Nishant Sinha

Echoing my comments to the attached post here: Conceptually, RDF-star is a superset of LPGs (you can model all LPGs in RDF-star but not vice versa). The LPG model makes a hidden abstraction assumption that the "real" edge is a label "hiding" a blank node statement reifier. Similarly, a hypergraph is equivalent to RDF-star. if you assume that a node is a pointer to a set or a linked list. RDF-star (its Turtle representation especially) is complex because it is arguably the most fundamental representation of a graph, and graphs can be extraordinarily complex. It is the machine language of the graph world: you can build out multiple abstractions (such as LPGs or hypergraphs) that hide some of the complexity, but at the end of the day, it will always be reducible to some form of rdf-star. I would argue that we probably need a better form of Turtle (which is an abstraction) and need to get away from the belief that RDF is equivalent to OWL. It's not. OWL is a way of encoding logical formalisms, just as SHACL is a way of describing structure, and SKOS is a way of describing (many forms) of inheritance taxonomy. They are all languages built on RDF, and I think as people begin to more fully incorporate reification into their modeling processes we will (possibly) see other abstraction languages built on top of RDF-Star as well (SHACL is evolving in that direction, I'm not sure that OWL is). https://lnkd.in/gTZ5Edwb

59

5 Comments

Will AI increase the rate of progress in science or not? A new paper from Sayash Kapoor and Arvind Narayanan (Could AI slow science?) says it probably won't AI is already used in science (See figure) “to uncover trends in data using sophisticated pattern-matching algorithms; hand-written machine learning models specified based on expert knowledge; or even generative AI to write the code that researchers previously wrote.” But “AI has already led to widespread errors. Even before generative AI, traditional machine learning led to errors in over 600 papers across 30 scientific fields. In many cases, the affected papers constituted the majority of the surveyed papers, raising the possibility that in many fields, the majority of AI-enabled research is flawed.” More importantly, AI modeling doesn’t lead to an improved understanding of scientific phenomena. “Solving problems and writing papers about them can be seen as a ritual that leads to the real prize, human understanding, without which there can be no scientific progress.” One Fields Medal-winning mathematician “emphasizes that the point of mathematics is not simply to figure out the truth value for mathematical facts, but rather the accompanying human understanding.” The same holds for AI. If AI solves “open research problems without leading to the accompanying understanding, AI could erode these useful byproducts by reducing incentives to build understanding.” Another reason that AI will likely not improve the conduct of science is that the AI-for-science suppliers want flashy “AI discovers X!” headlines “so that they can sustain the narrative that AI will solve humanity’s problems, which buys them favorable policy treatment.” He says “we should be skeptical of #AI-for-science news headlines. Many of them are greatly exaggerated. The results may fail to reproduce, or AI may be framed as the main character when it was in fact one tool among many.” If you are developing tools, “How about tools for finding errors in scientific code or other forms of quality control? Or “mathematicians have repeatedly said that tools for improving human understanding are much more exciting than trying to automate theorem-proving, which they view as missing the point.” The way we evaluate these tools should also change. For a lit review tool: 1) Does it save a researcher time and produce results of comparable quality to existing tools? 2) How does the use of the tool impact the researcher’s understanding of the literature compared to traditional search? 3) What will the collective impacts on the community be if the tool were widely adopted”? Currently, only the first question is considered thus guaranteeing that evaluations are “guaranteed to provide a highly incomplete and biased picture.” In conclusion, “The impact of AI on science as an institution, rather than individual scientists, is a different question that demands a different kind of analysis.” #technology #artificialintelligence

