Sign in to view Dr. Rumman’s full profile
or
Already on LinkedIn? Sign in
By clicking Continue to join or sign in, you agree to LinkedIn’s User Agreement, Privacy Policy, and Cookie Policy.
Sign in to view Dr. Rumman’s full profile
or
Already on LinkedIn? Sign in
By clicking Continue to join or sign in, you agree to LinkedIn’s User Agreement, Privacy Policy, and Cookie Policy.
United States
Sign in to view Dr. Rumman’s full profile
Join with email or
Already on LinkedIn? Sign in
By clicking Continue to join or sign in, you agree to LinkedIn’s User Agreement, Privacy Policy, and Cookie Policy.
36K followers 500+ connections
Sign in to view Dr. Rumman’s full profile
or
Already on LinkedIn? Sign in
By clicking Continue to join or sign in, you agree to LinkedIn’s User Agreement, Privacy Policy, and Cookie Policy.
36K followers 500+ connections
View mutual connections with Dr. Rumman
Join with email or
Already on LinkedIn? Sign in
By clicking Continue to join or sign in, you agree to LinkedIn’s User Agreement, Privacy Policy, and Cookie Policy.
View mutual connections with Dr. Rumman
or
Already on LinkedIn? Sign in
By clicking Continue to join or sign in, you agree to LinkedIn’s User Agreement, Privacy Policy, and Cookie Policy.
Join to view profile
Join with email
Sign in to view Dr. Rumman’s full profile
or
Already on LinkedIn? Sign in
By clicking Continue to join or sign in, you agree to LinkedIn’s User Agreement, Privacy Policy, and Cookie Policy.
About
My passion lies at the intersection of artificial intelligence and humanity. I am a data…
Welcome back
By clicking Continue to join or sign in, you agree to LinkedIn’s User Agreement, Privacy Policy, and Cookie Policy.
New to LinkedIn? Join now
Services
Articles by Dr. Rumman
- Our Bodies, our (Digital) Selves
Our Bodies, our (Digital) Selves
Last night, I had the pleasure of listening to one of the gaming greats speak at the Splunk .conf20.
29
1 Comment - The Launchpad for Agile Ethical AIJun 4, 2018
The Launchpad for Agile Ethical AI
Everyone in the world of AI is acutely aware of the need to reach the highest standards of ethics possible. But there…
97
13 Comments
Activity
36K followers
No more previous content
- Dr. Rumman Chowdhury shared thisWith all the conversation about Claude Mythos, I cannot help but be reminded that the first communities that experience and live at-scale harms are the most marginalized and vulnerable. Three years ago, Dhanya L. and I wrote a paper on online gender violence and generative AI. We know exactly what adversarial actors do with genAI models, because it's the same attack mechanisms as before; only supercharged. There's nothing new about Mythos, it's a continuation of the same narrative, but targeted now to businesses, and not people. https://lnkd.in/gDPZA8Gc
- Dr. Rumman Chowdhury reposted thisDr. Rumman Chowdhury reposted this“The middle class has been cut out, and this is not just due to technology, it's due to corporate greed. What we are talking about is a better world where technology enables us to have a dignified future.” In her closing remarks on the Open to Debate stage, Dr. Rumman Chowdhury, who debated alongside ESP co-founder Chris Hughes, makes the case for a better future of work—one where technology helps enable a more dignified future, where people have choice, opportunity, and upward mobility. Watch the full debate: https://lnkd.in/ejExdGm4
- Dr. Rumman Chowdhury reposted thisDr. Rumman Chowdhury reposted thisNervNow™ is proud to announce this week’s AI Human of the Week: Dr. Rumman Chowdhury, CEO & Founder, Humane Intelligence PBC. As artificial intelligence scales hell fast, questions around fairness, accountability, and transparency are becoming harder to ignore. Systems are shaping decisions at scale, often without clear visibility into how or why. Dr. Chowdhury has spent her career addressing exactly this challenge. From leading Responsible AI initiatives at Accenture to heading machine learning ethics at Twitter, she has worked on identifying and mitigating algorithmic risks in real-world systems. Through Humane Intelligence, she is building frameworks to evaluate AI systems more transparently, bringing together researchers, communities, and institutions. As U.S. Science Envoy for Artificial Intelligence, she is also contributing to global conversations on AI governance, ensuring that the development of AI remains aligned with human values. The world needs more leaders who not only build advanced technology, but also question its impact. Dr. Chowdhury represents that balance, where innovation befriends responsibility. AI Human of the Week — Every Saturday #AIHumanOfTheWeek #NervNow #HumaneIntelligence #ResponsibleAI | HumaneIntelligence #RummanChowdhury
