Commentary and a selection of the most important recent news, articles, and papers about AI.
Today’s Brief Commentary
I’m writing this at the tail end of my trip to Riyadh, Saudi Arabia, to speak at the huge (> 100,000 people) LEAP / DeepFest 2025 event. It’s been a busy week, and yesterday, I was on the panel “Quantum Leap: The Intersection of Quantum Computing and AI.” Areiel Wolanow moderated, with Kathrin Kind and Daniela Herrmann also speaking. Unfortunately, Mohammad AlKhayyal from Saudi Aramco was not able to join us.
I have an extended write-up of my comments, and I’ll note here the link when it is published.
As usual, Gen AI dominates the links. Of these, I want to highlight Cision’s guide to using AI in PR and the VentureBeat article about Qwen.
Regarding PR, whatever AI tells you, stop saying everything you do is a “breakthrough.” You only get so many of these that you can self-declare to the world. After that, let third parties label your innovations as such. It’s a bit like Aesop’s “The Boy Who Cried Wolf”: if you constantly say you have breakthroughs, no one will believe you when you actually do.
As to Qwen, most tech people who follow LLMs know about this from China, and it is well-regarded. Take time to get beyond the press and the Gen AI models from the usual suspects such as OpenAI. Perplexity is a front-end to many models, and they are constantly testing the various offerings for speed and quality. Also, look at what IBM is doing with Granite.
Certainly, pay attention to where your data is going when you use LLMs on the cloud and other security issues. This has made sense for years. However, code and model quality transcends political bombast, which is a great argument for open source.
Contents
- General News, Articles, and Analyses
- Semiconductor Chipsets and Infrastructure
- Generative AI and Models
- Research and Technical
- Related Articles and Papers
General News, Articles, and Analyses
The Complete Guide to Using AI in PR | Cision
https://www.cision.com/resources/guides-and-reports/complete-guide-generative-ai-pr-comms/
Excerpt: In the interest of increasing those numbers – and helping PR teams take advantage of all that AI has to offer – we created a comprehensive guide to help you better understand how to use AI in public relations (with a focus on generative AI), how to tackle its challenges, and how to make the most of its benefits.
Semiconductor Chipsets and Infrastructure
“Boring stuff changes the world”: AMD’s Jason Banta offers a glimpse into the future for AI and AMD | Laptop Mag
https://www.laptopmag.com/laptops/amd-interview-roadmap-2025-ai-and-gaming
Author: Madeline Ricchiuto
Date: Wednesday, February 5, 2025
Commentary: See the link in the Related section to our writeup of AMD‘s recent earnings announcement.
Excerpt: AMD is somewhat unique as far as PC chipmakers go. The company has robust CPU and GPU lines, while Nvidia and Intel are playing catchup on at least one processor front. AMD has also been implementing NPUs into its mobile CPUs for the past three generations. Qualcomm is a new player in the computing space and hasn’t jumped into either the gaming or discrete GPU sectors yet.
Generative AI and Models
Alibaba’s Qwen2.5-Max challenges U.S. tech giants, reshapes enterprise AI | VentureBeat
https://venturebeat.com/ai/alibabas-qwen2-5-max-challenges-u-s-tech-giants-reshapes-enterprise-ai/
Author: Michael Nuñez
Date: Tuesday, January 28, 2025
Commentary: I’m noting this about Qwen from a couple of weeks ago because DeepSeek isn’t the only interesting LLM and architecture coming out of China.
Excerpt: For CIOs and technical leaders, Qwen2.5-Max’s architecture represents a potential shift in enterprise AI deployment strategies. Its mixture-of-experts approach demonstrates that competitive AI performance can be achieved without massive GPU clusters, potentially reducing infrastructure costs by 40-60% compared to traditional large language model deployments.
