Sajid Sadi

The future of AI isn't a sudden leap, but a steady climb. INSEAD experts predict a fusion of rule-based and neural systems, promising enhanced accuracy and broader knowledge. Yet, the holy grail of human-like AI remains elusive. What's your take on AI's evolutionary path? #ArtificialIntelligence #FutureOfTech #AIEthics #InnovationTrends #TechLeadership

3

Sharing my latest article on Overconfidence in State of the Art LLMs. Explore the challenges and limitations these models face in critical applications. #AI #MachineLearning #LLMs #ArtificialIntelligence #TechInsights #Innovation #EnterpriseSolutions #AIResearch

23

3 Comments

The new issue of Interactions is out, with a number of great articles. https://lnkd.in/gCmWQ973. This issue has a number of articles focused on the role of UX and HCI in thinking about AI. #humancomputerinteraction ABOUT ACM INTERACTIONS ACM Interactions magazine is a mirror on the human-computer interaction (HCI), interaction design (IxD), and user experience (UX) communities and beyond. It offers a multiplicity of perspectives in the form of conversations, collaborations, relationships, and new discoveries focusing on how and why we interact with the designed world of technologies. Our focus is broad: from interactions with devices to the impact of platforms and infrastructures on our experiences of digital technologies. Interactions has a special voice that lies between practice and research, with an emphasis on engaging HCI, IxD, and UX researchers and practitioners in dialogue about fundamental research and practical application. The magazine is published bimonthly by the Association for Computing Machinery (ACM), the largest educational and scientific computing society in the world. Interactions is the flagship magazine for the ACM's Special Interest Group on Computer-Human Interaction (SIGCHI), with a global circulation that includes all SIGCHI members

173

1 Comment

Large language models (LLMs) are increasingly being explored for their potential in medicine, but applying them effectively to complex clinical scenarios remains a challenge. The paper "MDAgents: an adaptive collaboration of LLMs for medical decision-making" by Yubin Kim, Chanwoo Park, Hyewon Jeong, Yik Siu Chan, Xuhai Xu, Daniel McDuff, Hyeonhoon Lee, Marzyeh Ghassemi, Cynthia Breazeal, and Hae Won Park (NeurIPS 2024) introduces a novel framework aimed at addressing this gap. MDAgents structures collaboration among LLMs dynamically, assigning roles—either independent or group-based—based on the complexity of the medical task. This design mimics real-world clinical decision-making processes, where collaboration is tailored to the case at hand. Evaluated on 10 medical benchmarks, MDAgents achieved leading performance on 7 tasks and demonstrated an 11.8% accuracy boost when incorporating external medical knowledge and moderator reviews. A case study further illustrates how the framework synthesizes differing perspectives among LLMs to reach accurate, consensus-driven diagnoses. This framework stands out for its adaptability and its emphasis on collaboration. Ablation studies underscore the importance of the system’s components, from its ability to classify medical complexity to the integration of multi-modal reasoning. By reflecting the dynamic, consultative nature of clinical decision-making, MDAgents provides a thoughtful approach to enhancing LLM-assisted medical diagnosis. To me, this paper is interesting because it takes a practical step toward aligning AI capabilities with the intricacies of healthcare. By emulating real-world medical collaboration, MDAgents not only improves performance but also lays the groundwork for more robust and trustworthy AI systems in clinical settings. What are your thoughts on how AI frameworks can better reflect real-world decision-making processes in fields like healthcare? paper: https://lnkd.in/gd5fX4Ef #artificialintelligence #AIhealthcare #LLMs #AIAgents