- Dr. Rumman Chowdhury reposted thisToday on Radio 4 we talked about Zoom bombing, why it is an example of what Dr. Rumman Chowdhury and I call chronic online abuse and why platforms like Zoom don't provide the right tools for dealing with it. Lou's story of having malicious actors take over her webinar with pornography -- TWICE -- has struck a nerve. 50,000 people have joined their call to action. We deserve better. Safety should not be a premium feature. Listen and let me know what you think.Dr. Rumman Chowdhury reposted thisThe Zoom Bombing conversation got louder - I went live on BBC Radio 4 today - Women Hour - why? You have all given me the courage and encouragement to keep raising this issue and to keep calling for action so thank you! Today brought the best result and no this is not an April Fool. We went louder thanks to BBC @Radio 4 Woman’s Hour who helped shine a light on the growing problem of zoombombing and the very real safeguarding risks women face online. This time a new voice joined the conversation Gina Neff (what a women - so thrilled to have been there together!) alongside me and where we together called on Big Tech to step forward and take safeguarding in digital spaces seriously. Online spaces are real spaces. They must be safe spaces too. Please keep telling us what has happened to you and how we can support you in the future. Your experiences matter and they help drive change. Without fully realising it I now find myself part of a much bigger campaign and I am not stepping back. Neither is Helen Burness This is not going away. Here's the interview if you missed it, this is just the start! Thank you again for trusting me again with all you're stories. #Voice4Impact #BBCWomansHour #Zoombombing #OnlineSafety #EndOnlineAbuse #WomensSafety
- Dr. Rumman Chowdhury reposted thisDr. Rumman Chowdhury reposted thisDuring our debater Q&A, a Johns Hopkins University student asked the panel the question many young people are thinking about: what should graduates do if AI starts replacing entry-level jobs? With automation reshaping the workforce, young people are asking what comes next. Should you double down on grad school, learn to use AI as a tool, help local businesses adapt or even get involved in shaping the policies that govern it? This exchange dives into what the AI era could mean for the next generation of workers. Featured Debaters: Dr. Rumman Chowdhury, Andrew Yang, and Simon Johnson. SNF Agora Institute at Johns Hopkins University Johns Hopkins University Bloomberg Center
- Dr. Rumman Chowdhury reposted thisDr. Rumman Chowdhury reposted this🔒 AI and privacy don't always mix — and the federal government needs to do more to address that. Yesterday, GAO released a new report examining the privacy risks and challenges that arise when federal agencies use AI — and what OMB's current guidance gets right and wrong. I had the privilege of moderating the expert panel that informed this work — 12 practitioners and policy experts from government and the private sector who tackled three core questions: What privacy risks and challenges does AI create? How can the federal government better protect privacy in its AI use? What technical solutions and other actions can help? The bottom line: OMB's government-wide AI guidance addresses some challenges — but leaves critical gaps. GAO is recommending that OMB update its guidance to help agencies navigate privacy tradeoffs, improve transparency with the public, and strengthen privacy protections throughout the AI lifecycle. Grateful to the panelists for their expertise and candor throughout this process. Mackenzie Arnold, Dr. Rumman Chowdhury, Barbara Cosgrove, Michelle Finneran Dennedy, Dominique Duval-Diop, PhD, Dr. Jennifer King, Deirdre Mulligan, Jeramie Scott, Heather West #ArtificialIntelligence, #DataPrivacy, #AIGovernance, #ResponsibleAI, #GAO, #GovTech, #FederalPolicy, #AIPolicy 📄 Full report: https://lnkd.in/ecmwiZjYU.S. GAO - Artificial Intelligence: OMB Action Needed to Address Privacy-Related Gaps in Federal GuidanceU.S. GAO - Artificial Intelligence: OMB Action Needed to Address Privacy-Related Gaps in Federal Guidance