122

56 Comments

Quantum computing has a scaling problem. Classical networks have a physics problem.    Single quantum machines operate at hundreds of qubits, but real-world problems require millions. Millions of qubits with different modalities for different types of problems, and different generations to preserve investment. Meanwhile, classical networks are bound by the speed of light, limiting us on latency and security problems, costing us millions.    What if the solution to both was the same? Quantum networking.     Today, Cisco Quantum Labs is announcing software that makes both distributed quantum computing work and enables novel solutions to some hard classical problems as well. We're releasing three research prototypes:    — Quantum Compiler: The industry's first network-aware compiler that partitions quantum algorithms across multiple processors of any type. It uniquely accounts for network interconnects and includes another industry-first - distributed quantum error correction.    — Quantum Alert: Eavesdropper-proof security with guarantees from physics, not just promises from software. Using entanglement, it provides provable detection against "harvest-now-decrypt-later" attacks.    — Quantum Sync: Instantaneous decision coordination for classical systems. It uses entanglement to correlate actions across locations without waiting for messages to travel, eliminating costly delays.    All three are built on a unified set of quantum networking protocols and software such as entanglement distribution and entanglement swapping.    All of this was made possible by the incredible work of Ramana Kompella, Reza Nejabati, Sabitha Krishnamurthy and their teams.    Read more about it here: http://cs.co/9008AbIBg     Want to see it in action? We'll have live demos and make the compiler prototype available for download at our Quantum Summit on September 30 and October 1. Register here: http://cs.co/9006AbxAC  

525

14 Comments

Another excellent lecture from Aishwarya Naresh Reganti and Kiriti Badam's course delivered by Aman Khan who was a fantastic speaker. What Nobody Tells You About Building AI Agents: 10 Critical Evaluation Truths: 1️⃣ Systematic Evaluation is Non-Negotiable The companies building reliable AI systems aren't magically more talented - they're simply more rigorous about testing. Even non-technical team members can contribute significantly to crafting evaluation prompts. 2️⃣ Understand the Three Pillars of Agent Architecture Every modern agent has three core components that need distinct evaluation approaches: input processing, reasoning layers, and API action execution. Each has unique failure modes that require targeted testing strategies. 3️⃣ Identify All Five Distinct Failure Modes Most teams only check for 1-2 problems. Comprehensive evaluation requires testing: function calling accuracy, output correctness, path efficiency, API parameter selection, and human-aligned responses. Miss one, and your "working" agent will fail in production. 4️⃣ Path Optimization is the Hidden Performance Killer Agents can produce correct answers but take 3-5x more steps than necessary. This destroys user experience and drives up costs. Tracking step counts and preventing recursive loops is essential. 5️⃣ LLM-as-Judge Enables Scale Using LLMs to evaluate other LLMs is transformative. It allows assessment of thousands of interactions cost-effectively. While not perfect, it provides valuable signals for where to focus human evaluation efforts. 6️⃣ Few-Shot Prompting is Worth the Effort In real implementations, adding specific examples to evaluation prompts improved accuracy from 80% to 90%. Domain-specific examples significantly enhance judgment quality with minimal implementation complexity. 7️⃣ Meta Prompting Represents the Next Level Using GPT-4 to optimize evaluation prompts can match or exceed few-shot prompting effectiveness. This reduces the need for prompt engineering expertise by automating optimization. 8️⃣ Open Telemetry Provides Visibility Into the Black Box Implementing spans and traces visualizes complete request workflows from input to output. This integration with existing monitoring systems enables detailed analysis of performance and failure points. 9️⃣ Guardrails Are Your Insurance Policy Protection against inappropriate outputs isn't optional - it's essential. System design should include fallbacks for when agents encounter problematic inputs, and regular guardrail evaluations help maintain safety as threats evolve. 🔟 Multi-Method Evaluation is the Gold Standard Combining automatic evaluations, LLM judges, and human feedback creates truly robust assessment. Data-driven approaches produce more reliable improvements than subjective "vibe checks," and evaluation metrics should align with business goals. Strong evals has been top of mind for the agentiiv team - hope this list helps others building in the agentic space.

