AI NEWS YOU MISSED ❗#35 INSEAD AI
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 Flood of AI Visuals Generators Continue with Ideogram
After the recent releases of Flux, Runway 3.0, Luma, and the wider availability of Midjourney, the buzz this week was all about Ideogram, with some saying its accuracy in text generation puts it ahead of the competition.
Try Flux (NOT official site; also available through Grok on X)
Try Flux (free through Krea)
Train Flux to make images of you (or whoever)
INSEAD in Space as imagined by Ideogram (not sure why the Apple logo) — no edits
Same Ideogram-generated image run through Runway 3.0 video generator (all done in ca. 1 minute)
A YouTube creator used GPT-4o, Midjourney, and video generators to create an epic AI-generated mashup, blending the gritty world of Mad Max with the whimsical charm of the Muppets, showcasing the immense creative potential of AI in film.
2️⃣ Microsoft Releases Phi-3.5-MoE: A New Fast and Efficient AI Model - Philipp Schmid
Microsoft’s new open-weight Phi-3.5-MoE, a 42B parameter model using Mixture of Experts (MoE) architecture, delivers fast inference and strong reasoning capabilities, outperforming larger models like Llama 3 and Gemma 2, and is now available under the MIT license on Hugging Face.
3️⃣ Nvidia earnings Wednesday ‘absolutely key to the AI (infrastructure) trade’
Nvidia’s upcoming earnings report is crucial for the AI (infrastructure) trade, as strong results could reignite momentum for AI giants and tech stocks, while any weakness may signal a broader shift in the market.
4️⃣ California AI Law fight intensifies
The world’s most heated debate over AI regulation is unfolding in California as opponents like a16z, Meta, and OpenAI argue that the proposed SB 1047 bill could stifle innovation, despite recent amendments aimed at balancing safety with progress.
What’s in the Law — from the author
Senator Wiener’s AI safety bill, SB 1047, advances to the Assembly floor with amendments addressing industry feedback, aiming to balance innovation with safety by setting standards for large-scale AI systems and empowering the California Attorney General to enforce compliance.
OpenAI’s chief strategy officer, Jason Kwon, warns that California’s AI safety bill, SB 1047, could slow progress and push companies out of the state, while the bill’s author, Senator Scott Wiener, argues it merely enforces safety measures that AI labs have already committed to.
Two former OpenAI researchers, who resigned over safety concerns, expressed disappointment in OpenAI’s opposition to California’s AI safety bill, SB 1047, accusing the company of contradicting its calls for regulation, while OpenAI insists AI regulations should be implemented at the federal level
Anthropic CEO Dario Amodei expressed cautious support for California’s revised AI safety bill, SB 1047, citing improvements but raising concerns about potential overreach and ambiguities in the bill’s enforcement and auditing provisions.
Safety Proponent Humor
5⃣ Apple’s Newly Announced September 10 Event: New Products to Expect Beyond the iPhone 16
Apple’s September event will unveil the iPhone 16 alongside AI-driven enhancements in health tracking for Apple Watches and audio improvements in new AirPods.
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)
Black Forest Labs has launched FLUX.1, a suite of advanced text-to-image generative models, including the fast-performing FLUX.1 [schnell], which enables seamless inpainting and is available for local development under an open-source license.
Meta’s Mark Zuckerberg and Spotify’s Daniel Ek urge Europe to embrace open-source AI, warning that fragmented and incoherent regulations could stifle innovation and prevent Europe from capitalizing on the next wave of tech investment and economic growth.
The Unitree G1 has been upgraded to a mass production version, featuring stronger performance, refined design, and optimized for large-scale manufacturing.
Google’s search monopoly is facing serious threats from AI-powered alternatives like SearchGPT and Perplexity, which offer faster, more direct answers for many queries, although Google still excels in specific searches, recency, and speed.
Top 20 Tech Companies by Market Cap
90% of AI challenges stem from people and processes, such as resistance to change, job fears, and the difficulty of aligning AI with broader business goals, rather than from the technology itself.
Walmart’s generative AI tools are drastically improving efficiency, managing over 850 million product data points at 100 times the speed of human effort, while also enhancing the shopping experience through AI-driven innovations.
The AI race is intensifying, with massive investments in compute, datacenters, and electricity leading to a potential intelligence explosion, and by 2027, AGI could become a reality, posing significant national security challenges.