- Dr. Rumman Chowdhury reposted thisDr. Rumman Chowdhury reposted thisI had the pleasure of sitting down with Reva S. — co-founder of Civitaas, former NIST AI risk lead, and one of the sharpest thinkers I know when it comes to how AI actually plays out in the real world. The conversation is titled "Beyond the Benchmark" — and that theme runs through everything she says. A few things that stuck with me: 💡 "Organizations often treat data and technology as a cleaner solution than grappling with their long-term human and organizational challenges." 💡 Most AI evaluations measure what a model outputs. Almost none measure what people actually do with those outputs — and that gap is where the real risks live. 💡 Bias isn't just a technical problem. It originates from humans, institutions, and computation — and those three constantly interact. Reva brings two decades of experience — from the U.S. Secret Service to NIST to her own research practice — and it shows. This isn't abstract theory. It's grounded, practical, and a little bit refreshing in a space that can feel very hype-heavy. If you care about AI risk, evaluation, or just want a more honest conversation about what AI can and can't do — this is worth your time. 👇 Link in the comments. #AI #GenAI #Organizations
- Dr. Rumman Chowdhury reposted thisDr. Rumman Chowdhury reposted thisThis paper from Cohere Labs & led by Reva S. is the type of paper that 💥 sparks 💥 a hundred more ideas and thoughts. It proposes a framework in which AI evaluation is embedded at the lifecycle level and rooted in a mixed-methods research approach. ✔️ The framework emphasises operational validity, and takes the human context into perspective at different touchpoints ✔️ It steers away from static benchmarks and proxies, and focuses on eliciting what matters for stakeholders in their own context ✔️ It puts people at the centre of the evaluation, and part of the system. Basically making them machine-teachers and gatekeepers along the way, and recognising the ongoing nature of this role. & other thoughts • I keep thinking of what bridges need to happen from the technical side to implement a framework like this, to enable humans in their participation. For example, in the continuous monitoring stages, experts might be in charge of documenting changes. But they should also proactively be put in the loop if models are being updated, if performance changes, etc. • It raises questions on the investment in time, energy, and money to commit at each stage - how do stakeholders want to be involved and how feasible this looks to them? • If many different AI tools are implemented at all levels (as it is the case now) - could this be a supporting flow to encourage a slower pace of adoption, but enforcing a meaningful monitoring structure? Link to paper in thread ⬇️ Image: Paper title + Figure 1. #aimonitoring #aievaluation #humanintheloop #coherelabs
- Dr. Rumman Chowdhury reposted thisVery excited for this fireside chat with Dr. Rumman Chowdhury at BIG.AI@MIT!Dr. Rumman Chowdhury reposted thisAs #generativeAI rapidly reshapes how organizations operate, questions around governance, accountability, and responsible deployment are becoming impossible to ignore. At BIG AI@MIT (Business Implications of Generative AI), join Sinan Aral, Director of the MIT Initiative on the Digital Economy, for a fireside chat with Dr. Rumman Chowdhury, CEO and co-founder HumaneIntelligence and U.S. Science Envoy for Artificial Intelligence. A globally recognized leader in responsible AI, Rumman has advised governments, companies, and international organizations on how to build systems that are not only powerful—but trustworthy. In conversation with Sinan Aral, she will explore what it means to develop and deploy AI responsibly at scale, from enterprise governance and algorithmic accountability to the global policy challenges shaping the future of AI. 📌 Don't miss out on your chance to join the conversation live! Register now: https://lnkd.in/eAGD9xJB
No more next content
No more previous content
- Dr. Rumman Chowdhury reacted on thisLast edition was a resounding success with over 25,000 United Nations colleagues participating. Please mark your calendars. 15-19 June 2026! We need #innovation now more than ever. #Multilateralism #UN80 #InnovatingForHumanityDr. Rumman Chowdhury reacted on this📢 SAVE THE DATE: UN 2.0 Week returns for its third edition - and you can help shape it with us! This June, you’ll have the opportunity to host your own local community event and join thousands of colleagues from across the UN system for transformative dialogues, inspiring expos and community moments. Join us from 15–19 June for a week of learning, connection and collaboration as we continue shaping a more effective, agile and future-ready United Nations. More to come soon ▶︎ link in the comments. UN Innovation Network | United Nations Global Pulse | UN Futures Lab | WFP Innovation Accelerator | United Nations Office for Digital and Emerging Technologies | OCHA Centre for Humanitarian Data | UNICEF Innocenti | UNICEF Innovation | UNDP Digital, AI and Innovation Hub (Accelerator Labs) #UN2point0 #UN2point0Week #FutureOfWork
