The week's top generative AI updates - July 16, 2024

Welcome to the SymphonyAI generative AI weekly newsletter summarizing important generative AI industry developments and technology advancements. The notable highlights for this week include:??

Industry market predictions?

• Bloomberg Intelligence Predicts Generative AI Sales Could Soar 2,040%: Bloomberg Intelligence predicts generative AI revenue will reach $1.3 trillion in 2032, a significant increase from $63.5 billion in 2023, highlighting massive growth potential.?

• SkyQuest: AI Platform Market Sees Explosive Growth to USD 366.93 Billion by 2031: SkyQuest predicts that the AI platform market will to grow from $41 billion in 2023 to $367 billion by 2031, with a CAGR of 31%, driven by advancements in AI technology.?

• McKinsey & Company estimates generative AI could contribute $7.9 trillion annually to the global economy, underscoring its transformative impact across various industries and prompting strategic investor interest. Meanwhile, MIT economist Daron Acemoglu predicts only a 0.5% rise in labor productivity and 0.9% GDP growth in the US over the next decade, doubting rapid AI advancements.?

• AI Models to Cost $1 Billion to Train, $100 Billion Models Coming: Anthropic’s CEO says that AI models currently cost up to $1 billion to train, with projections that future models may cost $10 to $100 billion by 2025-2027, driven by advancements in AI technology and hardware. Training costs for current models like ChatGPT-4o are around $100 million, requiring extensive computational resources, including thousands of GPUs, with demand for AI-specific hardware increasing rapidly.?

• Natural Language Understanding (NLU) Market Worth $62.9 Billion by 2029: MarketsandMarkets estimates that the NLU market will grow from $19.2 billion in 2024 to $62.9 billion in 2029, with a CAGR of 26.8%, driven by increasing consumer demand.?
• IDC Survey: GenAI to Transform Tech Decisions, Operations, and Customer Engagement: An IDC survey reveals that generative AI will impact technology decision-making, financial and operational systems, and customer engagement, transforming competitive positioning and business models.?
• AI in Retail Market to Surpass USD 71.23 Billion by 2031: The AI technology used in retail will grow from $7.42 billion in 2023 to $71.23 billion by 2031, driven by technological advancements and increased adoption. AI technologies, like machine learning and natural language processing, will optimize pricing strategies, provide personalized customer experiences, and enhance operational efficiency in retail environments.??

AI Investments?

• AI’s Trillion-Dollar Bet: Tech Giants Gamble Big on Uncertain Returns: Tech giants will likely invest over $1 trillion in AI infrastructure, but experts question the profitability, citing modest expected increases in productivity and GDP growth. Chip and power supply shortages, including high-bandwidth memory and specialized packaging technologies, could limit AI growth as demand exceeds supply.?

• AI Industry Needs $600 Billion Annually to Cover Hardware Costs: AI companies must generate $600 billion annually to cover their extensive AI infrastructure costs, raising concerns about a potential AI financial bubble due to unsustainable investment levels.?

Technology?

• Google DeepMind’s JEST Speeds Up AI Training by 13x While Slashing Computing Needs: Google DeepMind’s JEST method makes AI model training for image and text processing 13 times faster, using only a tenth of the computing power required by traditional methods.?
• Google’s ImageInWords Could Boost Everything from Image Search to Text-to-Image AI: Google’s ImageInWords (IIW) system aims to enhance image descriptions by combining precise AI-generated instructions with detailed human annotations, significantly improving image model accuracy.?
• ChatGPT Accidentally Shared Its Secret Rules: ChatGPT inadvertently revealed internal instructions to a user, highlighting embedded guidelines for maintaining safety and ethical boundaries and sparking discussions about AI transparency and security.?

• OpenAI to Join Washington Lobbying Group BSA | The Software Alliance: OpenAI joins BSA | The Software Alliance to expand influence as global policymakers grapple with AI regulation, highlighting the company’s strategic move to shape AI policy and regulatory frameworks.?

• ChatGPT for Mac Now Available to All: OpenAI announced the ChatGPT app for macOS, allowing all users to easily access the AI chatbot using the keyboard shortcut Option + Space, enhancing user convenience and accessibility.?