11

Unpacking Vinod Khosla's Future AI predictions at the Imagination in Action summit last week at MIT Media Lab. ✨ Evolution of Bipedal Robots Robots are evolving from static, rule-based imitation systems to dynamic, continuously learning systems. The ability to self-learn and improve from their experiences and surroundings will become integral to robotic systems. This development will empower robots to adjust to new circumstances and respond effectively to environmental changes. ✨ Paradigm Shift in Software Throughout computing history, we have been optimizing how we interact with computers and machines. Traditionally, through programming languages, operating systems, and application software, we've created structured systems designed to perform specific tasks. However, the future of software, enhanced by AI, will shift towards adapting more fluidly to human needs. We are already seeing the beginnings of this transformation in the functionality of chatbots and concepts such as 'prompt inversion' or 'reverse prompting', where software interacts with users to better understand and achieve their desired outcomes. This shift will move us from the rigidity of traditional software to a more adaptable and user-centered software paradigm. ✨ Internet of the AI Age AI agents are set to become the primary users of the internet, necessitating a redesign of web services and infrastructure to accommodate AI-first interactions. Within the next five years, AI agents are expected to dominate internet usage, transforming how web services are designed and information is processed. ✨ Natural Language Programming Programming is becoming more accessible thanks to natural language processing, effectively democratizing programming skills. This will lower barriers to entry, making programming nearly cost-free and allowing more rapid realization of ideas—an excellent opportunity for innovators, founders and creators. ✨ Super Intelligent AI Systems We are beginning to see AI systems assisting in training other AI systems, with early steps involving AI-generated data. This is akin to the initial stages of AlphaGo, which didn't achieve superhuman capabilities until it began self-play. Future AI systems will use synthetic data and develop their own reward models, engaging in self-play and self-improvement. ✨ Universal Access to Expertise AI will soon provide widespread access to educational and healthcare expertise at little to no cost. Advances in AI will address issues like hallucination and ungrounded responses, ensuring an abundance of reliable, expert advice. ✨ Continuous Learning As technology continues to evolve rapidly, the ability to continuously learn and adapt will be crucial. Embracing a mindset focused on lifelong learning and a diverse skill set will be essential for navigating the future successfully. The emphasis is on learning to learn and diversifying one's learning approaches. John Werner, Paul Baier #FutureofAI #MITForge2024 #AITrends #IIA

33

1 Comment

It’s finally here! I’m thrilled to introduce you to Llama 3, our latest generation of models that is setting a new standard for state-of-the art performance and efficiency for openly available LLMs. Key highlights: • 8B and 70B parameter openly available pre-trained and fine-tuned models • Trained on more than 15T tokens, 7x+ larger than Llama 2's dataset! • State-of-the-art performance across industry benchmarks • New capabilities, including enhanced reasoning and coding • 3x more efficient training than Llama 2 • New trust and safety tools with Llama Guard 2, Code Shield, and CyberSec Eval 2. • Training underway on a leading-edge 400B+ parameter model  We’re also releasing a new version of Meta AI powered by Llama 3. We’re making it available in 13 new countries, integrating enhanced search capabilities and adding new ways to experience it on the web at meta.ai. Our latest updates to image generation on Meta AI is state of the art, allowing you to create, animate and share stunning images with a simple text prompt in seconds. We think Meta AI is the world’s most intelligent AI assistant and we can't wait for you to try it and tell us what you think. Read more about it here: https://lnkd.in/eB5yzi8R

2,385

80 Comments

Another demonstration of ScienceCast.org #AI enhancements of arXiv postings - free, simple interface https://lnkd.in/e_gXsF_9

3

1 Comment

🚀 Exciting Collaboration Alert! 🚀 NVIDIA and Hugging Face have joined forces to simplify the deployment of generative AI models. This partnership aims to streamline the integration of state-of-the-art foundational models into application development and production, enhancing performance and reducing operational costs for enterprises. 🔹 **NVIDIA NIM Inference Microservices**: - Designed for low-latency, high-throughput AI inference. - Supports a wide range of LLMs including Llama 3, Mixtral 8x22B, Phi-3, and Gemma. - Optimized for domain-specific applications in speech, image, video, healthcare, and more. - Achieves up to 5x faster token generation. 🔹 **Seamless Deployment on Hugging Face**: - Deploy models like Llama 3 8B and 70B with just a few clicks. - Leverage NVIDIA NIM on your preferred cloud service provider. - Near-100% utilization with multiple concurrent requests, generating text 3x faster. This collaboration makes it easier than ever for developers to go from AI model development to production, focusing on innovation and growth. #GenerativeAI #NVIDIA #HuggingFace #AI #TechInnovation #AIModels #DeepLearning #MachineLearning

3

Worth a read. I like this quote: “Regulation helps us avoid mistakes and enables new benefits as we move into an uncertain future.”