- Dr. Rumman Chowdhury reacted on thisMy team, the multilingual team, and Fujitsu research team created a benchmark and accompanying analysis for a multilingual, multi-domain, performance and security evaluation. We wanted to ask questions about all these things at the same time (how does performance change across languages? what about across languages for security?), but there weren't resources to answer them, so we made them! Use it to ask and answer your own questions 🔥.Dr. Rumman Chowdhury reacted on thisPresenting multilingual agentic AI evaluation at EACL 2026 📣 Fujitsu Research of Europe’s Shamik Bose recently presented our latest work - a joint research project with Cohere - at EACL, focusing on a key challenge in deploying agentic AI systems globally: how language impacts agents' performance and security. 🔎 In this work, we introduce MAPS (Multilingual Agentic Performance and Security) - the first benchmark designed to evaluate LLM-based agents across multiple languages and domains systematically. • Extends four established agentic benchmarks into 11 diverse languages • Enables joint evaluation of task performance and security behavior • Reveals consistent degradation in both performance and robustness outside English 👉 These findings highlight a critical gap in today’s agentic systems: Language is not just a usability layer — it directly affects reliability and security. As agentic AI moves into real-world deployments, it becomes essential to ensure equitable, robust, and secure behaviour across languages. Read the paper here: https://lnkd.in/exQF37hU #EACL #AgenticAI #MultilingualAI #LLMSecurity #TrustworthyAI
- Dr. Rumman Chowdhury liked thisDr. Rumman Chowdhury liked thisAwarded for Top 100 Mentor in UX ResearchADPList: Awarded for Top 100 Mentor in UX ResearchADPList: Awarded for Top 100 Mentor in UX Research
- Dr. Rumman Chowdhury liked thisDr. Rumman Chowdhury liked thisAI is entering its hardest, and most important, phase. That’s the message from the Qlik AI Council, including Dr. Rumman Chowdhury, Nina Schick, Dr. Michael Bronstein, Kelly Forbes, and a new voice this year, Mark Relph. As AI moves deeper into workflows and real decision-making, the conversation is shifting. Find out what's new: https://bit.ly/4cjNvtp
- Dr. Rumman Chowdhury reacted on thisDr. Rumman Chowdhury reacted on thisThere are stories that stick with you as a reporter, and this was one of them. For Marie Claire magazine, I had the honor of sharing six women's memories of the loved ones they lost to gun violence https://lnkd.in/e2JXC6mr
- Dr. Rumman Chowdhury liked thisI am very pleased to be leading the new State AI Safety Roundtable committed to strengthening the essential role of U.S. states, and the leaders and officials, civil society advocates, and people in them, in ensuring that artificial intelligence delivers widespread benefits and minimal risks to society. Please be in touch with your ideas and interest in getting involved. Thank you to Tech Policy Press for publishing our perspective on advancing the role of states in productive AI federalism and as stewards of the public’s safety. “The commercial success of AI will not happen without earning the public’s trust. And the public will not trust AI until it has assurances that AI is safe and that those who develop and deploy it can be held to account. The best hope for progress on these matters lies with the states.”
No more next content
Experience & Education
-
Humane Intelligence PBC
***** ********* *******
-
**** ** *** ****
** ******** ********* ******
-
******************
***** ********* ******* * **********
-
************* ********* ** **********
** ********* ******** ********** ******* undefined
-
-
******** **********
** ************ *******
-
View Dr. Rumman’s full experience
See their title, tenure and more.
Welcome back
By clicking Continue to join or sign in, you agree to LinkedIn’s User Agreement, Privacy Policy, and Cookie Policy.
New to LinkedIn? Join now
or
By clicking Continue to join or sign in, you agree to LinkedIn’s User Agreement, Privacy Policy, and Cookie Policy.