Introducing deep research | OpenAI
https://openai.com/index/introducing-deep-research/
Date: Sunday, February 2, 2025
Excerpt: Deep research is built for people who do intensive knowledge work in areas like finance, science, policy, and engineering and need thorough, precise, and reliable research. It can be equally useful for discerning shoppers looking for hyper-personalized recommendations on purchases that typically require careful research, like cars, appliances, and furniture. Every output is fully documented, with clear citations and a summary of its thinking, making it easy to reference and verify the information. It is particularly effective at finding niche, non-intuitive information that would require browsing numerous websites. Deep research frees up valuable time by allowing you to offload and expedite complex, time-intensive web research with just one query.
Introducing the MIT Generative AI Impact Consortium | MIT News
https://news.mit.edu/2025/introducing-mit-generative-ai-impact-consortium-0203
Author: Liam McDonnell
Date: Monday, February 3, 2025
Excerpt: How can AI-human collaboration create outcomes that neither could achieve alone? What is the dynamic between AI systems and human behavior, and how do we maximize the benefits while steering clear of risks? How can interdisciplinary research guide the development of better, safer AI technologies that improve human life?
Report: Amazon to Introduce GenAI-Powered Alexa on Feb. 26 | PYMTS
https://www.pymnts.com/amazon/2025/report-amazon-to-introduce-genai-powered-alexa-on-feb-26/
Date: Saturday, February 15, 2025
Excerpt: The generative AI-powered version of the voice assistant will be able to respond to more than one request at a time, take actions on behalf of users without their direct involvement, and remember their preferences when recommending music or restaurants, per the report.
Research and Technical
Optimizing Large Language Model Training Using FP4 Quantization
https://arxiv.org/abs/2501.17116
Authors: Ruizhe Wang; Yeyun Gong; Xiao Liu; Guoshuai Zhao; Ziyue Yang; Baining Guo; Zhengjun Zha; and Peng Cheng
Date: Friday, February 28, 2025
Commentary: Research from Microsoft.
Excerpt: The growing computational demands of training large language models (LLMs) necessitate more efficient methods. Quantized training presents a promising solution by enabling low-bit arithmetic operations to reduce these costs. While FP8 precision has demonstrated feasibility, leveraging FP4 remains a challenge due to significant quantization errors and limited representational capacity. This work introduces the first FP4 training framework for LLMs, addressing these challenges with two key innovations: a differentiable quantization estimator for precise weight updates and an outlier clamping and compensation strategy to prevent activation collapse. To ensure stability, the framework integrates a mixed-precision training scheme and vector-wise quantization. Experimental results demonstrate that our FP4 framework achieves accuracy comparable to BF16 and FP8, with minimal degradation, scaling effectively to 13B-parameter LLMs trained on up to 100B tokens. With the emergence of next-generation hardware supporting FP4, our framework sets a foundation for efficient ultra-low precision training.
Related Articles and Papers
Sutor Group Earnings Brief: AMD Announces Financial Results for Q4 2024 and FY 2024
Date: Thursday, February 6, 2025
Excerpt: Some highlights from AMD‘s earnings with some quantum and AI perspectives.
Sutor Group Intelligence and Advisory
Dr. Bob Sutor is the CEO and Founder of Sutor Group Intelligence and Advisory. Sutor Group provides broad market insights and deep technical expertise based on over four decades of experience with startups and large corporations. It advises Deep Tech startups, companies, and investors on quantum technologies, AI, enterprise software, and other emerging tech fields.
Sutor Group shares its knowledge and analysis through direct client engagements and seminars, reports, newsletters, books, written and on-air media appearances, and speaking and panel moderation at the top conferences and client events.
Disclosures
Bob Sutor is a former employee of IBM and Infleqtion and holds equity positions or stock options in each company. He is a Non-Executive Director for Nu Quantum and Advisor to the venture capital firm Forma Prime.
The information contained in this newsletter or article and the resources available on the Sutor Group Intelligence and Advisory website is not intended as, and shall not be understood or construed as, legal or financial advice. We are not attorneys, accountants, or financial advisors, nor are we holding ourselves out to be. The information is not a substitute for legal or financial advice from a professional who is aware of the facts and circumstances of your individual situation.
