Rich Heimann

Are you trying to get from 0 to 1 in your AI governance journey? What if you’re already doing good work but need to accelerate your maturity? Are your Safety- and Rights-Impacting use cases being handled appropriately? Is your AI inventory stuck in Excel and, well, a little chaotic? We’ve solved that. I’m extremely proud to announce the AI Governance Jumpstart by Credo AI and Booz Allen Hamilton. This offering tackles the challenges above with a simple, straightforward set of tools and practices. We’ve designed it to eliminate your “Governance Debt” and turbocharge your AI governance work to meet key OMB requirements. One thing we’ve learned over the years is that requirements that seem simple on the surface are never straightforward to implement in your organization. Read more about how we approached this problem in the article that Dave Prakash MD and I wrote about this offering and reach out if you’re interested in learning more. A big thank you to the visionaries behind this work: Sarah Curtis, the Product Manager, and Stuart Friedman, Sandra Luksic, and Jack Williamson from my team who stressed over every detail these past six months. And Ehrik Aldana and the entire Credo team for bringing their secret sauce and world-class tooling. This is our first joint offering together and we’re excited to bring it to market. https://lnkd.in/exBShWv5

40

⚡️ A new edition of ICONCLASS is released! 🖼️ https://iconclass.org/ What is new in this edition of the Browser? This fourth edition continues to use the ever expanding dataset of sample images, but presents them in a different way, allowing you to zoom in and inspect the metadata for selected images. New data and new functionalities Visual search: based on computer vision technology, algorithms help to find visually similar pictures with re-usable Iconclass encoding. The Iconclass Bibliography is now integrated with the Browser and continuously updated. Updating the Iconclass Bibliography also means we are creating deep links to other subject and art authority systems, such as the Warburg Institute's Iconographic Database, the Princeton Index of Medieval Art, Wikidata, and Getty's Art & Architecture Thesaurus. Primary textual sources important for iconography are also tagged with Iconclass, providing subject access to e.g. the text of the Bible, Ovid's Metamorphoses, Cesare Ripa's Iconologia and Andrea Alciato's Emblematum libellus. For more information: https://lnkd.in/ejeGvhat

58

1 Comment

To achieve AAL2 for #passkeys as per NIST guidelines, User Presence = ✅️ and User Verification = ✅️ Blog post here: https://lnkd.in/gjg3yxtk

1

Yesterday had some significant announcements from the United States Department of Defense #FFRDC and #UARC community, with the next leaders of three major organizations being announced on the same day. Mark Peters will become MITRE's new CEO (https://lnkd.in/ev8vf2iw). Melissa Choi will become MIT Lincoln Laboratory's next Director (https://lnkd.in/eBqvFgUx). John Beieler will become University of Maryland Applied Research Laboratory for Intelligence and Security's next Executive Director (https://lnkd.in/eeS2XaHt). Personally I think this is a momentous point in time for the DOD and the FFRDC/UARC community. Technology's importance to the future of warfighting could never be stronger, but so too is the importance of technology in pre-conflict economic competition. DOD is in a unique position to integrate across these domains, and having three new leaders with such strong S&T credentials puts the FFRDC/UARC community and the DOD in an excellent position to succeed.

147

3 Comments

I've used RTI messaging software for some federal contracts for the Air Force, but here is an ad/blog on Digital Twins which provides some foundational information about concepts in this arena including a tie-in to the Army’s Unified Data Reference Architecture (A-UDRA). Moving the data between physical and virtual systems is key. #digitaltwins #simulation #3dmodel #sensors #realtime #feedback #movingdata

9

1 Comment

State of Colorado has enacted SB 24-205 (informally, "The Mile High AI Act") into law, with Governor Jared Polis signing the bill yesterday. https://lnkd.in/eQbrF7np To my knowledge, this bill is the very first example of enacted AI law in the Americas that is separate from, and not subsumed under, other overarching rules on information privacy, human rights, consumer rights, or online safety (albeit touching these important areas appropriately). My very high level takeaways: - The primary concern is to manage the risk of algorithmic discrimination in AI systems deemed high risk. High risk is defined explicitly in the industry sectors and civic areas listed in the bill. - Vendors must provide model documentation describing data governance, intended use and evaluation, known limitations. - Users must have risk management procedures in place (including red teaming if appropriate), and provide consumers mechanisms for recourse if they feel like they suffered harm. (Disclaimers: I am not a lawyer and this is my personal opinion)

