This wil be a short weeknote this week as I have 2 days off to do some Christmas shopping.
There are two things worth mentioning.
Using GenAI to create a project vocabulary
Adrián asked me to review a bit of content for the Service Manual relating to the Well-being of Future Generations (Act). Only the English content is currently prepared. But I was acutely aware that there was a lot of content about the topic on the Future Generations Commissioner and Welsh Government websites – including a great deal of subject-specific terms.
So I came up with an idea to use GenAI to create an initial vocabulary to help create the Welsh content. To go a step further, I think that there are situations where using this approach at the start of a project to facilitate the co-design of Welsh and English content together could work.
I’ve written about what I did here:
Using GenAI to create a project vocabulary
Welsh and Technology Event, Welsh Language Commissioner
Over 170 people attended the event online. This was an event about technology, and there were other presentations about:
- Welsh Government’s Hello Blod,
- the capabilities and limitations of LLMs with Canolfan Bedwyr
- a good session by Harry Thompson of Marvell Consulting and their bilingual approach by default.
Jo and I presented ideas for modern service design by: - solving the right problem
- focusing on the users
- collaborating
- considering technology as an infrastructure for good design decisions.
A question (valid enough, in terms of motivation) came at the end – prodding why we shouldn’t artificial intelligence to fill a gap in Welsh language provision? That is
Gruff Prys from Canolfan Bedwyr said in the first place (and I’m paraphrasing here…) that we should not use LLMs as an easy solution to fill a gap in langauge provision. Osian Llywelyn, Deputy Welsh Language Commissioner, reiterated that the principle of fair and equal service is important here.
It feels like this was a good discussion to have. Its unfortunate that we didn’t have time to pursue it with the time available. We often hear discussions about ‘what tech we should use’ or ‘how to adopt’ tech in certain contexts. I really don’t feel we have enough discussions about why we use technology in the first place.
The role of technology in problem solving
I didn’t contribute at the time, because I was still collecting my thoughts. But it’s clear to me that the temptation to see technology as a solution to organizational problems is very strong. But often that happens at the expense of users, because it’s easy to see ‘efficiency’ in terms of ‘making life easier’ for ourselves, the providers.
Offering equal and user-friendly Welsh-language provision is the organisation’s problem to solve, not the users’. The use of technology in itself won’t solve the users’ problems.
What users generally care about is that they need to use a service or fulfil a task as easily and effortlessly as possible. Ultimately, the users want to get on with their lives. Users don’t care what technology facilitates that.
Consider a service where:
- only a telephone line is available in English
- only chatbot is available in Welsh
The organisation adopted a chatbot, because the couldn’t hire enough staff for the call centre. Is that equal service for the user? If we assume that a chatbot’s capacity to solev a user’s problem is very limited, then it is highly unlikely that Welsh users will to be able to use a service without picking up the phone and eventually using the English service. Poor service for all. A headache for the Commissioner.
But what if the chatbot was a great solution to answer 90% of users’ questions? Splendid. But if so, why wouldn’t the service include:
- a chatbot that’s available in Welsh and English
- a streamlined bilingual telephone line for the other 10% of queries?
The point is, technology is useful, only if you’ve considered:
- the context of the users and their needs
- the context of the service administrators (such as telephone line workers)
- what problem you’re really solving
This note went on longer than expected, and I have shopping to do. Bye!