Peter Norvig

Lin Guan and Yifan Zhou have led a very interesting study investigating the effectiveness of Vision Language Models in critiquing robot behaviors in terms of the desirability of their "style." If you are wondering how our work on LLM's inability to verify correctness squares with this study on using them as style critics, see my earlier post https://lnkd.in/gBYAr5qq ! This is the reason why being a style critic is one of the roles LLMs can play in LLM-Modulo framework: https://lnkd.in/gyAyKx4E Check out the details of the work @ https://lnkd.in/gfVZ4EYW ( Lin Guan also has a nice twitter thread on the work here: https://lnkd.in/gnieU-qM )

89

2 Comments

Small models becoming smarter! Phi-3 reinforces the importance of data quality.

24

Thanks Hamdi Amroun, PhD - this is a great paper to share. In the real world business context we often need AI to reason over large pieces of text, and often you need/want multi-document summarization (MDS). 'When testing five LLMs with benchmarks using news and conversation datasets, they found up to 75% of content in MDS summaries was hallucinated, with notable increases towards summaries' end. Alarmingly, even for non-existent topics, models like GPT-3.5-turbo and GPT-4 generated fabricated content 79% and 44% of the time, respectively. Analysis of 700+ generated insights showed that hallucinations often arose from failures to follow instructions or overly generic content.' I agree with Hamdi's sentiments that despite improvements in post-processing techniques, robust solutions are urgently needed to reduce hallucinations, as inaccurate summaries can lead to business risks (notably misinformation and misrepresentation). I would add that there is still a lot of utility in LLMs/Multimodal models but we need to be vigilant - this is not about AI malevolence or intentional deception. LLMs do not possess consciousness or intent; they generate content based on patterns in data they were trained on.

6

1 Comment

Great to see more press coverage of the possibilities for AI + robotics to create new ways to help people and do things that have not already been imagined. I think the future of robotics will be different than what we see today, stranger and more amazing. 

5

1 Comment

This is a monumental breakthrough—a fully error-corrected quantum computing chip. Bravo, Google! What an incredible achievement. Sundar Pichai, big high-fives to you and the team—this is truly WOW-worthy! Now, consider the implications. Daisy-chain several of these chips into a cluster, reaching the magical 1,000-qubit threshold or beyond, and it’s game over for many of today’s math-based encryption methods. Large prime factorization, pseudo-random number generation, and other hard math problems could crumble under this tech. And imagine the potential when this processing horsepower is paired with a large quantitative model (LQM) powered by advanced AI. The possibilities are staggering—impossible to fully quantify. Oh the fun I would have in the labs at Google with this tech! :) The AI genie is out of the bottle, and breakthroughs like this have ramifications that will alter the course of mankind in both good and bad ways. https://lnkd.in/gVYAvNTy

10

4 Comments

Sharing my experiment on creating a Streamlit based chatbot, that combines inputs from Agentic RAG using llamaindex (with openai and MistralAI), and langchain google genai (with Gemini-1.5-Flash), to extract information from images embedded within the PDF files. Was a nice experience to build this chatbot. Still many open questions, ways to improve the solution, in a production scenario..but nice to combine features from multiple libraries and come up with a solution. Please check it out! Thanks. #chatbot #gemini #openai #langchain #llamaindex https://lnkd.in/gPdtxhQF

19

From where I sit in the SF Bay Area, cradle of innovation for at least the last 30 years, I'm a true believer in the promise of the #AI revolution. That said, I find it fascinating to consider views on the opposing side. I'm sharing below two opinion pieces from the last two weeks suggesting that the AI revolution is fundamentally overhyped and a bubble ripe for bursting, and that AI tools are actually not useful or transformative. Contrast this with venture capitalist Kai-Fu Lee's prognostication that 50% of white collar jobs will be displaced by #AI within three years, and that OpenAI and the other cutting edge movers in the space are headed to a trillion+ dollar valuation. Which side will win out? Where will we be by 2030 in the AI race? I'd love to hear your thoughts in the comments.

