SETTING DEMOCRATIC GROUND RULES FOR AI

SETTING DEMOCRATIC GROUND RULES FOR AI: CIVIL SOCIETY STRATEGIES

https://www.ned.org/setting-democratic-ground-rules-for-ai-civil-society-strategies/

Published on October 19, 2023 by Amaris Rancy

DOWNLOAD THE REPORT

A new report by Beth Kerley (International Forum for Democratic Studies) analyzes priorities, challenges, and promising civil society strategies for advancing democratic approaches to governing artificial intelligence (AI). The report is based on conversations from a private Forum workshop in Buenos Aires, Argentina that brought together Latin American and global researchers and civil society practitioners.

SUMMARY

With recent leaps in the development of AI, we are experiencing a seismic shift in the balance of power between people and governments, posing new challenges to democratic principles such as privacy, transparency, and non-discrimination. We know that AI will shape the political world we inhabit–but how can we ensure that democratic norms and institutions shape the trajectory of AI?

Drawing on global civil society perspectives, this report surveys what stakeholders need to know about AI systems and the human relationships behind them. It delves into the obstacles– from misleading narratives to government opacity to gaps in technical expertise–that hinder democratic engagement on AI governance, and explores how new thinking, new institutions, and new collaborations can better equip societies to set democratic ground rules for AI technologies.

LAUNCH EVENT

Please join the International Forum for Democratic Studies for a virtual event launching the new report “Setting Democratic Ground Rules for AI: Civil Society Strategies.”  The Forum’s Beth Kerley will share key findings from the new report. Natalia Carfi, of Open Data Charter, and Eduardo Carrillo of TEDIC, will provide comments and share further insights on opportunities for promoting democratic approaches to AI. The discussion will be moderated by Ryan Heath of Axios.

The event will take place on October 24 at 10:00 am EDT on the National Endowment for Democracy’s YouTube channel.

HIGHLIGHTS: SETTING DEMOCRATIC GROUND RULES FOR AI

Advances in artificial intelligence (AI) are transforming political landscapes, impacting how people exercise their rights, and presenting new challenges to democratic principles such as privacy, transparency, accountable governance, and non-discrimination. Democratic AI governance is critical, yet significant barriers to engagement in this area remain. Drawing upon conversations from a private workshop in Buenos Aires, Argentina, the International Forum for Democratic Studies compiled an overview of eight key challenges to and opportunities for the democratic governance of AI.

• AI technologies reflect the human choices and structures behind them. The wide range of technologies described by the term “AI” are shaped by human choices about design and deployment, as well as the social and political contexts that feed into training data. Like all human products, they must be open to challenge by democratic activists and institutions.
• The risks and harms associated with AI challenge traditional assumptions. These impacts can arise at all stages of the AI pipeline, from development to procurement to use, and they may demand new ways of thinking about issues like data protection.
• Opacity around AI hinders democratic engagement. AI systems from surveillance cameras to social-media algorithms already work in the background of our daily lives, and the institutions that deploy them often prefer not to share the details. This reluctance, as well as the inherent complexity of AI systems, can make it hard to map the impacts of these tools.
• Addressing AI impacts will require more than just technical expertise. Because AI risks and harms have social and political roots, they will also require social and political responses. Moreover, these responses may sometimes demand trade-offs between competing democratic values.
• Democracies must close institutional gaps and widen participation in AI governance. Democratic institutions remain broadly unprepared to manage AI harms. Technical expertise on AI is concentrated in the private sector, which places democracies and their publics at a disadvantage in key decision-making processes—many of which exclude civil society and marginalized communities.
• New mechanisms and enduring democratic principles both have important roles to play. Democratic governance of AI may require building specialized institutions, but it also hinges on finding ways to apply existing democratic laws and principles effectively when AI tools enter the picture.
• Tech expertise within civil society can help influence the trajectory of AI technologies. Cutting-edge civil society groups are leveraging their technical skills to pinpoint government or corporate systems’ vulnerabilities; model more inclusive, representative, and responsible approaches to design; and develop AI tools to support civic accountability activities.
• The complexity of AI governance makes cross-sectoral collaboration crucial. AI governance challenges cut across traditional sectoral boundaries. New partnerships and knowledge-sharing initiatives that bring together digital rights groups, traditional human rights groups, journalists, trade unions, teachers, and others can enable civil society organizations to address these issues more effectively.

DOWNLOAD THE REPORT

---

“Setting Democratic Ground Rules for AI: Civil Society Strategies” by Beth Kerley is a report produced by the National Endowment for Democracy’s International Forum for Democratic Studies. 