24

1 Comment

Check out the #AI News Roundup for the Week of April 15 from the Mississippi Artificial Intelligence Network (MAIN)! One of the objectives of MAIN is to raise awareness about Artificial Intelligence (AI). MAIN helps facilitate this awareness by staying up-to-date on AI news. As part of this effort, MAIN compiles a variety of AI news every week and creates a news roundup. #MississippiArtificialIntelligenceNetwork #MAIN #MississippiAI #Mississippi #AI #ArtificialIntelligence #AINews #AIAwareness #News #NewsRoundup

11

1 Comment

Thomas Woodside's latest blog post delves into the contentious debate surrounding the "emergent capabilities" of large language models (LLMs). Despite widespread discussions, there remains significant confusion about what emergence in LLMs truly entails and its implications for policy. This explainer explores: ♦️ How the concept of emergence, from complex systems theory, relates to deep learning and how the term has been applied to LLMs specifically. ♦️ Real-world examples and potential risks associated with these emergent capabilities, such as unexpectedly effective autonomous hacking. ♦️ The academic and practical responses to emergence, including efforts to develop metrics that predict LLM behavior more accurately. Thomas emphasizes the need for more robust methods to anticipate and mitigate the risks posed by new AI capabilities, particularly those that might arise unexpectedly as LLMs scale. Dive deeper into this critical exploration of AI's unpredictable frontier, and discover why both skepticism and rigorous research are crucial as we advance:

7

2 Comments

CAP Index Part 2 of the Crime Data Series in LPM with a general Esri dashboard on crime in #Charlotte. User has the ability to filter based on #NIBRS Location Type, Crime Type, and Charlotte specific neighborhoods. #CrimeAnalysis #PlaceMatters https://lnkd.in/eyhnxDe2

12

2 Comments

Addressing Risk Aversion in Adopting User-Centered Design (UCD) within the DoD:The Department of Defense (DoD) faces challenges related to risk aversion when adopting UCD practices. However, there are strategies to mitigate this: Low-Friction Business and Operations: Creating a flexible and responsive environment Start Small: Begin with small UCD initiatives. Usability testing, for instance, allows early validation of assumptions and helps avoid costly rework. Embrace Feedback: Encourage open communication with end-users. Their insights can guide development and reduce risks associated with assumptions. Educate Leadership: Educate decision-makers about the value of UCD. Demonstrating successful outcomes can build confidence in the approach12. AI Applications for Logistics Optimization:Artificial Intelligence (AI) is transforming logistics in various ways:Demand Forecasting: AI-driven demand forecasting improves accuracy, optimizing capacity planning, inventory management, and transportation. Dynamic Replenishment: Real-time AI systems ensure optimal product availability by considering orders, inventory levels, promotions, and local events. Predictive Maintenance: AI predicts machinery or equipment failures, reducing unplanned downtime. Route Optimization: AI optimizes delivery routes, enhancing efficiency and reducing costs. Warehouse Management: AI automates warehouse operations, improving stock management345. Specific Projects Successfully Delivered by the Digital Capabilities Directorate (DCD):The DCD focuses on digitally transforming business and operations. While specific projects may not be publicly disclosed, their goals include:Faster Research: Accelerating research, experimentation, and innovation. Streamlined Transitions: Ensuring seamless entrance into acquisition and implementation. Better Decisions: Making analytically rigorous technical, business, and operational decisions. https://lnkd.in/exFcn_uK #Medium

1

All roads go through the AI Safety Institute The AI Safety Institute, located within the National Institute of Standards and Technology (NIST), and established by the Department of Commerce, features prominently in the national security memo (https://lnkd.in/e8TyFFfP). Disclaimer: I represent George Mason University in the AISI Consortium AISI's role is now crystallized as central to AI Safety, Security, and Trustworthiness. By the numbers: AISI mentioned 20 times in the memorandum and plays a critical role in ensuring the safety, security, and trustworthiness of AI, especially for the frontier AI models. Here is a summary of all the mentions of AISI in terms of its envisioned key responsibilities and actions: Serving as primary point of contact for AI developers with with private sector AI developers for model testing. Keyword: voluntary Conducting preliminary testing of frontier AI models: Within 180 days, AISI will pursue voluntary preliminary testing of at least two frontier AI models [..] to evaluate their capabilities and potential threats to national security. Issuing guidance for AI developers: Within 180 days, AISI will issue guidance for AI developers on testing, evaluating, and managing risks related to dual-use foundation models. Quite a challenging task. My own take is that we are far behind in understanding what it is we should measure and how we should measure. All we agree on is that we should measure. Developing benchmarks for assessing AI systems: Within 180 days, AISI will develop benchmarks [..] in various areas, including science, mathematics, code generation, and general reasoning. Aggressive timeline. Communicating findings to developers: If AISI identifies potential threats to public safety from a dual-use foundation model, it will be the primary channel for communicating findings and risk mitigation recommendations to the model's developer. My take: risk mitigation is an ongoing area of research. Submitting reports to the President: Every 270 days, AISI will submit a report to the President through the APNSA [..]. This is probably a good thing. The government does need to have a bird's eye view of the quickly-evolving landscape of AI models. Note here the emphasis on frontier.