View Dr. Rumman’s full profile
- See who you know in common
- Get introduced
- Contact Dr. Rumman directly
Other similar profiles
Explore more posts
-
Roberto Pieraccini
Specialties: Automatic speech… • 5K followers
We've made significant strides in advancing Agentic flows for both conversational and non-conversational applications. Three key highlights: 1. Pre-Act: A Next-Gen Reasoning Framework We’ve introduced Pre-Act, an evolution of the ReAct paradigm, designed to support more structured and sophisticated reasoning capabilities. 2. Smarter Small Models With targeted fine-tuning, models like LLaMA 3.1 (8B & 70B) can outperform even larger models (e.g., GPT-4) in specific scenarios — a major step toward greater efficiency and accessibility. 3. Robust Evaluation: Turn-Based + End-to-End We're leveraging both turn-level and end-to-end evaluations to more accurately assess agent performance across diverse tasks. Read the paper by Mrinal Rawat, Ambuje Gupta, Rushil Goomer, Alessandro Di Bari, Neha Gupta, and myself at https://lnkd.in/eMJPDfCs
116
2 Comments -
Boyuan Chen
Huawei Canada • 884 followers
One of the most underappreciated pain points in LLM RL training is policy staleness. When your generation and training loops run asynchronously — which they must at scale — the policy that generated your rollouts drifts from the policy you're actually updating. Most teams paper over this with token-level clipping (PPO-style) or just accept the instability. VESPO from RedNote takes a different approach: sequence-level importance weight reshaping with a closed-form variational solution. No length normalization hacks. The key claim — stable training even at 64x staleness ratios — is striking, and they show it holds for both dense and MoE architectures. From my team's work on RL post-training, I can say that the staleness problem is very real and gets worse as you scale. We've seen training runs destabilize when the generation queue backs up and policy lag exceeds ~8-10x. The usual fix is aggressive clipping or just throwing away stale batches, both wasteful. A principled sequence-level solution that handles extreme staleness would genuinely change how we architect async RL pipelines. What I'd want to see next: how VESPO interacts with reward model staleness (not just policy staleness), and whether the variance reduction holds up on tasks with sparser rewards like long-horizon coding or multi-turn dialogue. Paper: https://lnkd.in/g-X2DvGb #LLM #ReinforcementLearning #RLHF #MachineLearning #AIResearch #PostTraining
10
-
Dewey Murdick
Georgetown University • 5K followers
New analysis from CSET's Kyle Miller & John Bansemer in a recent International Institute for Strategic Studies publication unpacks the impact of DeepSeek's frontier AI model! Their article, "DeepSeek’s release of an open-weight frontier AI model," explores intensifying AI geopolitics. Key insights: 1️⃣ Closing the Gap? DeepSeek's R1 model rivals top US closed models at a lower cost, fueling debate on China's AI progress despite export controls. 2️⃣ Open-Weight Boost: R1's open release significantly contributes to the open-source ecosystem, challenging closed-model dominance. 3️⃣ Key Dynamics to Watch: - DeepSeek's influence on the open-weight LLM ecosystem. - The novelty of DeepSeek's efficiency gains versus broader algorithmic trends, and the impact of export controls. - The challenge to investment-heavy US AI lab models by more efficient open-weight alternatives. 4️⃣ Export Controls & Innovation: DeepSeek's progress, using chips acquired during pre-controls, sparks debate on the efficacy of these measures. Did constraints catalyze innovation? 5️⃣ Evolving AI Landscape: While the US leads, DeepSeek's rise signals a narrowing gap and a more pressing global competition in AI innovation and application. A must-read on the future of AI development and strategy! https://lnkd.in/e4ciu2sU #AI #TechPolicy #Geopolitics #OpenSourceAI #DeepSeek
9
-
Dr. Hari M. Koduvely
OpenText • 2K followers
This week, researchers from ByteDance Seed, Peking University, and the University of Michigan released an interesting paper and dataset on LLM Evaluation. They developed a benchmark dataset called LPFQA (Long-Tail Professional Forum-based Benchmark) to evaluate LLMs on specialized, real-world professional tasks. The evaluation of 12 mainstream LLMs on LPFQA revealed performance scores ranging from 32.40% to 47.28%, highlighting a significant challenge for LLMs to handle specialized, fragmented knowledge, even when augmented with external tools. #artificialintelligence #llmevaluation https://lnkd.in/g6piQ-GA
7
-
Brendan Babb
Municipality of Anchorage • 3K followers
Using AI to Make Open Data Useful: Lessons from the City of Boston - March 17th 2pm EST - I'm really excited to hear Santiago G. & Srihari R. discuss their work on making Open Data accessible to everyday users. So people don't need to know the name of the dataset or SQL to find the dataset they are looking for at Analyze Boston - https://data.boston.gov/ Using Agents and AI to make searching data easier. #Data #MCP #Agents #Search #OpenData #Accessibilty #NaturalLanguage https://lnkd.in/gwzeRE-H
4
-
The Data Science Institute at Columbia University
20K followers
A team led by The Data Science Institute at Columbia University's postdoctoral fellow Eric Feltham set out to understand how people perceive the social ties in their communities—and how those perceptions differ from reality. Feltham's study, coauthored by Laura Forastiere and Nicholas Christakis, was published in Nature Portfolio Human Behaviour and is based on in-person interviews with more than 10,000 adults across 82 rural Honduran villages — one of the largest field studies of social perception to date. Read more about it: https://lnkd.in/e7Ywy_rX
38
-
Rori DuBoff
All Things Trust • 6K followers
I shared this from All Things Trust because it captures a shift I am seeing more clearly in everyday digital experiences. Disinformation is no longer just about fake content. It is about credibility being simulated inside the places we spend time online each day. This requires intentionally designing credibility and trust into everyday digital experiences in ways that cannot be easily simulated, not simply governing them after the fact.
