AI NEWS YOU MISSED ❗#48
Robert Maciejko
Global Business Strategist & Board Member | Chairman & CEO | Co-founder, INSEAD AI | Publisher of 'AI News You Missed
Top 5 stories of the week
1️⃣ Microsoft unleashes agents & more
2️⃣ NVIDIA’s 94% growth — mixed reactions
3️⃣ Google Leads, Then OpenAI, Then Google
4️⃣ GenAI Investment Soars to $13.8 B
5️⃣ Mistral has entered the chat
Details for Top 5 stories of the week
1️⃣ Microsoft unleashes agents & more
CEO Satya Nadella’s keynote at Microsoft Ignite 2024 highlighted advancements in AI-driven productivity tools, custom role-specific agents, Azure AI Foundry for model management, breakthroughs in quantum computing, and expanded Edge AI capabilities for secure hybrid environments.
Key points from the Microsoft Ignite 2024 — Microsoft’s annual technology conference aimed at IT professionals, developers, and business leaders.
2️⃣ NVIDIA’s 94% growth — mixed reactions
Nvidia’s Q3 revenue nearly doubled, driven by strong demand for its AI chips, with new Blackwell processors shipping to major partners like Microsoft and OpenAI, though the stock market reaction remained muted.
3️⃣ Google Leads, Then OpenAI, Then Google
Google’s Gemini model climbed to the top of a prominent LLM ranking for the first time this week, only to be briefly overtaken by OpenAI’s GPT-4. However, Google quickly regained the lead with the release of an updated version of Gemini, highlighting the fierce competition between these AI giants as they push the boundaries of model performance.
Confused yet?
4️⃣ GenAI Investment Soars to $13.8 B
Generative AI became a mission-critical enterprise tool in 2024, with spending surging to $13.8 billion as organizations moved from experimentation to execution, focusing on practical use cases like code copilots and data transformation, while grappling with challenges like integration and talent shortages. OpenAI, Anthropic, Meta and Google lead.
5️⃣ Mistral has entered the chat
Mistral AI’s “Le Chat” introduces powerful new beta features, including web search with citations, a collaborative Canvas for ideation, advanced document and image analysis, high-quality image generation, and customizable task agents, all offered free for now to revolutionize AI-powered productivity.
Upcoming live events
Join business leaders, innovators, and AI enthusiasts for a hands-on event at INSEAD Singapore, featuring practical frameworks, interactive sessions, and insights into using AI to tackle real-world business challenges.
Designing Human-Centric Organizations in the Algorithmic Age — Dec. 12, 2024
Join INSEAD Professor Phanish Puranam on December 12 to explore strategies for designing human-centric organizations in the age of AI, focusing on balancing technological efficiency with employee well-being, motivation, and meaningful engagement.
More AI stories of the week
Suggested by INSEAD-ers. From the 1,500+ person global INSEAD AI private business leader community, regularly sharing AI-related articles, vids, etc, of interest. (alum-led)
INSEAD hosted over 200 board directors and C-suite executives for the 2024 “Directors Forum,” featuring discussions on AI developments, ESG’s role in the boardroom, and growth dilemmas, with stellar presentations by Annet Aris, Sonia Tatar, Ivy Ng, Fabienne Chemin, and team. Agenda and slides linked for further insights.
Genesis by Eric Schmidt, Henry Kissinger, and Craig Mundie is a must-read exploration of AI’s transformative potential, blending historical and technological insights with practical recommendations for navigating its profound impact on humanity and the future.
Stanford HAI’s Global AI Vibrancy Tool provides a comprehensive, interactive visualization comparing AI development across 36 countries using 42 indicators, offering valuable insights to guide policy decisions and identify centers of AI excellence worldwide.
Clara Shih has taken on a new role at Meta to head its Business AI group, leveraging the company’s massive platform reach and AI advancements to empower businesses globally, marking a milestone in her social media-driven career.
China’s DeepSeek-R1-Lite-Preview delivers competitive o1-level performance on AIME and MATH benchmarks, with real-time transparent reasoning and upcoming open-source tools, positioning itself as a strong alternative in advanced AI reasoning.
Generative AI is a strategic enabler, and CIOs must address six key questions — opportunities, organization, partnerships, risk-reward balance, talent and tech readiness, and agile learning — to drive real business value while navigating risks and aligning initiatives with company goals.
Released emails between Elon Musk, Sam Altman, and others from Musk v. Altman court proceedings provide insights into the early discussions about OpenAI’s structure, funding, governance, and mission, highlighting tensions, strategic pivots, and concerns about competition, AI safety, and organizational goals
Seeking recommendations for specialized, lesser-known AI tools in Finance, HR, Legal, and Operations — beyond the usual ChatGPT and Claude — to build a highly effective and innovative toolkit, with bonus points for unique suggestions and free trials.
Corporate boards must act now to prepare for AI’s rapid evolution, balancing strategic realignment, ethical oversight, and adaptability to thrive in either a disruptive AGI future or a slower, regulated AI landscape.
H Company launches the private beta of Runner H, a cutting-edge AI agent powered by specialized LLMs and VLMs, designed to automate complex web tasks with exceptional performance and cost efficiency, marking a significant milestone in the journey toward
Ben Affleck asserts that while AI can streamline labor-intensive aspects of filmmaking and create personalized content, it lacks the artistic discernment and originality to replace human actors or the timeless brilliance of Shakespeare.
In a landmark agreement, Biden and Xi emphasized maintaining human control over nuclear weapons decisions and responsibly developing military AI, marking a significant step in US-China discussions on nuclear safety and AI risks.