• Apple Launches Public Demo of Its ‘4M’ AI Model: The?4M (Massively Multimodal Masked Modeling)?model can create images from text, detect objects, and manipulate 3D scenes using natural language, emphasizing its versatile AI applications.?

• Apple Poised to Get OpenAI Board Observer Role as Part of AI Pact: Apple will gain an observer role on OpenAI’s board, enhancing collaboration and insights into AI developments, aligning with Microsoft’s involvement in OpenAI as a major supporter and technology provider.?

• Microsoft Launches Phi-3 Mini-128K-Instruct AI Model: Microsoft’s Phi-3 Mini-128K-Instruct is a 3.8 billion-parameter AI model trained with Phi-3 datasets, focusing on high-quality, reasoning-dense properties. Intended for commercial and research use, the model excels in memory/compute-constrained environments and latency-bound scenarios, providing strong reasoning for tasks like coding and mathematics.?
• Windows 11 to Integrate More AI, Boosting Third-Party Apps: Microsoft plans to enhance Windows 11 with more AI features, supporting third-party developers to integrate AI into their applications, and using powerful NPUs for local and cloud-based AI tasks.??
• Anthropic Looks to Fund a New, More Comprehensive Generation of AI Benchmarks; Anthropic launched a program to fund the development of new AI benchmarks, aiming to evaluate the performance and impact of AI models, including generative models like Claude.?The program seeks to create benchmarks assessing AI’s abilities in areas such as cyberattacks, enhancing weapons of mass destruction, and manipulating people, focusing on AI security and societal implications.??

• Anthropic Advocates for Third-Party AI Model Evaluations: Anthropic announces an initiative for third-party evaluations of AI models to assess capabilities and risks, addressing the need for standardized assessment methods in the AI community. Evaluation areas include AI safety level assessments, advanced capability metrics, and infrastructure for developing evaluations, aiming to effectively measure and mitigate various AI risks.?

• ?Meta Drops AI Bombshell: Multi-Token Prediction Models Now Open for Research: Meta released pre-trained models using a multi-token prediction approach, potentially revolutionizing large language model development and deployment with enhanced performance and reduced training times.?

• Meta to Release Largest Llama 3 Model with 400 Billion Parameters: Meta plans to release its largest Llama 3 model in late July/early August, boasting over 400 billion parameters, significantly surpassing the scale and capabilities of its previous models and competitors.?

• Meta Introduces AI Model for 3D Object Retexturing: Meta launched its 3D Gen model, which is capable of generating high-fidelity 3D images from text in under a minute, with additional retexturing capabilities for both generated and artist-produced images.

SymphonyAI news?

• Webinar Tue, Jul 30 7:30 AM - 8:30 AM PDT: Cut through the hype – unlock true enterprise AI value. Join Constellation Research’s?Ray Wang, Microsoft’s?Eve Psalti, MIT’s Dr.?Daniela Rus, and SymphonyAI CTO?Raj Shukla?for a lively discussion about real world enterprise success with AI and what predictive and generative AI innovations we should expect in the near future. Register?here.?

• Case study: Learn how industrial giant Nippon Gases prevents unplanned downtime with AI. Watch the video?here.?

• South African bank ABSA recognized by ICA Compliance Awards?for its work using predictive and generative AI to combat financial crime.?

• GigaOm speaks with SymphonyAI CEO Sanjay Dhawan: Gigam CEO Ben Book sits down to speak with Sanjay to discuss leadership and the market landscape of technology.?

• Reuters:?SymphonyAI targets second half 2025 IPO with $500 million in revenue run rate: SymphonyAI reached $500 million in revenue run rate last year and achieved profitability, after growing revenue at a rate of about 25%.??

• SymphonyAI’s Innovative Enterprise AI SaaS Products Recognized by Leading Industry Analysts and Influencers Worldwide: This year, SymphonyAI introduced new AI SaaS products, including SensaAI for Sanctions Screening, IRIS Foundry and IRIS Workspace, Apex Enterprise IT Copilot, and Media AI Copilot.?