That the Portuguese have many complaints about the service they receive from public services is nothing new. But the president of the Agency for Administrative Modernization (AMA) believes that artificial intelligence can help solve this problem. In an interview with ECO, João Dias explains that this technology can be useful not only by making employees more productive, but also by interacting directly with citizens, giving them the answers they are looking for.
According to the official, several tests are already underway in this direction. For example, the ePortugal website already offers an avatar based on ChatGPT and the State has also experimented with using artificial intelligence to analyze citizen complaints. Another possibility on the table is to apply this technology to public procurement, to filter applications.
The president of the AMA – which coordinates all the Citizen’s Bureaux in the country – stresses, however, that this transformation is only possible with properly trained civil servants, an effort he guarantees is being made.
This is one of two parts of João Dias’ interview with ECO. In the other part(which you can read here), he talks about the new service model at the Citizen’s Bureau, which will start at the end of this year, and also about the new government.
At a summit organized by the AMA, it was concluded that the public sector has difficulty attracting and retaining specialized talent. Salaries are the biggest obstacle. How can the employing state compete and attract talent?
There is a great deal of competition in the market for talent, especially in areas closely linked to technology. There is a lot of demand from private companies and even international companies. For the public sector, this means an even greater challenge, because we don’t have all the flexibility in recruitment, salaries, career development and meritocracy measures that private companies have. But we also have some assets.
What are these assets?
Fortunately, many young people are entering the job market today with a different paradigm. They value other dimensions than just the salary variable. In the case of AMA, and we have a very strong technology area, what we’ve noticed is that there are people who want to dedicate themselves to a cause and contribute to the greater good. This is a variable that weighs heavily on many people’s decisions. On the other hand, in the public sector, we have ideal conditions – often with access to very interesting information and data – for young people who are completing master’s degrees or doctorates to make practical applications of what they are studying, or even to do their theses.
He said that AMA has a strong technological component. Public administration as a whole is undergoing a technological transformation. Are officials being properly prepared in terms of training?
There’s little point in having state-of-the-art technology without… Well, AMA was one of the first organizations in Portugal, and perhaps at European level, to make Copilot – which is an artificial intelligence tool based on ChatGPT – available to most employees to increase their productivity. We did that, but we had a training program for people to get the most out of the technology. The two things go hand in hand. The greater the technological leap we make, the greater the commitment and care we have to take with training. We have taken this care and I think it exists in the state as a whole. I think the state is increasingly aware of the importance of training.
And did they manage to increase productivity with the introduction of this tool, as they wanted?
It’s very recent, but the oral feedback we’re getting from many of our employees is absolutely extraordinary. At the AMA, we have to follow all the European legislative production on very diverse digital issues. It’s a huge effort to read and analyze. Having a tool that helps systematize, summarize and make analyses is absolutely extraordinary. We already have many areas of activity where we are running dedicated pilots to use this technology. For example, in the area of public procurement. When we launch a large tender in the area of technology or outsourcing, several candidates come and all bring their proposals with hundreds of pages. Imagine having a tool that immediately makes a checklist of what you need to know. It’s no human substitute. But there’s a lot of routine work, with very little added value, that this technology speeds up and frees up the human, who can then do the qualitative analysis that matters.
In what other areas are you testing artificial intelligence?
Another field of application where we are also testing and the success rate has been 99% – now we’re actually going into production – is the analysis of citizen feedback, such as complaints, suggestions or compliments. We coordinate all the Citizen’s Bureaux in the country, as well as the Citizen’s Bureau. People can write a comment, a complaint, a suggestion, a compliment in the book or they can do it digitally. Analyze all this feedback by hand…
Was it Herculean?
It took up a lot of people. With artificial intelligence, we’re not only able to transfer the physical format of the book sheets to a digitized database, but we’re also able to have everything in that database and have artificial intelligence help us understand what kind of comments people are making the most and what difficulties they’re experiencing the most. This not only allows us to increase productivity, but also helps us to retain insights into the voice of the citizen to help us improve the service.
It’s no secret that there is a lot of criticism of public services from citizens. So can artificial intelligence help resolve this situation?
Yes. From the data we have from all the pilots we’re doing, I believe it can help a lot. In addition to operations, there is a whole other field of application for artificial intelligence, particularly generative artificial intelligence. We’re already testing it and we’ve been pioneers at international level.
What field is that?
If you go to the ePortugal website, you’ll see an avatar on the first page, which is the digital virtual assistant. Behind this digital figure is generative artificial intelligence, based on ChatGPT, which we train on the basis of models and information prepared by us.
What is the objective in this case?
It allows you to interact as if you were a human. In the old model, you had to call someone or browse the site. What this does now is that you don’t have to navigate, in other words, you ask it what you want to know and it writes it down. We ran a pilot in May for the Digital Mobile Key service and it went so well that we’re already working on extending the topics and services that this virtual assistant covers. This is to say that, in addition to the more operational issues – and it’s important that back-office processes in the State move quickly and efficiently – there is a whole new window of opportunity, which is in direct interaction with the citizen. In the future service model, we see that the virtual assistant can be this first informational contact.
Do you therefore foresee a reduction in the number of employees?
No. Unfortunately, there’s no point in hiding it, today we have queues in stores and people waiting to be served in contact centers. Often people are queuing to get information or a referral. If we give them the opportunity to talk to a virtual assistant, and the person gets the answer they need, I think we’re providing a good service. The person is no longer in the queue, they get the solution they need and, with fewer people on hold, the assistants start dealing with the more complex issues. Maybe they’re not under so much pressure because they know there isn’t a huge queue, and they can deal with the person they’re dealing with with more time, calm and depth. Basically, it’s a partnership between technology on the first level and then the human part.
Underlying these transformations are some questions. Is the state sufficiently trained in cybersecurity?
We have to distinguish between two types of data. There is data that is personal and falls under the so-called General Data Protection Regime. This is almost sacred. It is an unquestionable premise that we have to be – and we are – absolutely rigorous in preserving this data. Then there’s another layer of data. This is operational data, which is very useful for managing services. This is information that we naturally also want to collect.