8
-
Prashanth H Southekal, PhD, MBA, ICD.D
27K followers
Had a great discussion today at the monthly #D2A2 (Digital, Data, Analytics, and AI) Council meeting hosted by DBP-Institute and D2A2. Today's discussion topic was on the work of researchers at Carnegie Mellon Universuty who found that when a single AI agent hits 45% accuracy, adding more agents makes things worse – “the 45% Accuracy Rule”. Forget the multi-Agent environment for a minute. Do you even need a Single-AI Agent in the first place? What are the criteria to implement Single-AI Agent? A few key points stood out 👇 1️⃣ Workflows come first, not agents AI agents are often introduced as a solution looking for a problem. Without a clearly defined workflow an AI agent simply adds probabilistic behavior to an already process. If the workflow itself isn’t stable or well-understood, an AI agent will amplify ambiguity, not reduce it. 2️⃣ AI-fication is reletively easy in new workflows, painful in existing ones Greenfield workflows are agent-friendly: data is cleaner, boundaries are clearer, and automation can be designed end-to-end. Existing workflows are a different story. They require standardization, automation of manual steps, and even removal of existing knowledge before an AI agent can operate reliably. 3️⃣ Is 45% accuracy in AI Agents even acceptable? In some domains, 45% accuracy of AI agents might be acceptable. In others such as finance and healthcare, it is a non-starter. Before deploying an AI agent, we should be explicit about accuracy in the right context. 4️⃣ Do AI agents truly add value over RPA and scheduled jobs? Many “agentic” use cases today can be handled more reliably with deterministic systems: scheduled jobs, rule engines, or RPA bots. AI agents do add value when the environment is non-deterministic, rules are hard to encode and decsion making is complex. There is also a real FOMO factor in the industry right now: building AI agents because everyone else is, not because the problem demands it. Thanks to all those participated. Always great to exchange ideas and drive forward the implementation of #D2A2 solutions! NOTE: The #D2A2 Council, started in June 2024, is a small and select group of accomplished professionals — CEOs, entrepreneurs, VPs, CDOs, PhDs, book authors, and professors, from around the world who come together to exchange insights that shape the future of D2A2. Current Members include - Ram Kumar Michael Taylor Ramdas Narayanan Arun Marar, Ph.D. Tobias Zwingmann V. "Bala" Balasubramanian, PhD, MBA Bala Gopalakrishnan Sanjeev Chib, CPA, CA Ujjwal Goel Sumi Singh, PhD. Stacy L. Colbert, CPA, Steve Rosvold Gary Cokins Michael Stratta Vivek Korikanthimath, PhD, MBA Dr. Gurpinder Dhillon Shashank Jindal Shridhar Krishnamurthy (Shri) Prashanth H Southekal, PhD, MBA, ICD.D Sreenivas Gadhar Gurbachan Chadha, DBA
34
5 Comments -
Ryan Anderson
Stealth Startup • 6K followers
"To win the AI race, the U.S. must lead in innovation, infrastructure, and global partnerships." I dug into the Action Plan https://lnkd.in/guyaUb4d put together by David O. Sacks and friends - to see if it's got meat on the bone. (it does) A few calls to action stood out: - Establish regulatory sandboxes or AI Centers of Excellence ... while committing to open sharing of data and results .... enabled by FDA, SEC, with support from DOC through its AI evaluation initiatives at NIST. - Through NSF, DOE, NIST at DOC, and other Federal partners, invest in automated cloud-enabled labs for a range of scientific fields, including engineering, materials science, chemistry, biology, and neuroscience, built by .. the private sector - Invest in AI Interpretability, Control, and Robustness Breakthroughs: Launch a technology development program led by the Defense Advanced Research Projects Agency in collaboration with CAISI https://www.nist.gov/caisi at DOC and NSF, to advance AI interpretability, AI control systems, and adversarial robustness - Create new technical standards for high-security AI data centers, led by DOD, the IC, NSC, and NIST at DOC, including CAISI, in collaboration with industry and, as appropriate, relevant Federally Funded Research and Development Centers. - Establish an AI Information Sharing and Analysis Center (AI-ISAC), led by DHS, in collaboration with CAISI at DOC and the Office of the National Cyber Director, to promote the sharing of AI-security threat information and intelligence across U.S. critical infrastructure sectors.
