Welsh language technology and AI

Accurate, reliable, and useful Welsh language technology can help us increase the daily use of Welsh. This is one of the 2 main targets of Cymraeg 2050: A million Welsh speakers. Technology can make it easier for us to use Cymraeg in all aspects of life, from education and work to chatting with friends, family, and colleagues.

Over the past few years, we have done a lot of work to improve technology to help more people to use Welsh in everyday life. 

Our current priorities for Welsh language technology are:

• improving technology to increase the daily use of Cymraeg
• making sure everyone can access Welsh language technology
• improving Welsh language artificial intelligence (AI) and speech and language technologies (by sharing data and other means)

The Written Statement on Welsh language technology explains what we are doing to achieve this.

We’ll publish regular updates on this work.

We think that Welsh language interfaces, tools and resources should be available as easily as possible. We’ll continue to:

• work with partners to reduce ‘friction’ while using Welsh language tools and services so people can use their Cymraeg in more situations
• continue to promote our ‘trio writing’ methodology to make content and services as easy as possible to use yn Gymraeg

Digital Iaith Gwaith logo

The Iaith Gwaith (Working Welsh) logo is an effective way of showing that we can speak Welsh. It also helps us find other Welsh speakers. You may have seen the orange speech bubble on a poster at a reception or on a lanyard to advertise that a Welsh language service is available. 

We’ll be working with the Welsh Language Commissioner, who’s responsible for Iaith Gwaith (Working Welsh), to make it even more prominent in the digital world.

We want people to feel more confident to start more conversations and use more services yn Gymraeg. 

This will be part of our wider collaboration with the Commissioner on developing the use of Welsh language digital technology and services in workplaces and beyond.

Welsh language technology resources

Spelling and grammar checkers are already available. We’ll make sure that these tools are up to date. We’ll support their improvement and make them available on more platforms.

We’ll regularly update this page and our Helo Blod list of Welsh language tools and resources. 

We’ll continue to support the community of developers in Wales. We’ll work with them so they can create more content and services in Welsh.

Cymraeg belongs to us all. We want to make sure everyone can access Welsh language technology. This will make it easier for us to use more of our Cymraeg. When we say us, we mean everyone, wherever we are on our language journey.

We want to support digital advances to help people become more confident wherever they are on their language journey. Tools like the Learn Welsh Level Checker are already available from the National Centre for Learning Welsh. We’ll work with the Centre to develop the technology behind their tools to make it even easier for technology to help people use Cymraeg every day.

Parents and carers

We want to make it easier for parents and carers of school children to use more Cymraeg. We’ll:

• work with the schools’ sector to make it easier for parents, carers, and teachers to support children’s Welsh-language education
• share more news and updates, via Hwb, on appropriate technology and content
• expand our list of Welsh language tools and resources. We will include a section for parents and carers, to help children use more Cymraeg 

People with specific needs

An important part of our Welsh language technology work is helping people with specific needs. This could include:

• disabled people
• people with additional learning needs (ALN)
• people with needs resulting from an illness
• people with a cognitive or physical impairment

We’ll work to help everyone use more of their Cymraeg in tools where Welsh isn’t supported to the degree it needs to be.

We’ll identify gaps and offer data and resources to motivate companies to support Welsh. This includes screen readers and pen-readers for school assessments and better embedding of bilingual synthetic voices for people with visual impairments.

Harmful content

We want everyone to feel safe and protected from harmful content when they use technology in Welsh. We’ll work to motivate regulators and companies to consider how they will flag Welsh language issues.

Modern language technologies, like the large language models (LLMs) used by artificial intelligence (AI), rely on large amounts of training data to be:

• more accurate
• culturally relevant
• unbiased

We all need to contribute to improving AI and other technologies by sharing more:

• Welsh text
• video
• other Welsh data 

We already share Welsh language data with projects around the world. We share Welsh language data with:

• Amazon for their Massive database
• the AINA project, led by the Government of Catalonia in collaboration with the Barcelona Supercomputing Centre

We do this because we want to encourage more people to share more data. This can be by:

• sending data to Microsoft as a formal offer as part of our partnership
• openly licencing suitable data

For us to use more Cymraeg in AI, the output needs to be accurate, safe, and culturally relevant. The same is true for Welsh language and bilingual speech recognition. This needs to be able to recognise and accurately transcribe the Welsh we all speak with our colleagues, and our friends and families. 

As part of our partnership with Microsoft, we’ve already collaborated to develop a simultaneous interpretation facility on Microsoft Teams. 

This partnership now also include work to improve Copilot in Welsh: Microsoft’s AI tool. 

As we all become more used to speaking with AI assistants or using AI to summarise our meetings, we want to be able to chat with technology in the way that Welsh speakers chat in real life, for example ‘code switching’, which is when we use a bit of Saesneg in our Cymraeg and vice versa. We’re already working with international partners so that our work on this can help multilingual communities around the world.