IGF 2020 WS #270 AI-Powered Technologies for social inclusion

Thematic Track

Organizer 1: Arturs Vasilevskis, Tilde
Organizer 2: Anna Kotarska , European Language Resource Coordination
Organizer 3: Andrejs Vasiljevs, Tilde
Organizer 4: Liga Strautniece, Tilde

Speaker 1: Arturs Vasilevskis, Private Sector, Eastern European Group
Speaker 2: Anna Kotarska , Technical Community, Eastern European Group
Speaker 3: Jānis Ziediņš, Government, Eastern European Group
Speaker 4: Andrejs Vasiljevs, Private Sector, Eastern European Group

Moderator

Andrejs Vasiljevs, Private Sector, Eastern European Group

Online Moderator

Liga Strautniece, Private Sector, Eastern European Group

Rapporteur

Liga Strautniece, Private Sector, Eastern European Group

Format

Other - 60 Min
Format description: 60 min auditorium Tutorial / Group Discussion. Our planned workshop is a learning session where presenters will share their expert knowledge on language technologies and present best practices. Following tutorial part of the session, we would engage the audience and raise discussions about different practices and policies that can help to reach Sustainable Development Goals.

Policy Question(s)

1. ow do we ensure fair representation online and diverse access to content in one’s language? 2. What strategies could be developed to promote (better) information access online for older people, people living with disabilities, language minorities, refugees and other disadvantaged groups? 3. What is the impact of AI and other data-driven technologies in the exercise of social inclusion? 4. What societal and economic benefits are enabled by state-owned language technology platform?

- Challenges of multilingual community - Ensuring equal opportunity for language minorities and visually impaired - Providing social inclusiveness - Increase access to information - Data security, possible sensitive information leakage through unsecured, unapproved free online translation platforms - Opportunities for AI-Powered technology adaptation

SDGs

GOAL 9: Industry, Innovation and Infrastructure
GOAL 10: Reduced Inequalities

Description:

With 7,100+ languages our world is truly multilingual, yet many governments exclude millions of people by limiting their access to public services due to language barriers. The number of international migrants and refuges has reached 272 million, - a number that would constitute 4th most populous country in the world. Additionally, since information traditionally has been only in written form it is not accessible to people with sight disabilities. This is an ethical issue that brings attention to how the government includes all members of society to create an environment where everyone can access public services independently. And it is not just about accessing public services, it also about being able to access important and relevant information. For example, millions of people who speak lesser-known languages don't have a single resource on COVID-19 yet. The information gap is enormous – and lives are at stake. Governments must look for ways to ensure social inclusiveness for all and to reach Sustainable Development Goals otherwise we are left in a world with continuing cultural, economic, and political exclusion for minorities. Afterall, social inclusion is a human right and governments should provide it for all people. Equitable participation in health, education, economic and legal environments relies on freely available information access for all. Translation is integral to the delivery of multilingual public services and information exchange in multilingual societies, however, due to large amounts of content and dynamic information flow, it can’t be sustained by human translators alone. Artificial Intelligence powered technologies can ensure that public services are accessible to everyone, even if they don’t speak the official language. Public service websites can be translated in real time with a custom machine translator created specifically to understand the terminology. If a person is visually impaired, he / she can access the public services with speech technologies through the Automatic Speech Recognition or Automatic Speech Synthesis technology and in case they need assistance on any matter, an intelligent virtual assistant is always on duty. Latvian Public Administration has deployed a multi-language technology platform Hugo.lv that ensures social inclusiveness, information accessibility and data security. Hugo.lv benefits multiple stakeholder groups– language minorities, visually impaired, civil service and the state. It offers 3 main functionalities - text translation, speech recognition (audio to text), and speech synthesis (text to audio). It solves two major challenges at once – it increases accessibility to public services for language minorities and people with visual impairment, as well as help automate the translation process for civil service and ensure data security. Governments deal with large amounts of sensitive and confidential data that often need to be translated in short period of time. As this can’t be sustained by human translators alone, there is an urgent need for feasible machine translation service that can ensure the right degree of linguistic quality while guaranteeing complete data security and confidentiality. Often, text translated by public administration are sensitive and confidential as, for example, translations for procurement documentation, contracts, etc., and generic online translation service providers do not pay enough attention to data privacy and security. Custom secure machine translation can provide automation which increases the speed, mobility, and productivity of the translation process while speech recognition can significantly increase text transcription for legal materials. A translation platform owned by the state would manage the processed data, therefore, ensuring that there is no leakage of sensitive data and provides number of major possibilities and improvements for the public and civil service. This proof of concept for AI-Powered eGovernance has demonstrated great benefit to Latvian Public administration in terms of ensuring social inclusiveness and by optimizing translation workflow and ensure data security and protection. State owned language technology platform is customizable, flexible and scalable solution that other countries can implement in their strategy for social inclusiveness and data protection. It also shows a great potential for large scale international organizations. The planned agenda for our workshopl is as follows: 1. Introduction of AI for multilingual world, social inclusiveness and data security 2. Passing on know-how for state owned language technology platform – presenting the best practices adapted by Latvian Public Administration Use 3. Engage the audience in discussion about challenges of multilingual community, increasing access to information, ensuring equal opportunity, language diversity, society inclusiveness and policies that can advance multilingualism and data security. The discussions would be facilitated by online moderator. We hope to raise discussions with the presenters as well engage the audience with each other by sharing their challenges and outlook on the future possibilities. The aimed practical outcome is to help other countries, organisations and individuals to utilise AI-powered language technologies to build stronger communities that will help to fulfil UN’s 2030 Agenda for Sustainable Development.

Expected Outcomes

The desired outcome of our workshop is to educate the audience about possibilities of language technologies for social inclusiveness and data security. The goal is to raise discussions about language and technology policies that can support multilingual societies and individuals. The audience will learn about ready-to-use language solutions that offer great scalability, flexibility and customisation. After the session, we would launch a publication that explains how language technologies can tackle different challenges posed by language barriers and how countries can adapt in in their strategy for social inclusiveness. There could also be follow up events with interested parties to discuss this topic in greater detail and explore the opportunities for their country.

The first part of the session will be a tutorial where presenters will share their expert knowledge. The second part of the session will be an open discussion where moderator will encourage audience engagement in conversation about the topic. Onsite audience will be able to share their experience and ask question to the presenters or other audience members. Online audience will be able to also participate in discussion.

Relevance to Internet Governance: “AI-Powered technologies for social inclusion” engages the IGF community to discuss language technologies that can help to provide information accessibility and availability of e-services to language minority groups and ensure data security and confidentiality for sensitive government data.

Relevance to Theme: Governments have the opportunity to create inclusive information society by engaging all language groups and providing equal opportunity for accessing for everyone to access e-services. Language technologies can provide social inclusiveness, but it also contributes to data track. Governments require large translations amounts that contain sensitive and confidential information. If a civil service employee needs to instantly translate foreign-language document, he / she might use unapproved, unsecure online translation solution that doesn’t guarantee the necessary security measures. Such actions can put the security and future of country at stake if sensitive and confidential data is leaked.

Online Participation

 

Usage of IGF Official Tool.