47

In this newly released paper, "Fully Autonomous AI Agents Should Not be Developed," Hugging Face's Chief Ethics Scientist Margaret Mitchell, one of the most prominent leaders in responsible AI, and her colleagues Avijit Ghosh, PhD, Alexandra Sasha Luccioni, and Giada Pistilli, argue against the development of fully autonomous AI agents. Link: https://lnkd.in/gGvRgxs2 The authors base their position on a detailed analysis of scientific literature and product marketing to define different levels of AI agent autonomy: 1) Simple Processor: This level involves minimal impact on program flow, where the AI performs basic functions under strict human control. 2) Router: At this level, the AI has more influence on program flow, deciding between pre-set paths based on conditions. 3) Tool Caller: Here, the AI determines how functions are executed, choosing tools and parameters. 4) Multi-step Agent: This agent controls the iteration and continuation of programs, managing complex sequences of actions without direct human input. 5) Fully Autonomous Agent: This highest level involves AI systems that create and execute new code independently. The paper then discusses how values - such as safety, privacy, equity, etc. - interact with the autonomy levels of AI agents, leading to different ethical implications. Three main patterns in how agentic levels impact value preservation are identified: 1) INHERENT RISKS are associated with AI agents at all levels of autonomy, stemming from the limitations of the AI agents' base models. 2) COUNTERVAILING RELATIONSHIPS describe situations where increasing autonomy in AI agents creates both risks and opportunities. E.g., while greater autonomy might enhance efficiency or effectiveness (opportunity), it could also lead to increased risks such as loss of control over decision-making or increased chances of unethical outcomes. 3) AMPLIFIED RISKSs: In this pattern, higher levels of autonomy amplify existing vulnerabilities. E.g., as AI agents become more autonomous, the risks associated with data privacy or security could increase. In Table 4 (p. 17), the authors summarize their findings, providing a detailed value-risk Assessment across agent autonomy levels. Colors indicate benefit-risk balance, not absolute risk levels. In summary, the authors find no clear benefit of fully autonomous AI agents, and suggest several critical directions: 1. Widespread adoption of clear distinctions between levels of agent autonomy to help developers and users better understand system capabilities and associated risks. 2. Human control mechanisms on both technical and policy levels while preserving beneficial semi-autonomous functionality. This includes creating reliable override systems and establishing clear boundaries for agent operation. 3. Safety verification by creating new methods to verify that AI agents remain within intended operating parameters and cannot override human-specified constraints

326

19 Comments

Tried a small experiment on vibe coding for assessing the productivity of GPT-5. Tried similar experiments with Gemini Pro before, and also with perplexity pro. Together, as human-AI collaboration, we built up a system that can automate complex decision-making on faculty timetable preparation for a mid-size institute given all the constraints. This is a non trivial problem for a planner, given the variabilities involved. With a set of carefully written prompts, we ended up with about 500 lines of working code in under ten minutes. It read the data, optimized the schedule, resolved conflicts, and exported a clean result. I spent most of my time stating the rules and checking use cases, not typing syntax or correcting errors. As a trained academician we should not indulge in sweeping generalizations such as "gone are the days of the coders," based on one experiment. There has to be extensive and intensive research with industry-academia collaboration to understand fully the impact of such a tremendous productivity boost. When tools can draft usable scaffolds in minutes, the real leverage moves upstream and downstream: scoping the problem, modeling constraints, validating outputs, and stitching systems together. The value is in understanding the process, the data, the risks, and the goals—then making the whole thing work end to end. This is where academia and industry have to look beyond, starting from how courses can be restructured to become more cross discipline and digital focused to usher in "transformation". For decades, educational institutions—particularly local colleges—have measured success by the number of “job‑ready” developers they produce. This made sense in an era when software development was a scarce and highly manual skill. But in a world where AI can generate production‑grade code in minutes, this approach risks creating a talent surplus in areas where the demand curve is flattening. We should not be preparing our students merely to compete with algorithms—they will lose that race. Instead, the conversation must shift toward digital transformation—not just in the IT sector, but across industries and societies. The opportunities shaped by Industry 4.0 are far broader than writing code. For instance, rather than IT, operational technology OT has a great strategic role to play, and IT OT integration is a vast unexplored realm by behavioral and management science, going beyond the obvious technology. Similarly, although supply chain disruption is a vast research area, it is rarely taught as a curriculum except for being a minor part of OM. Other areas of interest can be interoperability and integration with business partners, Megaproject execution — deploying digital platforms to manage complexity, and risk. Platformization and digital ecosystem creation across technology landscapes are going to of critical essence for businesses, large and small. It is time that we work on these concepts in broader, cross disciplinary ways.