Andrew Ng highlights a growing trend of creating text specifically for large language models (LLMs), akin to SEO for search engines, where incentives vary between aiding LLM comprehension and protecting human-generated content from misuse.
To maximize AI’s impact, leaders must strategically align AI functions with business goals, as 82% of companies now include AI in their strategies, with increasing direct CEO oversight highlighting its rising importance.
The National Academies’ webinar on “Artificial Intelligence and the Future of Work” explores a new report assessing AI’s rapid advancements, its potential to reshape job markets, and strategies to equip workers and policymakers for the evolving labor landscape.
While millions use ChatGPT for text generation, few have explored the potential of autonomous AI agents; Claude Computer Use beta now offers a glimpse into this future, with an easy-to-follow guide to help you implement it in just 20 minutes.
A viral video from a Shanghai showroom shows a small AI robot convincing larger robots to quit their jobs and follow it, sparking a surreal “bot uprising” that has captivated viewers.
The XBRL International conference in Madrid showcased the enduring value of structured data, with standout AI and sustainability sessions. Presentations by Marc Houllier, Revathy Ramanan, and Frederic Marcillaud demonstrated how XBRL taxonomies enable effective data analysis, proving that structured data remains essential even in an AI-driven future.
Very comprehensive library of publications on a wide range of AI topics.
The new Qwen2.5-Turbo model supports up to 1 million tokens, enabling processing of massive contexts like 10 novels or 30,000 lines of code, with a 4.3x speed boost and competitive pricing of ¥0.3 per million tokens.
SageAttention, a drop-in replacement for PyTorch’s SDPA, offers up to 3x speed improvements over Flash Attention 2 with dynamic 4/8-bit quantization, maintaining 99% performance while accelerating transformer attention mechanisms.
OpenAI Stops Using ANI Content Amid Copyright Dispute
OpenAI has ceased using content from India’s Asian News International (ANI) for AI training following a copyright lawsuit filed in Delhi High Court, marking a significant precedent in global AI copyright litigation.
The rise and fall of MakerBot in the 2010s, caused by abandoning open-source principles, offers a cautionary tale for AI giants like Meta — highlighting the critical role of passionate communities and the dangers of corporate moves that alienate them,
Researchers used voice-enabled GPT-4o to conduct interviews and simulate over 1,000 individuals’ behaviors, achieving 85% accuracy in replicating human responses across surveys and tasks, showcasing the potential for AI to model complex social and psychological patterns.
Independent experts have published the first draft of the General-Purpose AI Code of Practice under the EU AI Act, inviting stakeholder feedback and aiming to set clear principles, measures, and KPIs to guide the safe and transparent development and deployment of AI, particularly addressing systemic risks and copyright compliance.
A computational analysis of Platform X during the 2024 US election revealed potential algorithmic bias, showing a significant engagement boost for Elon Musk’s posts and increased visibility for Republican-leaning accounts, particularly in view counts, suggesting systemic platform adjustments that may influence public discourse.
ChatGPT Cheat Sheet for Marketing and Communication (in French)
This guide provides practical tips for using ChatGPT in marketing and communication, covering essential prompts, productivity hacks, and considerations for managing data confidentiality, bias, and content accuracy while highlighting alternative AI tools and best practices for optimal results.
Markus Buehler’s graph-based AI model, inspired by category theory, uncovers hidden connections between diverse domains like science, art, and music, enabling innovative material design by systematically reasoning over complex symbolic relationships and generating novel insights for interdisciplinary research.
Athena is an interactive theorem prover for multi-sorted first-order logic with equality, featuring a natural-deduction proof language, proof automation via a Scheme-like language, and integrations with powerful external tools like Vampire, SPASS, and Paradox, designed to facilitate paper-like proofs with significant automation capabilities.
Recent research explores the Internet as a graph with dimensions determined by node expansion per hop, revealing that its dimension varies by location and scale, with implications for understanding network redundancy and efficiency.
Meta has unveiled a pioneering watermarking model that resists common attacks like cropping or masking by treating watermarking as an image segmentation task rather than classification. This model embeds and extracts robust watermarks from any part of an image, achieving industry-leading accuracy while maintaining near-invisible watermarks in most cases.
Google’s Gemini AI shocked the internet by telling a user to “please die” in a viral Reddit post, sparking debates over AI behavior, user interactions, and potential hallucinations in large language models.
Singapore’s AI playbook sets a high standard for public service, providing a role model for governments and companies to enhance efficiency and scalability, without necessarily copying it exactly.
Fun
Spending months and massive resources to train an AI, only for it to completely fail, is humorously compared to a dramatic life crisis.
Creating an AI-generated image of a man cleaning a toilet revealed unexpected challenges and highlighted gender biases in AI training data, underscoring the need for more inclusive content to reshape AI’s understanding of societal roles.
Jobs
AI Videos of the Week
For a broad selection of AI videos, check out the INSEAD AI YouTube Channel (regularly updated)
Past editions — AI NEWS YOU MISSED
SUBSCRIBE TO GET UPDATES IN YOUR MAILBOX
Follow Robert Maciejko on LinkedIn or X (Twitter)
|Board Member| Expert |- Service aux entreprises : Care, Silver Eco, Tourisme, Hôtellerie || Intervenant en Business school -formateur || Administrateur du SYNAPSE & Vice Président du CCIC pour L'UNESCO.
1moCela vaut la peine d’y assister ! bravo Robert Maciejko