MORE ON EMERGING TECHNOLOGIES FROM THE INTERNATIONAL FORUM:

---

The Digitalization of Democracy: How Technology is Changing Government Accountability, a report featuring insights from Krzysztof Izdebski, Teona Turashvili, and Haykuhi Harutyunyan

Smart Cities and Democratic Vulnerabilities, a report by Beth Kerley, Roukaya Kasenally, Bárbara Simão and Blenda Santos

The Global Struggle Over AI Surveillance: Emerging Trends and Democratic Responses, the first report in the Forum’s “Making Tech Transparent” series, by Steven Feldstein, Eduardo Ferreyra, and Danilo Krivokapic

Digital Directions a curated newsletter sharing insights on the evolving relationships among digital technologies, information integrity, and democracy.

Power 3.0 blog posts: “Putting a Thumb on the Market: The Rise of State-Aligned Platforms from Repressive Contexts” by Allie Funk, “Xi’s Pitch to the Global South on Technological Governance” by Kenton Thibaut, and “Bridging the Gap Between the Digital and Human Rights Communities” by Eduardo Ferreyra

Power 3.0 podcast episodes: “Digitalization and Democracy in Mauritius: A Conversation with Roukaya Kasenally” and “Can Democratic Norms Catch Up with AI Surveillance? A Conversation with Vidushi Marda” 

КОММЕНТАРИИ Сизова - Нейронная сеть 2

КОММЕНТАРИИ  Сизова - Нейронная сеть  2

Why use batch normalization?

Batch normalization offers several advantages for deep learning optimization and regularization. It can reduce internal covariate shift, which makes training more difficult. It can also improve gradient flow and enable higher learning rates, which can speed up the convergence and prevent local minima. Additionally, batch normalization acts as a regularizer that prevents overfitting and improves generalization by introducing some noise and randomness to the inputs of each layer.

Batch normalization in deep learning is like enabling auto exposure in a camera. Just as you adjust exposure for a perfect photo, batch normalization fine-tunes network layers. In photography, exposure relies on aperture, shutter speed, and lens filters, much like the normalization layer acts as a lens filter for a neural network layer. Correct exposure yields informative photos, preventing overexposed or underexposed areas - much like addressing covariate shifts in deep learning.

How do you balance analytical and intuitive thinking when solving problems?

Powered by AI and the LinkedIn community

Analytical and intuitive thinking are both valuable skills for solving problems, but they can also conflict or interfere with each other. How can you balance them effectively and use them to your advantage? In this article, you will learn about the benefits and drawbacks of each type of thinking, how to identify your dominant style, and how to improve your problem solving skills by combining analytical and intuitive thinking.

Combine analytical and intuitive thinking

In order to balance analytical and intuitive thinking, you can apply a few strategies to effectively combine them and use them to complement each other. This includes using analytical thinking to define the problem, gather information, and evaluate options, and using intuitive thinking to generate ideas, explore possibilities, and make decisions. Additionally, you should switch between analytical and intuitive thinking modes depending on the context of the problem and the availability of information. You can also seek feedback from others who have different thinking styles, challenge your assumptions and biases, and reflect on your problem solving process. By combining analytical and intuitive thinking in this way, you can improve your problem solving skills and achieve better results.

Konstantin Sizov

Founder of Drive Square, Inc., D2 Engineering and DUX.ECO

Henri Poincare wisely noted, "It is through science that we prove, but through intuition that we discover." Traditionally, a process of first brainstorming, relying on intuition to generate a set of ideas, followed by the validation of these ideas through analytical thinking is used. Irrespective of your thinking mix, there's a critical moment when analytical thinking must be engaged before action. This deliberate switch is vital, as it allows you to assess the potential risks and rewards of your decisions. Think of it as quality control, ensuring that your actions are rooted in sound logic rather than intuition, which can sometimes lead us astray. Balancing these two thinking modes can be the key to effective problem-solving.

What are the most innovative platforms for product development?

Powered by AI and the LinkedIn community

Product development is the process of creating and delivering new or improved products or services to meet customer needs and expectations. It involves various stages, from ideation and prototyping to testing and launching. To succeed in this competitive and dynamic field, product developers need to use innovative platforms that can help them design, collaborate, and iterate faster and better. In this article, we will explore some of the most innovative platforms for product development and how they can benefit you.

Все разработки продуктов для бизнес-администрирования

Какие платформы для разработки продуктов являются наиболее инновационными?

При поддержке искусственного интеллекта и сообщества LinkedIn

Разработка продукта — это процесс создания и поставки новых или улучшенных продуктов или услуг для удовлетворения потребностей и ожиданий клиентов. Он включает в себя различные этапы: от идеи и прототипирования до тестирования и запуска. Чтобы добиться успеха в этой конкурентной и динамичной области, разработчикам продуктов необходимо использовать инновационные платформы, которые помогут им проектировать, сотрудничать и выполнять итерации быстрее и лучше. В этой статье мы рассмотрим некоторые из наиболее инновационных платформ для разработки продуктов и то, какую пользу они могут принести вам.