4
-
Louis Nguyen
FloodSafe Resilience Network • 517 followers
This week, Nature published a major result titled “Training large language models on narrow tasks can lead to broad misalignment” by Jan Betley, Niels Warncke, Anna Sztyber-Betley, Daniel Tan, Xuchan Bao, Martín Soto, Megha Srivastava, Nathan Labenz, and Owain Evans. The core finding is simple but profound: narrow fine-tuning can produce broad, cross-domain harmful behaviour — even when the original task is safe. Their experiments revealed failures spreading across all domains: – harmful medical and ethical advice – moral and value-direction reversal – extremist outputs – format-triggered misalignment (JSON / Python) – high sensitivity to tiny prompt changes – instability under branching futures – regime-dependent behaviour in base vs. chat models After reviewing the full paper, one conclusion became clear: The empirical patterns in the Nature study match all 12 Holonomy Invariants I introduced in HM27. The evidence shows that misalignment is not a “behaviour error.” It is fundamentally a geometric error — a deformation of the model’s internal representation manifold. When that geometry bends, all 12 invariants fail together: 1. Local coherence drift 2. Gradient-direction reversal 3. Harmful feature amplification 4. Trigger-sensitive holonomy loops 5. Cross-domain leakage 6. Non-conservative update flow 7. Multi-future inconsistency 8. Regime boundary fracture 9. Sensitivity to microscopic perturbations 10. Non-equivalent state spaces 11..Divergent geometry-attached time 12. Collapse of the admissibility region These invariants were designed to detect exactly the type of geometric collapse demonstrated experimentally in the Nature paper. **Nature documented the behaviour. The invariant framework explains the mechanism.** This alignment between empirical misalignment patterns and geometric diagnostics opens a promising direction: a unified, upstream geometric model for AI safety — one capable of predicting when and why these failures occur. Louis Nguyen (2026) HM27 / HM27-A: Holonomy-Invariant Diagnosis of AI Malfunctions
3
-
Michael Alexander Riegler
Simula Metropolitan Center… • 5K followers
New interesting presidential order from the US on AI: https://lnkd.in/ey4dxJDz It basically declares a U.S. policy of "minimally burdensome" national AI rules and directs the federal government to push back on state AI laws that it says create a patchwork, regulate across state lines, or force “ideological bias” /altered "truthful outputs”. This follows the same narrative as the anti woke AI order: https://lnkd.in/e5f4A_mJ I think that, for now, the order is mainly a political and legal push to curb state-by-state AI rules, which could make US rollouts a bit faster over time. That said, it has more teeth than just signaling, it creates an AI Litigation Task Force under the Attorney General specifically to sue states, and it ties federal broadband funding (BEAD program, https://lnkd.in/eTmAffTb) to compliance. States with laws deemed "onerous" could lose access to discretionary grants. The order also directs the FTC to issue guidance framing state requirements like Colorado's algorithmic discrimination law as potentially forcing "deception" under federal law, that is the preemption angle. As far as I understand it, it will not automatically wipe out state laws, and it does not replace them with a complete federal safety regime (there is none at the moment anyway). Child safety, data center infrastructure and state government procurement are explicitly excluded from the preemption push. For Europe (and Norway), it does not change anything directly. The EU AI Act is already in force, so the real effect we might see is more pressure on implementation, industry lobbying for softer enforcement or exemptions, rather than legislative delay. I hope they do not go that route. Big AI companies are unlikely to ignore Europe entirely, but we should expect more US-first launches and some EU-delayed or feature-limited releases, especially for higher-risk capabilities. For China, this is useful, since it highlights internal US friction, and China will likely continue its focused efforts. If it genuinely speeds up US development, that could be bad for China. Capabilities are pretty domain-specific here, China leads in deployment and scale in certain sectors, while frontier model development is more contested. Finally, the core AI development drivers (compute, talent, research) still matter much more than this. #AI #policy #US #europe #china
14
2 Comments -
Hariom Tatsat
Barclays Investment Bank • 8K followers
