The week's top generative AI updates - July 23, 2024

Welcome to this week’s SymphonyAI generative AI weekly newsletter summarizing all the important AI industry developments and technology advancements you need to know.

Industry projections

• Enterprises Embrace Generative AI, But Challenges Remain: A new survey reveals that enterprises are heavily investing in generative AI, with 73% of respondents planning to spend over $500,000 and 46% over $1 million in the next year.?

• AI Startup Funding More Than Doubles in Q2, Crunchbase Data Shows: Investments in AI startups surged to $24 billion from April to June, more than doubling from the previous quarter, driven by strong interest in AI technologies across sectors. Overall startup funding increased by 16% sequentially, reaching $79 billion.?

• TechInsights: Generative AI to Account for 1.5% of World’s Power Consumption by 2029: AI accelerators will potentially consume 2318 TWh of global power between 2025 and 2029, accounting for 1.5% of the world's electricity consumption, driven by the high computational demands of AI applications.??

• Enterprise Strategy Group Reveals Fewer Than Half of Organizations Achieve GenAI Efficiency Gains: The survey found that 43% of organizations have seen improvements in operational efficiency after deploying generative AI, highlighting challenges in implementation. The survey revealed that 71% believe IT infrastructure needs modernization and cite privacy, latency, and cost as considerations.??

• Tenable: 73% of Indian Enterprises to Use GenAI in the Next 12 Months: The study found that 73% of Indian enterprises plan to adopt generative AI within the next 12 months for security and IT support of broader business goals. GenAI is a potential game-changer for preventive threat response, automation of security measures, and improving actionability.?

Technology news?

• OpenAI Introduces Scale to Measure AI System Progress Toward AGI: OpenAI has developed an internal scale to track the progress of its AI systems towards artificial general intelligence (AGI). ChatGPT is currently at Level 1, with Level 5 representing full AGI capabilities.?

• OpenAI Working on New Reasoning Technology Under Code Name "Strawberry": OpenAI is developing a novel AI project, code-named "Strawberry," to enhance model reasoning capabilities. The project aims to enable AI to navigate the internet autonomously and perform deep research, beyond generating answers to queries.??

• OpenAI Anticipates Decrease in AI Model Costs Amid Adoption Surge:? At VB Transform 2024, OpenAI discussed the significant reduction in inference costs for AI models like GPT-4o, noting an 85-90% decrease since the introduction of GPT-4. OpenAI's latest model, GPT-4o, introduced at the Spring Update event, is half the price and twice as fast as GPT-4 Turbo.???

• OpenAI and Los Alamos National Laboratory Announce Research Partnership: OpenAI and Los Alamos National Laboratory (LANL) partnered to study the safe use of multimodal AI models, such as GPT-4o, in bioscientific research. The collaboration is to develop safety evaluations for AI systems used in laboratory settings, to advance scientific research safely and effectively.??

• Anthropic’s Claude Adds a Prompt Playground to Quickly Improve Your AI Apps: Anthropic introduced a "Prompt Playground" feature within its Claude language model to help developers enhance AI applications. The tool automates parts of prompt engineering, so users can generate, test, and evaluate prompts more efficiently.???

• Anthropic Introduces Fine-Tuning for Claude 3 Haiku on Amazon Bedrock: Anthropic launched fine-tuning capabilities for Claude 3 Haiku on Amazon Bedrock for customizing the AI model for specific tasks such as classification, API interactions, and data interpretation. Early adopters report a 73% increase in positive feedback for agent responses and a 37% improvement in key performance indicators, highlighting the practical benefits of this customization.?

• Claude Artifacts Now Shareable and Customizable for All Users: Claude's Artifact feature now allows sharing with others, even without a Claude account. Artifacts are interactive content pieces, such as code or graphics that can now be shared and reused. This marks a significant step in making advanced AI functionalities more accessible and practical for a broader audience.?

• DeepMind’s PEER Scales Language Models with Millions of Tiny Experts: Google’s DeepMind has introduced?Parameter Efficient Expert Retrieval (PEER), a novel architecture that allows mixture-of-experts (MoE) models to scale to millions of specialized "tiny experts" without increasing computational costs.??

• Google's Search Market Dominance Grows as ChatGPT Falters: Google's search market share increased to 91.1% in June thanks to AI integration, while ChatGPT web visits fell 12. Analysts suggest that Google's AI advancements, despite initial rollout issues, are aiding query growth and increasing monetization across the Google ecosystem, contributing to its market dominance.?

• Gmail Introduces AI-Powered Summaries for Efficient Email Management: Google Gmail rolled out AI-powered summaries to help users quickly understand the content of long emails, for productivity and email management. The new feature uses advanced NLP for concise summaries, so users can grasp key points without reading the entire message.?

• Gemini's Background Mode and App Extensions Enhance AI Capabilities: Google Gemini's new features, including background mode and app extensions, will improve AI capabilities, potentially surpassing Apple's intelligence offerings in versatility and integration. The background mode allows Gemini to operate continuously without user intervention while App extensions give third-party apps the ability to integrate Gemini AI functionality.?

• Google DeepMind Gemini 1.5 Pro Improves Robot Navigation: Google is using Gemini AI to improve robot navigation and task completion. The DeepMind robotics team demonstrated Gemini 1.5 Pro’s long context window for better information processing from natural language instructions.??

• AWS Introduces Standalone Guardrails API and Updates Bedrock: AWS has made its Guardrails a standalone API, enhancing its Bedrock service to offer better support for enterprise generative AI applications. This new API allows users to adjust parameters for compliance and security.?

• AWS App Studio Simplifies Enterprise App Creation with AI: AWS launched App Studio, a generative AI-powered service to build enterprise applications by simply describing needs in natural language to streamline development. The service handles all aspects of application creation, from coding and testing to deployment and maintenance.?

• Amazon Launches Its Fourth-Generation Graviton4 Chip as Competition Intensifies: AWS introduced the fourth-generation Graviton4 processor, with three times the compute power and memory of its predecessor, 75% more memory bandwidth, and 30% better performance. The $544 billion semiconductor market is expected to exceed $1 trillion by 2033, driven by increasing AI demand.??

• Meta Researchers Distill System 2 Thinking into LLMs, Improving Performance on Complex Reasoning: Meta's FAIR team has developed a "System 2 distillation" technique to enhance large language models (LLMs) by teaching them complex tasks without requiring intermediate steps, thus improving efficiency and performance on reasoning tasks.??

• Meta AI Develops Compact Language Model for Mobile Devices: Meta AI researchers have unveiled MobileLLM,?an efficient language model designed for smartphones and other resource-constrained devices. This approach challenges the assumption that effective AI models must be large, focusing instead on optimizing models with fewer than 1 billion parameters.?

SymphonyAI news

• Partner news: L&T Technology Services and SymphonyAI partner to provide AI-based business transformation to global customers through Apex Enterprise Copilot?

• News coverage, Crunchbase H2 Forecast: Who’s Next To IPO???

• Blog: What is the 50 percent rule in financial crime prevention/sanctions screening?? FAQ and everything you need to know?

• Byline in Streaming Media: AI has become the secret weapon for FAST (free, ad-supported streaming TV) channels competing in a crowded market?