Leveraging AI in the public sector with open data
Publication Date/Time
2018-05-09T09:15:00+00:00
USING AI IN THE PUBLIC SECTOR IS POSSIBLE, BUT HOW? 

Artificial Intelligence (AI) software has been called the most
disruptive force in technology and it offers an unprecedented
opportunity to transform both the private- and public services sector.
It is a priority for the European Union and the European Commission
recently presented a series of measure
[http://europa.eu/rapid/press-release_IP-18-3362_en.htm] to boost
Europe's competitiveness in this field. In the public sector, AI
applications are still only used to a small extent though. A key
question is: how can the potential of AI applications in the public
sector be unleashed? Open Data can play an important role in realising
this.

 

WHAT CAN AI DO FOR SOCIETY? 

AI techniques can unravel deeper insights from sets of data than
traditional statistical techniques. Examples of AI are: speech
recognition, natural language processing, chatbots or voice bots. In
most cases, AI applications have at least one of the following seven
functions: monitoring, discovering, predicting, interpreting,
interacting with the physical environment, interacting with humans and
interacting with machines.
[https://data.europa.eu/sites/default/files/img/media/201819-aifunctions.png]
_Figure 1 AI functions_

 

By using AI, organisations can boost their productivity and optimise
their services through automating processes, optimising rational
decision-making and simplifying complex tasks. While the adoption of
AI applications in the public sector is not going as fast as in the
private sector, several noteworthy AI practices in the public sector
can be named. An AI system of The UK Ministry of Justice for instance
can sift, index and summarise documents as a human investigator would
do, but much faster and without human error. The AI sorts documents
into "privileged" and "non-privileged" piles, a time-consuming legal
process that would have otherwise been done by humans over many
months. The robot helped the team sift through 30 million documents,
processing up to 600,000 documents every day. Nevertheless, the use of
AI in the public sector is rather an exception than a common practice.

 

THREE PHASES OF AI 

In general, AI can be distinguished into three generations or phases.
In phase 1, AI is able to automate processes that are routine and do
not change over time. In phase 2 AI is able to adapt to- and learn
from changes in the automated process. In phase 3, AI proactively
provides new and innovative insights by being able to analyse and
learn from its previous actions.

Currently, this third phase of AI is the most advanced technology in
the private sector, but the expectation is that more complex AI levels
with larger potential value will be created in the future. In the
public sector on the other hand, some government departments have
adopted AI mainly for its ability to automate processes without making
use of its predictive analytic capabilities as seen in the private
sector. The most advanced AI level in the public sector at the moment
is in phase 2.

 

BRINGING AI AND OPEN DATA TOGETHER 

It is up to governments to take the best possible advantage of the AI
revolution. But what is needed for this? Three key ingredients are
crucial:

 

 	* _(Open) Data_

Obviously, AI systems need data to function. To function properly 1)
data volume, 2) data variety and 3) data veracity are crucial for
getting AI systems to work. This is especially important to avoid
biases and possible discrimination as machines start learning by
themselves on the basis of the data that has been provided. In
addition, Open Data is needed to assist public organisations in
fostering innovation. Open Data [/elearning/en/#/id/co-01] is data
that is made available by mostly public organisations but also by
businesses and individuals for anyone to access, use and share. Access
to open data will unlock the potential of data-hungry AI applications
for organisations in both the public sector and the private sector.

However, there are several barriers
[https://www.europeandataportal.eu/sites/default/files/edp_analytical_report_n5_-_barriers_in_open_data.pdf]
that still hinder the re-use of open data. The barriers faced can be
clustered into six categories:

 	* Political barriers;
 	* Organisational barriers;
 	* Legal barriers;
 	* Technical barriers;
 	* Financial barriers;
 	* (lack of) Awareness.

Next to open data, two other factors are crucial to fuel the AI
revolution in the public sector:

 	* _An innovative mindset_

Making progress on AI means that the mindset of 'business as usual'
needs to change. For organisations in the public sector it is
therefore important to foster a culture that embraces innovation,
brings out creativity and eliminates barriers to innovation.

 	* _Necessary skills _

It is vital to have the necessary skills in the workforce to be able
to identify where AI should and should not be used, and to help build
effective tools and systems. Increasing investment in a competent
workforce will help in upgrading the skills and knowledge on AI use in
the public sector.

 

WHAT WILL THE FUTURE BRING? 

The combination of AI and Open Data is a winning one. However, efforts
have to be taken to realise the benefits of AI in governmental
organisations. An important step in the right direction is for
governments to make their data available for re-use. By doing this,
this data can be used for innovative purposes such as AI. Only then
can governments prepare themselves for the future and move to a new
level of AI. Next to this, the ambition of the European Commission
with a comprehensive European approach to artificial intelligence
[https://ec.europa.eu/digital-single-market/en/artificial-intelligence]
and robotics shows that AI is seen as an important priority that
Europe is responding to.

It is expected that the possibilities of AI will further expand in the
future as the amount of data increases, for example because of more
real-time sensors that are increasingly being installed in cities and
on highways. Sensors can function as an additional source for AI
applications and some cities already have a policy
[https://data.eindhoven.nl/explore/dataset/eindhoven-open-data-principes/?flg=nl]
on opening up the data of sensors in public spaces. Large-scale use of
AI can have a significant impact in many domains, for example in the
healthcare sector, the educational sector, transportation, public
order and safety. Open data can help realise this potential.

  