7

🚀 Exciting news from the Defense Information Systems Agency (DISA)! 📢 DISA's Chief Data Officer, Caroline Kuharske, announced at the AFCEA International's TechNet Cyber conference that DISA is preparing to unveil an updated data strategy that will cover fiscal years 2025 through 2027. This new strategy will build upon the existing data strategy and focus on integrating data governance and policies into the backend systems. The previous data strategy, implemented from 2022 to 2024, aimed to improve data architecture and enable transparency and data sharing. DISA is now in the final stages of implementing this plan and transitioning to an integrated data architecture. But they're not stopping there! The upcoming strategy will also include initiatives to integrate emerging technologies, such as artificial intelligence (AI), into data operations. DISA will collaborate with its emerging tech directorate and leverage AI for data reporting and automation. This aligns perfectly with the Department of Defense's Chief Digital and Artificial Intelligence Office's 2023 Data, Analytics, and Artificial Intelligence Adoption Strategy. But DISA isn't working alone. They will gather insights from the DoD CDAO and partner with U.S. Cyber Command and JFHQ-DoDIN to deliver critical and actionable data to the warfighter. It's all about leveraging the power of data and AI to support our defense forces and ensure their success. This is an exciting step forward in the world of data strategy and technology integration. DISA's commitment to innovation and collaboration is commendable, and it's great to see them embracing the potential of AI and emerging technologies. Let's stay tuned for the official unveiling of DISA's updated data strategy in the coming months. I'm sure it will pave the way for even greater advancements in data governance and operations. What are your thoughts on this news? How do you think AI and emerging technologies will impact data operations in the defense sector? Share your insights in the comments below! 👇 #DISA #datastrategy #AI #emergingtechnologies #defenseindustry #innovation

1

I appreciate the NIST team. I think they do an excellent job keeping up with evolving technology and risk management. The supplemental guidance is very helpful when trying to understand the intent of various controls. #compliance #ITGC #audit

3

Should we fund AI monitoring and crime prevention systems? Enter "Real Time Intelligence Center (RTIC)," powered by AI. "Investigators can request information from [our] network of community stakeholders." "This free registry lets us know that you have a camera system, so we can contact you and request footage as needed." "The RTIC would also help gather and analyze advanced technology and data to monitor incidents and track criminal activity." https://lnkd.in/eSyAfd-x #AI #AIalignment #change1

3

4 Comments

Tune in for the ATARC Optimizing Operations: Gen AI in Federal Agencies webinar on July 18, 2024 from 1:30 to 2:30 PM ET! Generative AI is transforming the federal government by enhancing efficiency, decision-making, and service delivery. This webinar explores its #applications, such as automating tasks, analyzing large datasets, improving #cybersecurity, and developing predictive models. Follow along as experts discuss strategic implementation, addressing challenges like #data privacy, security, and system integration to create a more agile and responsive #government. Join here - https://ow.ly/plaL50SyLQp

1

😎Great news in the space of data-for-AI creation and management. The National Geospatial Intelligence Agency (NGA) in the US will issue a call worth more than $700 million for AI data labeling to help train AI-driven computer vision systems — including for its sprawling Maven program. 🛡Needless to say that any labeling mistake or "glitch" in the defense industry - like in the life sciences, legal and financial industries - can lead to serious consequences for humans and/or machines. 👉If you need a clear definition of data labeling explaining what it matters here is a good one⬇ "Data labeling is the process where the human actually identifies the object, and then, in a way that is understandable by the model, informs the model. And so you have to actually label it in a very specific way. In other words, data labeling works somewhat like how users pick out the squares showing pieces of a motorcycle or those showing birds to prove they are “not a robot” and get access to a website. It is a time-consuming task, and requires large numbers of individuals to review each piece of data and ensure its accuracy."

4

2 Comments

Academia has the unique ability to explore new ideas, conduct in-depth research, and challenge conventional approaches. By integrating these strengths into national security strategies, we can enhance our ability to anticipate, detect, and mitigate threats. Collaboration between academic institutions and national security agencies can drive innovation, research, and the development of the next generation of cybersecurity professionals. This is a powerful reminder that cybersecurity requires a collective effort across all sectors, including the invaluable contributions of our academic institutions. #NationalSecurity #Cybersecurity #Academia #Innovation #Collaboration

12

1 Comment