheetCompressor to optimize performance and achieve state-of-the-art accuracy.

• Natural Language Integration: Enables LLMs to generate and interpret spreadsheets from natural language prompts.

• Enhanced Decision-Making: Facilitates AI-assisted data analysis and decision-making in enterprises.

SpreadsheetLLM opens new avenues for AI-driven data management, making it easier for enterprises to analyze and leverage spreadsheet data for informed decision-making.

Business

Proton Scribe: Revolutionizing Email Productivity with Privacy

Proton introduces Proton Scribe, a smart, privacy-first writing assistant integrated into Proton Mail, designed to help businesses compose and proofread emails efficiently. This tool ensures your data remains secure while boosting productivity. Proton Scribe can be run locally, guaranteeing that your sensitive information never leaves your device.

Key Features :

• Email Drafting: Quickly generate professional email drafts by simply entering your topic.

• Proofreading and Formalization: Improve drafts by eliminating typos and adjusting the tone for formal communication.

• Privacy-First: Built with zero-access encryption, ensuring no data is logged or saved.

• Local Operation: Run entirely on your device, ensuring maximum data security.

Proton Scribe combines productivity and privacy, allowing businesses to maintain high security standards without sacrificing efficiency.

Do It Yourself (DIY)

Unlocking the Power of Writesonic for Entrepreneurs

Harness the power of Writesonic to enhance your business content!

Here’s how:

Website Content Refresh: Use Writesonic to revitalize your website by updating sections or generating fresh ideas for product pages and “About Us” sections.

• Product Descriptions: Quickly create compelling product descriptions.

👉 Click here to Discover how Writesonic can transform your content strategy and help your business thrive.

Healthcare

AIHealthTech Insider - Issue #6: Transforming Healthcare with AI

Stay at the forefront of medical innovation with AIHealthTech Insider, your ultimate source for groundbreaking AI advancements in healthcare. Learn how AI is revolutionizing early Alzheimer’s detection, enhancing cancer diagnostics, and creating 4D heart scans for precise medical imaging.

🔗 Subscribe now for FREE weekly updates!

Don't miss out on the future of health and AI! ➡️

Event

International Conference on Machine Learning (ICML)

• The 41st ICML conference will take place from July 21 to July 27, 2024 in Vienna, Austria at the Messe Wien Exhibition Congress Center.

• ICML 2024 will be a physical conference with some streaming elements. It includes sessions, workshops, tutorials, and an expo.

• You can find more details and register on the official ICML website.

Giveaway

🎉 We're Giving Away Free AI Software, Courses, And Amazon Gift Cards!

(Enter to Win With 1- Click)

This $2,000+ AI software bundle includes AI tools, dozens of courses, Amazon gift cards, premium templates, and more.

It could 10X your personal and business goals.

And We're Giving It ALL Away For FREE!

🎉 👇Enter For Free With 1-Click Today! 👇

Startup Success: Instant Access to Free AI Entrepreneur Courses!