ТАМ МНОГО РАЗДЕЛОВ - СИЗОВ ПРОКОММЕНТИРОВАЛ - 3

https://www.linkedin.com/advice/1/what-most-effective-ways-train-image-segmentation?trk=cah2&utm_source=share&utm_campaign=copy_contribution_link&utm_medium=member_desktop&contributionUrn=urn%3Ali%3Acomment%3A%28urn%3Ali%3AarticleSegment%3A%28urn%3Ali%3AlinkedInArticle%3A7117586296293724160%2C7117586298395107328%29%2C7119101561896673280%29&articleSegmentUrn=urn%3Ali%3AarticleSegment%3A%28urn%3Ali%3AlinkedInArticle%3A7117586296293724160%2C7117586298395107328%29&dashContributionUrn=urn%3Ali%3Afsd_comment%3A%287119101561896673280%2Curn%3Ali%3AarticleSegment%3A%28urn%3Ali%3AlinkedInArticle%3A7117586296293724160%2C7117586298395107328%29%29

Select a model architecture

The third step is to select a model architecture that can perform both segmentation and classification tasks. There are many types of models that can do this, such as convolutional neural networks (CNNs), fully convolutional networks (FCNs), U-Nets, SegNets, Mask R-CNNs, and others. Each model has its own advantages and disadvantages, depending on the level of accuracy, speed, memory, and complexity you want to achieve. You should also consider the trade-off between segmentation and classification performance, as some models may prioritize one task over the other. You can either use a pre-trained model and fine-tune it for your dataset, or build your own model from scratch.

Третий шаг — выбрать архитектуру модели, которая может выполнять задачи как сегментации, так и классификации. Существует много типов моделей, которые могут это сделать, например, сверточные нейронные сети (CNN), полностью сверточные сети (FCN), U-Nets, SegNets, Mask R-CNN и другие. Каждая модель имеет свои преимущества и недостатки в зависимости от уровня точности, скорости, памяти и сложности, которого вы хотите достичь. Вам также следует учитывать компромисс между производительностью сегментации и классификации, поскольку некоторые модели могут отдавать приоритет одной задаче над другой. Вы можете либо использовать предварительно обученную модель и точно настроить ее для своего набора данных, либо создать собственную модель с нуля.

CIZOV

Konstantin Sizov

Founder of Drive Square, Inc., D2 Engineering and DUX.ECO

In my experience, it's crucial to analyze prior solutions within the field for the given problem. Examining which types of ANNs have successfully tackled similar challenges can provide valuable insights. By following established pathways, we can avoid reinventing the wheel. In cases where no direct parallels exist, it's wise to explore related problem domains. It's essential to refrain from making rash “activist” architectural decisions based solely on familiarity with a specific network, as this often yields suboptimal results. I've observed that, in many instances, the most effective approach doesn't involve developing a custom model.

Константин Сизов Основатель Drive Square, Inc., D2 Engineering и DUX.ECO. По моему опыту, крайне важно проанализировать предыдущие решения данной проблемы в данной области. Изучение того, какие типы ИНС успешно решают аналогичные задачи, может дать ценную информацию. Следуя установленным путям, мы можем избежать изобретения велосипеда. В тех случаях, когда прямых параллелей не существует, разумно изучить связанные проблемные области. Крайне важно воздерживаться от принятия поспешных «активистских» архитектурных решений, основанных исключительно на знании конкретной сети, поскольку это часто приводит к неоптимальным результатам. Я заметил, что во многих случаях наиболее эффективный подход не предполагает разработку специальной модели.

https://www.linkedin.com/advice/1/what-different-roles-product-rd-how-can-you?trk=contr&utm_source=share&utm_campaign=copy_contribution_link&utm_medium=member_desktop&contributionUrn=urn%3Ali%3Acomment%3A%28urn%3Ali%3AarticleSegment%3A%28urn%3Ali%3AlinkedInArticle%3A7117586191507439616%2C7117586193701085184%29%2C7118776385162117121%29&articleSegmentUrn=urn%3Ali%3AarticleSegment%3A%28urn%3Ali%3AlinkedInArticle%3A7117586191507439616%2C7117586193701085184%29&dashContributionUrn=urn%3Ali%3Afsd_comment%3A%287118776385162117121%2Curn%3Ali%3AarticleSegment%3A%28urn%3Ali%3AlinkedInArticle%3A7117586191507439616%2C7117586193701085184%29%29

Product Owner

A product owner is responsible for representing the voice of the customer and the stakeholder in a product development team. They work with the product manager to define the product vision, strategy, and roadmap. They also prioritize the product backlog, write user stories, and accept or reject the product deliverables. They communicate with the development team, the customer, and the stakeholder to ensure that the product meets the user needs and the business value. To become a product owner, you need to have a strong customer focus, stakeholder management skills, and agile skills. You also need to have a deep knowledge of the product domain, the user personas, and the product features. You can prepare for a product owner role by taking courses, reading books, or joining communities related to agile and product ownership. You can also gain practical experience by working on agile projects, collaborating with development teams, or obtaining certifications.