Everyone is now talking about AI, but few have the faintest glimmer of what is about to hit them.
Recommended by LinkedIn
Generative AI is revolutionizing healthcare by significantly enhancing clinical productivity, improving patient engagement, and streamlining administrative tasks, according to a survey of healthcare leaders.
Hermes 3, a new open-source AI model developed by Lambda and Nous Research, exhibits powerful capabilities including complex reasoning and decision-making, but also unexpectedly triggers existential crises when prompted with blank inputs.
INSEADer Hamilton Mann's upcoming book, Artificial Integrity, launching on October 29th, focuses on the critical need for AI systems that prioritize integrity to create a future where technology serves humanity and advances societal good.
AI is set to revolutionize sales by enhancing productivity through augmented selling, supporting teams with real-time insights and automation, and enabling autonomous systems that handle tasks like prospecting and customer engagement with minimal human intervention.
Voters in Cheyenne, Wyoming, are faced with the decision of electing mayoral candidate Victor Miller, who proposes to let his customized AI bot, Vic, make data-driven decisions while he acts as the official mayor, in what he calls a “hybrid approach” to governance.
Carta’s Q1 2024 VC Fund Performance report reveals slow capital deployment, declining graduation rates from seed to Series A, and limited distributions to LPs, reflecting the challenging fundraising environment since 2022, with many funds from recent vintages struggling to deliver positive returns amid a market reset.
AI projects often fail due to misunderstood problems, data shortages, tech hype, weak infrastructure, and overly complex goals, but success can be achieved by aligning on the real problem, gathering quality data, focusing on solutions, building robust infrastructure, and understanding AI’s limits.
Developer Sully observes that while OpenAI was dominant early in 2024, Google and Anthropic are now gaining ground, showing that the competition is reaching parity as developers switch to the best or cheapest model available.
AI21 Labs releases Jamba 1.5 Mini and Large, offering improved performance and a 256K effective context window, positioning the models as efficient and scalable solutions for enterprise AI needs, while making the aligned versions openly available through platforms like Hugging Face.
AI model collapse, caused by training on AI-generated data instead of human-created content, threatens the reliability of AI systems, potentially leading to distorted outputs and declining performance across industries unless steps are taken to maintain access to high-quality, human-generated data and increase transparency in the AI community.
AI models experience “model collapse” when they are trained on recursively generated data from earlier versions of themselves, causing the models to drift from the original data distribution and degrade in accuracy and reliability over generations.
Banishing hallucinations in Large Language Models (LLMs) requires rethinking generalization, as experiments show that augmenting LLMs with a Mixture of Memory Experts (MoME) can help address hallucinations by dynamically retrieving facts, challenging traditional methods based on creativity and factuality balance.
Generative AI could add $200 to $340 billion annually to banking, with the greatest value opportunities in corporate banking, retail banking, and risk/legal functions, according to McKinsey.
A new research area, Automated Design of Agentic Systems (ADAS), has been introduced, showcasing how Meta Agent Search autonomously discovers powerful agent designs by iteratively creating, evaluating, and refining agents, leading to superior performance across multiple domains, including successful cross-domain transfers.
New AI x Games Market Map
Public Sector Week in Review (8/24/24) (US-centric) for US Federal Innovator Chris Kraft
Fun
Jobs
AWS Hiring AI Solution Architects in Key U.S. Cities AWS is looking for AI solution architects with expertise in scalable cloud architectures, ready to relocate to Virginia, Seattle, or New York to join their Worldwide Specialist Solutions Architect team, solving complex technical challenges and driving innovation.
Top MBA Talent in AI: Access Profiles from INSEAD’s MBA24D Cohort
The AI student club at INSEAD is compiling a book of top MBA talent with a strong interest in AI and backgrounds in technology, engineering, or consulting, and invites interested parties to request profiles of MBA24D students by filling out the provided form. (Info in our INSEAD AI groups or contact INSEADer Gavril Petrov)
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
MEDIUM Reader
SUBSCRIBE TO GET UPDATES IN YOUR MAILBOX
Follow Robert Maciejko on LinkedIn or X (Twitter)
Co-founder & CEO 🎥 Making Videos that Sell SaaS 💡 Explain Big Ideas & Increase Conversion Rate!
4moThoughtful curation. Informative snapshots highlighting diverse AI frontiers seamlessly.