Commentary and a selection of the most important recent news, articles, and papers about AI.
Today’s Brief Commentary
Large Language Models (LLMs) don’t continuously learn, and David Linthicum covers this ground well in his YouTube video in the Generative AI and Models section. I strongly recommend that you watch it. I think today’s LLM + RAG (Retrieval Augmented Generation) combination is a bit of a hack, or at least the current technology’s state before we get to the next generation of generative AI. Moving RAG to an agentic model might be on the pathway to doing that, and I provide a link that talks about that.
When I play online MMORPGs such as Star Wars: The Old Republic, I like the quests, but I especially enjoy exploring the worlds. Google’s Unbounded extends this beyond current games with their open environments to include more personalized narratives. I’m not sure how this will work with multiple players, and at that point, I think we’ll have to stop calling them games. They’ll be immersive online environments where players’ worlds intersect and interact. Of course, this isn’t a new concept because Second Life and its descendants have been around for almost twenty years. See the link in the Games section below to learn more about Unbounded.
If you are interested in novels about virtual worlds, I can recommend three, all by Neil Stephenson:
- Snow Crash (1992)
- Reamde (2011)
- Fall; or, Dodge in Hell (2019)
Finally, I don’t usually thank authors for their research papers, but I must do so for the NVIDIA employees and colleagues who published the survey paper “Artificial Intelligence for Quantum Computing” (link in the Quantum and AI section). Read this and get smarter.
General News, Articles, and Analyses
Startup Looks to Develop Physical Artificial Intelligence, Raises $400M
Author: Liz Hughes
(Thursday, November 7, 2024) “San Francisco-based AI-powered robotics startup Physical Intelligence announced this week it has raised $400 million in funding bringing its valuation to $2.8 billion.”
Waymo Launches AI Model for Autonomous Driving
Author: Liz Hughes
(Monday, November 11, 2024) “The End-to-End Multimodal Model for Autonomous Driving (EMMA) was specifically trained and fine-tuned for autonomous driving, leveraging Gemini’s world knowledge to better understand complex road scenarios.”
Agentic AI
How agentic RAG can be a game-changer for data processing and retrieval | VentureBeat
https://venturebeat.com/ai/how-agentic-rag-can-be-a-game-changer-for-data-processing-and-retrieval/
Author: Shubham Sharma
Commentary:
I think the LLM + RAG combination is a short-term solution, almost like putting a bandaid on a model. We’ll need better unified schemes in the future.(Tuesday, November 12, 2024) “Organizations have already started upgrading from vanilla RAG pipelines to agentic RAG, thanks to the wide availability of large language models with function calling capabilities and new agentic frameworks.”
AI Chipsets and Infrastructure
Google puts Nvidia on high alert as it showcases Trillium, its rival AI chip, while promising to bring H200 Tensor Core GPUs within days | TechRadar
Author: Wayne Williams
(Friday, November 8, 2024) “Trillium offers substantial advancements over TPU v5e predecessor”
Amazon Invests $110M in AI Research of Trainium Chips
https://aibusiness.com/generative-ai/amazon-invests-110m-in-ai-research-of-trainium-chips
Author: Liz Hughes
(Tuesday, November 12, 2024) “The investment would support generative AI research at universities using Trainium chips. The program, dubbed Build on Trainium, would provide researchers the ability to develop new AI architectures, machine learning libraries and performance enhancements for large-scale distributed AWS Trainium UltraClusters, groups of AI accelerators working in unison on complex computational tasks, Amazon said.”
Games
Exploring Google’s Unbounded AI: The Future of Gaming Worlds – Geeky Gadgets
https://www.geeky-gadgets.com/google-unbounded-ai-gaming-worlds/
Author: Julian Horsey
(Wednesday, November 13, 2024) “Imagine stepping into a game where the world around you isn’t just a static backdrop but a living, breathing entity that evolves with every decision you make. Well, Google is turning that dream into reality with their latest innovation, “Unbounded.” This isn’t just another game; it’s a new leap into the future of gaming, where you, the player, are not just a participant but a creator of your own narrative. By harnessing the power of generative AI, Google is setting the stage for a gaming experience that’s as limitless as your imagination.”
Generative AI and Models
Scholarly publishing world slow to embrace generative AI
Author: Kathryn Palmer
(Thursday, November 14, 2024) “As the technology’s reach into the information sector expands, a recent report from Ithaka S+R shows that academe is still grappling with how best to integrate it into the scholarly publishing process.”
Ep. 24 Why GenAI Models Don’t Continuously Learn and How It Can Be Fixed | AI Insights & Innovation
https://youtu.be/xlLiki97_g0?feature=shared
Author: David Linthicum
(Saturday, November 16, 2024) “This video delves into why large language models (LLMs) lack continuous learning and ongoing efforts to address this. Challenges like catastrophic forgetting, resource constraints, and data privacy hinder progress. Catastrophic forgetting occurs when models lose prior knowledge with new training. Efforts are underway to develop continual learning algorithms, meta-learning for adaptability, and techniques for handling non-stationary environments. Researchers focus on allowing models to learn from their environment continuously, similar to reinforcement learning. While significant progress is being made, achieving fully continuous learning in generative AI requires further advancements in algorithms and privacy solutions.”
ChatGPT on your desktop | OpenAI
https://openai.com/chatgpt/desktop/
“ChatGPT on your desktop. Chat about email, screenshots, files, and anything on your screen.”
Quantum and AI
[2411.09131] Artificial Intelligence for Quantum Computing
https://arxiv.org/abs/2411.09131
Authors: Alexeev, Yuri; Farag, Marwa H.; Patti, Taylor L.; Wolf, Mark E.; Ares, Natalia; Aspuru-Guzik, Alán; Benjamin, Simon C.; Cai, Zhenyu; Chandani, Zohim; ; …; and Costa, Timothy
(Thursday, November 14, 2024) “Artificial intelligence (AI) advancements over the past few years have had an unprecedented and revolutionary impact across everyday application areas. Its significance also extends to technical challenges within science and engineering, including the nascent field of quantum computing (QC). The counterintuitive nature and high-dimensional mathematics of QC make it a prime candidate for AI’s data-driven learning capabilities, and in fact, many of QC’s biggest scaling challenges may ultimately rest on developments in AI. However, bringing leading techniques from AI to QC requires drawing on disparate expertise from arguably two of the most advanced and esoteric areas of computer science. Here we aim to encourage this cross-pollination by reviewing how state-of-the-art AI techniques are already advancing challenges across the hardware and software stack needed to develop useful QC – from device design to applications. We then close by examining its future opportunities and obstacles in this space.”
