This week’s latest generative AI updates – August 13, 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. Notable highlights this week include: ?

SymphonyAI news?

• Blog: Find out how AI is supporting smarter payment management to help prevent fraud??

• Video: Learn how AI is improving store operations with an introduction to retail store intelligence?

• Whitepaper: Sanctions in focus – how new AI tech is adding business value?

• News: SymphonyAI named to Inc. 5000 annual list of the fastest-growing private companies in America?

• Webinar: Join the webinar with Constellation Research, Microsoft, MIT, and SymphonyAI on Thursday, Aug 22 at 3pm ET/12pm PT Cutting through the hype to unleash true enterprise AI value

Technology news?

The New A.I. Deal: Buy Everything but the Company?

• Google agreed to pay $3 billion to license?Character.AI's technology and hire its founders along with about 20% of its employees, in what appears to be an approach to avoid regulatory scrutiny.?

• Earlier, Amazon secured a $330+ million licensing arrangement with Adept, acquiring its AI technology and key talent while avoiding a full company purchase. The FTC is scrutinizing these AI deals for potential evasion of merger notification requirements.?

Big Tech Model Mania?

• Google launched Gemini 1.5 Pro, surpassing GPT-4o in performance, excelling in multilingual tasks, mathematics, and coding, signaling significant AI advancements and raising ethical discussions.?

• Google DeepMind released Gemma 2 2B, a lightweight AI model for developers, outperforming GPT-3.5 models, promoting transparency and safer AI outputs across various applications.?

• OpenAI introduced the GPT-4o Long Output model with 64,000-token capacity, in response to demand for detailed response capabilities for tasks like code editing and writing??

• Meta launched SAM-2, a model for rapid video segmentation that reduces computational resources and supports video analytics, content creation, and real-time processing.?

• NVIDIA introduced generative AI models and NIM microservices for OpenUSD for 3D content creation, supporting language, geometry, and physics, democratizing access to high-quality virtual production tools for multiple industries.?

AI Supercharge Security with Automation?

• Rezonate introduced Zoe, a generative AI assistant for identity security that automates access management tasks.??

• FireCompass launched Agent AI, the first AI for ethical hacking and autonomous penetration testing so organizations to proactively identify and address security vulnerabilities before they are exploited.?

New AI Development Tools??

• GitHub launched GitHub Models, providing developers access to generative AI models like Llama 3.1 and GPT-4o, simplifying AI experimentation and deployment, and transition to production via Azure.?

• Baidu unveiled a self-reasoning AI framework to reduce hallucinations in language models, for accuracy and contextual understanding, helping AI be more transparent and accountable.?

• Cohere introduced Prompt Tuner to optimize prompts for large language models, improving output quality through iterative refinement, for tasks like summarization, customer support, and code generation.?

• MIT researchers developed Thermometer, a calibration technique for LLMs, reducing overconfidence in incorrect answers and improving response reliability.?

• Hugging Face launched inference-as-a-service with Nvidia NIM microservices, improving token efficiency and rapid AI model deployment, using Nvidia DGX Cloud to streamline prototyping and optimize performance for developers.?

Gen AI models?

• Zyphra's Zamba2-2.7B small language model offers twice the speed and reduced memory usage compared to larger models, optimizing performance for on-device AI applications in resource-constrained environments.?

• aiOla's Whisper-Medusa speech recognition model is 50% faster than OpenAI’s Whisper, using “multi-head attention” for improved speech recognition speed and accuracy.?

• Runway’s Gen-3 Alpha Turbo video model generates videos 7x faster than its predecessor while maintaining high quality, for near-real-time video production.?

• Writer’s Palmyra-Med-70b and Palmyra-Fin-70b AI specialized LLM models outperform larger generalist models in healthcare and finance.?

• TWO’s SUTRA Dual2L multilingual LLM uses Llama 3.1 and supports over 50 languages via a dual-transformer architecture that separates concept learning from language learning.??

• Flux, an open-source AI image generator with 12 billion parameters, surpasses Midjourney in visual quality and prompt adherence, offering high-quality image generation.?

• MidJourney's V6.1 model improves image quality and speed, offering more detailed visuals with 25% faster generation.?

Gen AI-powered creation tools?

• ?Hostinger launched an AI tool that generates product descriptions from images, simplifying ecommerce management by automating title and description creation for consistency and productivity.?

• Wix launched AI tools for its website builder that generate SEO-optimized blog posts, complete with images, for content creation efficiency and organic traffic.?

• HitPaw launched Edimakor 3.0.0 for Mac, incorporating AI-based subtitle translation and copywriting tools for faster, localized content production.?

• Forum3 launched Spok, an AI marketing expert that uses LLMs and data intelligence to provide insights, optimize campaigns, and improve ROI for marketers.?

Gen AI market impact and projections?

• A recent Deloitte study found that 56% of companies use gen AI primarily for efficiency and productivity, with 35% focusing on cost reduction, but only 29% focused on innovation and growth. Companies rely on pre-built AI products, with 71% using productivity apps and 68% using standard AI tools. But 41% of companies are underprepared for AI talent and governance needs. Generative AI is expected to centralize power (52%), increase inequality (51%), and erode trust in institutions (49%), leading 78% of executives to call for stronger regulation.?

• F5's 2024 Digital Enterprise Maturity Index reveals that 29% of companies are accelerating digital transformation, with gen AI playing a critical role in automation and data management. 92% of digitally mature organizations have implemented zero trust security principles, reflecting a focus on robust security frameworks essential for AI integration.?

• An ISG study reveals that large enterprises plan to nearly double AI applications from 2023 to the end of 2024, reflecting increased AI spending from 2% to 3.7%. Challenges include regulatory impacts, internal AI governance, and managing unintended risks.??

• ISACA’s 2024 AI Pulse Poll shows that 23% of organizations are considering increasing AI-related jobs, while 39% expressed concerns that AI might negatively impact privacy, and 20% lack confidence in the ability to detect AI-powered misinformation?

• A report from the AI-Enabled ICT Workforce Consortium predicts AI will highly or moderately transform 92% of ICT jobs, with significant impacts on roles in business management, design, and quality assurance.?

• IDC forecasts that global AI platform software revenue will reach $153 billion by 2028, growing at a 40.6% CAGR. Cloud-based AI platform deployments are expected to outpace on-premises solutions, with a five-year CAGR of 50.9?

• The AI-in-security market is projected to grow to $13.45 billion by 2029, driven by increasing adoption of AI-powered products for cybersecurity and threat protection.??

• Tech companies are investing heavily in AI infrastructure but the cost of developing larger AI models is staggering, with estimates suggesting a single model like GPT-6 could cost $100 billion to train by 2027?

• Gartner research indicates 30% of generative AI projects will be abandoned by 2025, citing poor data quality, rising costs, and unclear business value as primary challenges facing AI adoption.?

• Gartner reports that generative AI for procurement has reached the peak of inflated expectations, with businesses eager to explore AI-driven automation but facing challenges in realizing tangible value.?

• IBM’s latest report reveals that 63% of executives cite model cost as the primary obstacle to adopting generative AI, with 58% highlighting the complexity of AI models as a significant concern. Enterprises currently use an average of 11 different AI models and anticipate a 50% increase in model diversity within three years, debunking the idea of a one-size-fits-all AI model.?

• A report by MIT Technology Review Insights reveals that while 67% of global enterprises have advanced cloud strategies, less than 10% have fully integrated AI.?

• A Cloudera survey reveals that many enterprises face challenges in building the necessary data infrastructure and acquiring needed skills to effectively implement AI solutions across their organizations.?