Cerbrec’s Post

What impact is AI having on the public sector? A study of 89 government bodies has gathered some answers 👇 Ruth Kelly, Chief Analyst at National Audit Office, and Director of the Use of Artificial Intelligence in Government report provides interesting insights. The interview identified that skills, barriers to knowledge sharing, and legacy tech are the main barriers to AI adoption. For instance, 74% of government body respondents reported that they need “support around sharing knowledge and insight”. Ruth also emphasises the importance of a streamlined and co-ordinated governance strategy across the public sector. Addressing these fundamentals can prepare the public sector for AI adoption. I’m interested to see where we can create spaces in our ecosystem to achieve these goals 💭 #PublicSector #AI #Government #NAO #GenAI #Technology 🔗 Link to the full interview: https://rb.gy/5o7t01

One of the most critical challenges in AI/ML development is mitigating unintended bias. Drawing on his history of designing, developing, and deploying IT systems across the telecommunications and public sectors, Immuta Field CTO Chris Brown has identified three key steps teams can use to combat bias in AI/ML data sets. To better understand this bias – and learn how to counter it – read our blog today: https://lnkd.in/eshK32vY #AI #ML #AISecurity #DataGovernance #AIBias

The White House on Monday touted progress in the area of #artificialintelligence, saying that federal hiring has surged and funding is flowing to regional #AI research efforts while the federal government is preparing new regulations for the AI sector. Read More: https://lnkd.in/d9-jN8UE #ISMGNews #AIToday #genai #generativeai Rashmi Ramesh

📢 A cluster analysis of national #AI strategies 📌 This analysis reveals that certain countries are prioritising the realisation of the promises of AI while others are more concerned with mitigating its risks. 📌 The authors categorise the inclusion of six different attributes (e.g. #datamanagement; capacity development) in the countries’ strategies. On this basis they cluster countries whose AI strategies are on net high, medium, or low in the selected attributes. #governmentpolicies

A new AI Workgroup in New York with public- and private-sector members has been directed to create a greater understanding of AI for public-sector use. Read more: https://buff.ly/4gnVC9j #ArtificialIntelligence

Sunday Select: Stanford Institute for Human-Centered Artificial Intelligence (HAI) published its 2024 AI Index Report. The Report is a crucial resource for AI developers, policymakers, and individuals interested in AI. 📚 The Report provides a comprehensive overview of the current state of AI globally. Here are a few #keytakeaways from a high-level review: 1. Whether through benchmarking or #certification, industry must develop auditable standards to evaluate model performance. Some examples of emerging benchmarks in healthcare AI are MedPerf and MedAlign. From a testing perspective, methods like reinforcement learning from human feedback (RLHF) and reinforcement learning from AI feedback (RLAIF) are likely essential for model development and surveillance. Consumer-oriented explainability requires consensus from a model evaluation standpoint. ➡ 2. We're seeing an exponential increase in policymaking activity. At the Federal level, 181 bills were proposed in 2023, more than double the amount proposed in 2022. In addition, state-level privacy laws (many of them), federal agency bulletins, and international AI laws (e.g., the EU AI Act) are sprouting up regularly. #AI developers must be mindful of the commercial implications of a volatile AI and privacy landscape. ➡ 3. Agentic AI (AI that can autonomously complete complex workflows with limited human supervision) is coming into focus. The growing adoption of agentic solutions may challenge human supervision's status as a hallmark of AI #governance in sensitive domains. More nuanced accountability, standards of care, and liability themes will likely emerge as organizations commercialize agentic AI. ➡ 4. With great #data comes improved model performance and greater responsibility. From a healthcare perspective, AI developers should look towards existing privacy frameworks under HIPAA to inform sensitive data storage and security postures. Moreover, developers of healthcare AI solutions should assess responsibilities under the HTI-1 Final Rule, whether or not it applies, to future-proof their governance strategy. Lastly, developers should analyze their AI solution against FDA SaMD guidance. ➡ 5. It's all about risk. A global survey on responsible AI cited #privacy, data security, and reliability as key considerations for AI leaders. AI-specific risks, like algorithmic bias and drift, are difficult to quantify. In the meantime, organizations should establish protocols to inform the development and deployment process. ISO 42001 may serve as a baseline internal standard for managing AI systems. ➡ Best practices for AI governance are emerging from a crowded field of principles, bulletins, and agency guidance. We expect additional agency updates prompted by #EO14110, an uptick in state privacy law activity, and EU AI Act developments throughout 2024. 🚀 Rebecca E. Gwilt Reema Taneja Kaitlyn O'Connor Carrie Nixon More info:

How is synthetic / insilico data revolutionising the healthcare and life sciences market? Discover the latest trends, challenges and opportunities in this rapidly evolving landscape in our recent article: https://lnkd.in/dXxqmCcu #Data #AI #clinicaltrials #insilico Thanks and credits to Matteo Susta and Lidia Letterelli. Because at Simmons & Simmons we care about data and healthcare !        

Data privacy continues to be a big consideration with regards to AI. https://lnkd.in/eMDvh9wC

AI is revolutionizing government services, making them more adaptive, secure, and citizen-centric. This blog is a great short read that explores Google's commitment to responsible AI, ensuring powerful tools are developed ethically and transparently. #DigitalTransformation #AIinGovernment #Googleforgov