3

This is a great introduction to the multi-agent system, including the human-in-the-loop concept in the OpenAI cookbook: https://lnkd.in/edqD5vED. I really like this notion of using intent before handoffs, including deciding what context to share or not.

56

Mindblown by the reasoning and dexterity capabilities of Figure's autonomous robots! 🤯 Figure 02 just announced, featuring advanced speech-to-speech reasoning and a Vision Language Model for quick decision-making based on visual input and error correction. Check out the video of Figure 01 below! #Robotics #Innovation #AI

2

Day 27- Interesting approach for tackling Agentic workflows, which are labor-intensive and limit the broader application of LLMs. AFLOW reframes workflow optimization as a search problem over code-represented workflows 🖥️, offering a more flexible and comprehensive approach than previous methods that relied on manual setup 🤖 or limited search spaces. ✨ Key Innovations: - Code Representation: AFLOW uses code to represent workflows. This allows precise control over what the language models (LLMs) do, includes conditional logic, and helps integrate complex structures like graphs. 🖥️ - Operators: AFLOW introduces "operators." These are groups of reusable nodes for common tasks, like Ensemble or Review & Revise. Operators make workflow creation simpler and improve search efficiency. 🔄 - Monte Carlo Tree Search (MCTS): AFLOW uses a modified MCTS algorithm to explore different workflow setups. This process evaluates workflows, learns from the results, and improves them. The tree structure helps keep track of past experiences for better choices. 🌳 - Soft Mixed-Probability Selection: This method balances exploration and exploitation. It uses a mix of equal and score-based probability to ensure a wide range of options while focusing on successful workflows. ⚖️ - LLM-Driven Expansion: An LLM helps create new workflows by changing existing ones and their prompts. It uses insights from past changes and performance data to guide its improvements. 📈 Results and implications: AFLOW was tested on six benchmark datasets across question answering, code generation, and mathematics. The results show that AFLOW creates workflows that exceed those of traditional methods and other automated systems, allowing smaller language models (LLMs) to perform as well as or better than larger ones, leading to cost savings 💰. By reducing the need for human expertise, AFLOW simplifies workflow design ✨, making LLM technology more accessible and scalable. Its ability to find cost-effective workflows could promote wider use of LLMs in real-life applications 🌍. Notable caveats: AFLOW mainly focuses on reasoning tasks with clear evaluation methods. 🧠 Expanding its capabilities to handle more complex tasks, especially those that involve personal judgments or changing environments, could be valuable for future research. 🔍 Looking into how to include reinforcement learning could also improve workflow optimization and create more flexible AI systems. 💡 paper link- https://lnkd.in/gbbCXZbC #GenAI #0to100xengineer