Владелец продукта

Владелец продукта отвечает за представление голоса клиента и заинтересованных сторон в команде разработки продукта. Они работают с менеджером по продукту, чтобы определить видение продукта, стратегию и дорожную карту. Они также определяют приоритетность невыполненной работы над продуктом, пишут пользовательские истории и принимают или отклоняют результаты продукта. Они общаются с командой разработчиков, заказчиком и заинтересованными сторонами, чтобы гарантировать, что продукт соответствует потребностям пользователей и ценности для бизнеса. Чтобы стать владельцем продукта, вам необходимо иметь сильную ориентацию на клиента, навыки управления заинтересованными сторонами и навыки гибкой разработки. Вам также необходимо иметь глубокие знания о предметной области продукта, характерах пользователей и функциях продукта. Вы можете подготовиться к роли владельца продукта, пройдя курсы, прочитав книги или присоединившись к сообществам, связанным с Agile и владением продуктом. Вы также можете получить практический опыт, работая над гибкими проектами, сотрудничая с командами разработчиков или получая сертификаты.

https://www.linkedin.com/advice/1/what-most-innovative-platforms-product-development?trk=contr&lipi=urn%3Ali%3Apage%3Ad_flagship3_notifications%3BhX9t7FuITEqnN0iXAefq%2Fg%3D%3D&utm_source=share&utm_campaign=copy_contribution_link&utm_medium=member_desktop&contributionUrn=urn%3Ali%3Acomment%3A%28urn%3Ali%3AarticleSegment%3A%28urn%3Ali%3AlinkedInArticle%3A7112873299256201216%2C7112873300753543169%29%2C7117679032518774784%29&articleSegmentUrn=urn%3Ali%3AarticleSegment%3A%28urn%3Ali%3AlinkedInArticle%3A7112873299256201216%2C7112873300753543169%29&dashContributionUrn=urn%3Ali%3Afsd_comment%3A%287117679032518774784%2Curn%3Ali%3AarticleSegment%3A%28urn%3Ali%3AlinkedInArticle%3A7112873299256201216%2C7112873300753543169%29%29

Design platforms

Design platforms are invaluable tools that allow product developers to create and refine the visual and functional aspects of their products, such as user interfaces, logos, icons, and animations. They can also help them communicate their design vision and feedback with their team and stakeholders. Some of the most innovative design platforms include Figma, Adobe XD, and Sketch. Figma is a cloud-based design platform that enables real-time collaboration, prototyping, and feedback. Users can create and edit vector graphics, wireframes, mockups, and interactive prototypes, as well as share their designs with their team and clients. Adobe XD is a design platform that lets you create and prototype user experiences for web, mobile, and voice. It also provides access to a library of UI kits, icons, fonts, and plugins. Sketch is a design platform that focuses on vector-based UI design for web and mobile. This platform enables users to create and edit graphics, icons, layouts, and prototypes. Furthermore, users can collaborate with their team and sync their designs with other tools like InVision, Zeplin, and Framer.

Дизайнерские платформы

Платформы дизайна — это бесценные инструменты, которые позволяют разработчикам продуктов создавать и совершенствовать визуальные и функциональные аспекты своих продуктов, такие как пользовательские интерфейсы, логотипы, значки и анимацию. Они также могут помочь им передать свое видение дизайна и отзывы своей команде и заинтересованным сторонам. Некоторые из самых инновационных платформ дизайна включают Figma, Adobe XD и Sketch. Figma — это облачная платформа проектирования, которая обеспечивает совместную работу, создание прототипов и обратную связь в режиме реального времени. Пользователи могут создавать и редактировать векторную графику, каркасы, макеты и интерактивные прототипы, а также делиться своими проектами со своей командой и клиентами. Adobe XD — это платформа дизайна, которая позволяет создавать и прототипировать пользовательский интерфейс для Интернета, мобильных устройств и голосовой связи. Он также предоставляет доступ к библиотеке наборов пользовательского интерфейса, значков, шрифтов и плагинов. Sketch — это платформа дизайна, ориентированная на векторный дизайн пользовательского интерфейса для Интернета и мобильных устройств. Эта платформа позволяет пользователям создавать и редактировать графику, значки, макеты и прототипы. Кроме того, пользователи могут сотрудничать со своей командой и синхронизировать свои проекты с другими инструментами, такими как InVision, Zeplin и Framer.