54

7 Comments

ScienceDaily covered a study of Thinkscape.ai performed at Carnegie Mellon University to test its effectiveness for large-scale group brainstorming. In the study, groups of 75 people were able to brainstorm together in real-time using Thinkscape. These large groups were able to efficiently build on the ideas of others and rapidly prioritize the top outcomes. Compared to traditional enterprise collaboration tools, the participants using Thinkscape felt "more collaborative" , "more buy-in" ,"more heard" overall. They also reported producing better ideas and reaching a better outcome. Here are some quotes from the article... "Overall, our results suggest that CSI is a promising GenAI-based method for brainstorming and prioritizing on a large scale," says Ganesh Mani, Professor of Innovation Practice and Director of Collaborative AI at Carnegie Mellon's Tepper School of Business. "The average Fortune 1000 company has more than 30,000 employees, so the ability to engage large groups in real-time discussions, brainstorms, evaluations, debates, assessments, and prioritizations could be a powerful collaborative way to solve problems, plan projects, forecast outcomes, assess risks, capture employee feedback, and encourage the cross-pollination of ideas across large companies." Christopher Dishop, Post-Doctoral Fellow of Organizational Behavior and Theory at Carnegie Mellon's Tepper School of Business https://lnkd.in/grRSqeWK Unanimous AI David Baltaxe Joshua Sitzer Noah Dropkin Ganesh Mani #thinkscape #collectiveintelligence #swarmintelligence #swarmAI

32

1 Comment

🌍 SVO vs. SOV — Why Word Order Matters in AI Most LLMs like GPT-4 are trained on SVO languages like English. But languages like Hindi, Tamil, Marathi, and Bengali follow SOV structure — where the verb comes last. 🔹 English (SVO): “I eat rice.” 🔹 Hindi (SOV): “मैं चावल खाता हूँ” (I rice eat) This isn’t just grammar trivia. It’s a fundamental challenge in AI. LLMs trained predominantly on English struggle to understand or generate SOV languages fluently unless fine-tuned properly — or built from the ground up on Indic data. This is why building truly multilingual LLMs isn’t just about translation — it’s about respecting the cognitive architecture of each language. India’s linguistic diversity isn’t a bug — it’s a feature. Let’s build AI that reflects that. #AI #LLM #NLP #IndicLanguages #MultilingualAI #GPT #OpenSource #LanguageTechnology #LLMinIndia #SVOvsSOV #BhashaAI

472

22 Comments

instead of ‘thinking machine labs’ we could build a ‘thoughtful machines lab.’ of course, thoughtfulness — which implies deliberation and good judgment — might require that we learn how to be thoughtful ourselves. we might need to wonder how we could differentiate AI from machines: how could we stop building AI in terms originally abstracted from levers, gears, incline planes, etc.? what if AI systems could be more than/different from machines? (asking this question is quite different from suggesting AI is alive or has sentience or is human.)

20

7 Comments

Stop Overthinking: A Survey on Efficient Reasoning for Large Language Models (Rice University, March 2025) A call to #humans and #LLM’s a like;) 😂 😂 Paper: https://lnkd.in/gX3xkZ47 Abstract: "Large Language Models (LLMs) have demonstrated remarkable capabilities in complex tasks. Recent advancements in Large Reasoning Models (LRMs), such as OpenAI o1 and DeepSeek-R1, have further improved performance in System-2 reasoning domains like mathematics and programming by harnessing supervised fine-tuning (SFT) and reinforcement learning (RL) techniques to enhance the Chain-of-Thought (CoT) reasoning. However, while longer CoT reasoning sequences improve performance, they also introduce significant computational overhead due to verbose and redundant outputs, known as the "overthinking phenomenon". In this paper, we provide the first structured survey to systematically investigate and explore the current progress toward achieving efficient reasoning in LLMs. Overall, relying on the inherent mechanism of LLMs, we categorize existing works into several key directions: (1) model-based efficient reasoning, which considers optimizing full-length reasoning models into more concise reasoning models or directly training efficient reasoning models; (2) reasoning output-based efficient reasoning, which aims to dynamically reduce reasoning steps and length during inference; (3) input prompts-based efficient reasoning, which seeks to enhance reasoning efficiency based on input prompt properties such as difficulty or length control. Additionally, we introduce the use of efficient data for training reasoning models, explore the reasoning capabilities of small language models, and discuss evaluation methods and benchmarking." Project Website: https://lnkd.in/gbr46XNy Lohith Srikanth Pentapalli Kelly Cohen Scott Phillips Tri Nguyen Summer Crenshaw Zachary Huhn Monica Turner

21