7

3 Comments

In the race to advance AI, we often focus on algorithms and data. But what if the most fundamental bottleneck lies in the design of GPUs? I recently spoke with Sean Lie, co-founder and CTO of Cerebras Systems, about their audacious approach to AI computing. Instead of iterating on existing chip designs, they're reimagining the entire system from the ground up. This end-to-end approach is pushing Cerebras to innovate in unexpected areas. Take server design, a field long thought to be settled. When you're working with a chip the size of a dinner plate—57x larger than NVIDIA’s H100—even servers need an overhaul. It's a reminder that breakthrough innovation in AI will not come from algorithms alone, but from rethinking every layer of the stack, from bare metal to high-level software. Our full conversation here: https://lnkd.in/gWZCi4zE

37

Making decisions? In today's world, they must be data-driven, right? But with such an overwhelming amount of data, it’s natural to turn to AI. But can AI also handle ethical considerations? Is there an "AI Conscience"? Fortunately, Brian Green is joining us at the IIT Bay Area Leadership Conference on 10/19/2024 to address this important topic: "Ethical Decision Making in the Age of Data and AI." Brian is the Director of Technology Ethics at the Markkula Center for Applied Ethics at Santa Clara University (and if you are reading this on your iPhone, yes - the center is indeed named after Armas Clifford "Mike" Markkula Jr., the third co-founder, original angel investor, first chairman, and second CEO for Apple who was introduced to the two Steves by two Gods of Silicon Valley: Don Valentine and Regis McKenna.) Brian teaches AI ethics and has authored/co-authored several books, including Space Ethics (2021) and Ethics in the Age of Disruptive Technologies: An Operational Roadmap (The ITEC Handbook) (2023). I hope I have given you enough data to make a decision to attend the conference!

2

1 Comment

Large Language Models (LLMs) as super lossy compression. I have been thinking a lot about demystifying what LLMs do because lots of y'all's are very mystified. In the linked article, I try to reframe them for you, hoping it will help you understand what they good for and what they are not good for. Side note: I am still looking for my next adventure. If communicating complicated concepts so that regular people can grasp them is something you need, you should message me. I write code and stuff too and am usually fun to be around. #WrittenByMe

6

3 Comments

One of the biggest challenges in language models today is making them more interpretable. We often treat AI models as black boxes: data goes in, a response comes out, and the reasoning behind that response remains unclear. I remember an interview with Google CEO, where he was asked to explain how Gemini works. He said he didn’t know. This response resonated with the scientific community, as deep learning often similar to the human brain, but the interviewer was shocked! How can a model, released to millions, be so poorly understood? Two weeks ago, Anthropic released an important paper on model interpretability. They used a technique called "dictionary learning," borrowed from classical ML, which isolates patterns of neuron activations that recur across many different contexts. This paper sheds some light on this important challenge, which, if solved, will create more trust in these models and thus ease the integration of AI into our everyday lives. Highly recommend reading: https://lnkd.in/gPzEePx8

11

1 Comment

Making AI accessible with Andrej Karpathy and Stephanie Zhan Candid take by Karpathy on difficulties in training LLMs, dominance of companies with access to infrastructure-to-train in LLM space, and possible options, especially for small players. https://lnkd.in/diDMa4Rd #llm

13

Essential #AI thoughts from Fei-Fei Li. I was honored to hear her speak about AI years ago at orientation for new fellows at The Paul & Daisy Soros Fellowships for New Americans and I never forgot her wisdom and prescience. She heads up HAI (Human-Centered AI at Stanford University. Q: What do you you think about AI’s future, what worries you the most, and what gives you hope? A: It’s not AI’s future, it’s humanity’s future. We don’t talk about electricity’s future, we don’t talk about steam’s future. At the end of the day, it is our future, our species’ future, and our civilization’s future—in the context of AI. #futureofeducation #futureofwork

28

I'm thrilled to see Anthropic's latest announcement of the upgraded Claude 3.5 Sonnet and new Claude 3.5 Haiku models! 🚀 As someone who strategically leverages various AI platforms, including Claude, for specific use cases, these advancements are particularly exciting. The enhanced coding capabilities in 3.5 Sonnet are especially promising 💻✨, as they build on Claude's already impressive performance in this area. What really stands out to me is the groundbreaking computer use capability now in beta. 🤖🖱️ While I currently use AI selectively and thoughtfully for specific tasks, features like this - where AI can interact with computers like humans do - hint at the expanding possibilities ahead. I appreciate Anthropic's transparent approach in acknowledging that this feature is still experimental and may have limitations. This aligns with my philosophy of carefully choosing when and how to implement AI tools as they mature. 🎯 The fact that Claude 3.5 Haiku matches Claude 3 Opus's performance while maintaining cost efficiency is also noteworthy. 📊💪 This kind of advancement helps organizations like ours make informed decisions about scaling our AI usage as the technology evolves. Looking forward to exploring these new capabilities! 🌈 #AnthropicAI #Claude35 #AIInnovation #ComputerUse #FutureOfTech

2