6

2 Comments

The Human Side of AI Innovation: Stanford's Vision Through Andrew Ng and Fei-Fei Li's Eyes In an extraordinary dialogue that bridged technological brilliance with human warmth, Stanford's AI pioneers Andrew Ng and Fei-Fei Li shared their compelling vision for artificial intelligence. After watching this on YouTube, I felt truly inspired by their ideas and vision for AI’s future. Let me take you through this remarkable conversation that’s reshaping how we think about AI's purpose and its place in our world. Reimagining AI's Purpose The HAI Vision At Stanford's Human-Centered AI Institute (HAI), something revolutionary is happening. "We're still pre-Newton in our understanding," Fei-Fei candidly shared, reminding us that despite AI's rapid advancement, we're just beginning to grasp its true potential. But here's what makes their approach special: they're pioneering responsible innovation while keeping human values at the core of every development. From Vision to Reality AI in Action Picture this: An AI system that works like a guardian angel in hospital rooms, using privacy-preserving depth sensors instead of traditional cameras to prevent medical errors. This brilliant fusion of innovation and ethical consideration shows how AI can enhance healthcare while protecting patient dignity. These real-world applications demonstrate HAI's commitment to developing solutions that work for people, not despite them. Leading with Heart: Advice for Future Innovators Think long-term – pursue lasting impact, not just quick wins. Blend tech with humanity – balance technical rigor with ethical insight. Stay curious – ask big questions about intelligence and its purpose. Champion inclusivity – encourage diverse perspectives for well-rounded AI. Join the Conversation Standing at this tech-human crossroads, how will you shape AI’s future? Whether you’re a tech enthusiast, policymaker, or simply someone curious about AI, your voice is essential in this ongoing dialogue. Link: https://lnkd.in/ddNebSpe #HumanCenteredAI #AIInnovation #TechForGood #FutureOfWork #StanfordAI #ResponsibleAI #AILeadership #Innovation

16

2 Comments

Good news for #AI #innovation, and especially for developers, builders, and the technology community at large. Thank You Gavin Newsom and the Office of the Governor - California for a measured and thoughtful veto of the #SB1047 amendment. While no one disputes the need for #regulation in #AI, #California in particular has both an #opportunity and #responsibility to assume leadership. A deeper understanding of the technical ways in which AI #data is #retrieved, #processed, and #presented is key to an #informed regulation. Unlike traditional #relational #databases where the data you retrieve is the same as the data you store with #tables and #indexed data lookups, AI and #GenerativeAI is fundamentally different. With #vector #databases and #neuralnetworks, #LLMs can create data on the fly. This calls for qualified input from #academic #research, #developers, and #technology #providers to inform policy on #guardrails and #practical #industry implementations. Join the vetted community at Llama Labs, or follow me for more on this, and AI technology full-stack conversations, deepdives, resources, and events

15

In the wake of genAI hype, developers of all stripes are striving to design and implement application architectures that make use of the advantage of #LLMs and appropriately manage their disadvantages. 2024 is year 1 for optimizing genAI applications, and today you've probably heard of Retrieval Augmented Generation (RAG) as one of the common approaches. But what is #RAG in a business context? How can it be applied to enterprise data and applications? And are there other approaches to customizing the behavior of foundation models for language like #LLMs? To get more information on the first question, check out Rowan Curran’s recently published pair of Forrester blogs on RAG and what comes next in #genAI-infused application: #RAG is all the Rage (Part 1): https://lnkd.in/eTHwdJCU The Machine is Getting More Complex (Part 2): https://lnkd.in/eWWVzDve If you'd like to learn more about the latter - adapting LLMs to your enterprise - come to my session in September at #Forrester's Technology and Innovation forum: https://lnkd.in/eWnSMNWX

6

The #AIAct needs proper #implementation to be successful for #Europe. This new #legal study offers excellent practical#recommendations on #GPAI. This shows how we can operationalise #AIact rules successfully together and we hope it helps the AI Office! #CCIAeurope #ArtificialIntelligence #GeneralPurposeAI #digitalEU #BetterRegulation #AIOffice #AIPact CCIA Europe Boniface de Champris Yann Padova Sebastian Thess #EU #EUCommission #AI Kilian Gross Lucilla Sioli EU Digital & Tech Roberto Viola

5

This is a very interesting topic and debate related to #AI, #art, #positioning and many more themes. I saved it for my watch and food for thought list. Thought it will be useful for many people in my network, thus resharing. Thanks Jeremy! Insightful and thought provocative post like always. Keep it up! 😉👏 #debate #AI #Art #Humans #RebrandAI #DimitriosIoT

8

1 Comment