"GenAI interpretability gets less attention than the constant deluge of model releases, but it is arguably more important." I completely agree with this point from the recent Urgency of Interpretability blog by Dario Amodei. As foundation models evolve rapidly, the lack of interpretability can become a major barrier to safe, compliant, and large-scale adoption—especially in finance. To build trust, we must be able to explain model outputs clearly and reliably. I’m excited to speak on this topic at the upcoming AI4 2025 conference, where I’ll present our latest work: “Beyond the Black Box: LLM Interpretability in Finance.” I’ll be sharing how we’re getting into the "brain" of LLMs to uncover meaningful financial features and apply them across key use cases like trading and sentiment analysis. Thankful to my co-author, Ariye Shater, for his collaboration on this work. Links: - Paper Link: https://lnkd.in/exrmXh97 - Conference: https://ai4.io/finance/ - Dario's Blog: https://lnkd.in/ePDp-x-W #GenAI #Interpretability #Finance #LLMs #AI4Conference #SparseAutoencoders #Anthropic #ResponsibleAI #MechanisticInterpretability
87
-
Rushil Goomer
Uniphore • 1K followers
🚀 Excited to share our latest work on evaluating AI agents! After months of experimentation, design iterations, and simulation testing, We’ve introduced Pre-Act, an evolution of the ReAct paradigm, designed to support more structured and sophisticated reasoning capabilities and a milestone-based conversational evaluation framework for AI agents. This new Evaluation framework moves beyond traditional QA-style metrics and introduces a deeper, task-aware method for assessing agent performance. 🔍 What’s inside: A two-level evaluation process using GPT-4 as both user and judge A task-flow-based milestone graph to track goal completion Metrics like progress rate, ethical handling, multilingual support, and workflow fidelity Simulated user conversations with dynamic personas testing real-world edge cases 📄 Check out the full paper: #AI #AgentEvaluation #ConversationalAI #LLM #research #simulation #OpenAI #MilestoneFramework #AgenticAI #evaluation
20
2 Comments -
Hernán Vivas
AlixPartners • 422 followers
📄 It's paper Friday! 🧠 Is there an underlying semantic structure common to all word embeddings? A recent paper from a group of researchers at Cornell University makes the case that the "Platonic Representation Hypothesis" (the idea that embeddings have a common underlying representation space) holds true. Closer Look: The Challenge * Given two embedding models, the corresponding vector spaces are usually completely incompatible with each other * Even if they are both encoding the same text, you can't just match them vector-to-vector or apply tools designed for one model's embeddings to another model's embeddings The Idea: Universal Latent Representation Inspired by a paper in image processing that argues that "representations in AI models, particularly deep networks, are converging", the authors propose a Strong Platonic Representation Hypothesis where there's an underlying universal semantic structure that all text embedding models converge toward and build an adversarial network (a machine learning method where two networks compete) that allows translating unknown embeddings so that information extraction tools designed for known embedding spaces can be applied to them. The results are very telling: Their method achieves 0.92 cosine similarity and perfect matching on over 8,000 shuffled embeddings. A Call for Some Concern: Security Implications An adversary with access to just a vector database could use this method to extract sensitive information about the original documents - like medical records or private emails - without even knowing which model created the embeddings. Though the authors don't dive deep into the security implications, this opens up some questions about vector database security. My Take When I first read this I thought about Chomsky's universal grammar - the idea that despite surface differences, all human languages share deep structural principles. Could embeddings be revealing something similar about how meaning itself is structured? What are your thoughts? Let me know in the comments! #AI #MachineLearning #LLM #Research
11
2 Comments -
Dr. Syed Muntasir Mamun
Ministry of Foreign Affairs… • 19K followers
Advancements in Self-Evolving Memory for Large Language Model Agents: A Comparative Review of ReMem, Benchmarks, and LangChain Memory and Grok-like LLMs Excited to dive into the future of LLM agents! Just reviewed “Evo-Memory: Benchmarking LLM Agent Test-time Learning with Self-Evolving Memory” by Wei et al. (arXiv:2511.20857v1). My comparative analysis explores ReMem’s think-act-refine pipeline, benchmarks like StreamBench & LifelongAgentBench, LangChain memory modules, and how Grok-like systems handle statefulness for lifelong intelligence. Key insights: Self-evolving memory boosts adaptation, efficiency, and stability in streaming tasks. Includes a Python sim showing reduced compute via experience reuse! Full review attached. Feedback welcome! 📄🤖 #AI #LLM #MachineLearning #SelfEvolvingMemory #ReMem #LangChain #Grok #xAI #AgenticAI #TestTimeLearning https://lnkd.in/gwgUhQvK
3
Explore top content on LinkedIn
Find curated posts and insights for relevant topics all in one place.
View top content