18/03/2019 | Written by: Wouter Denayer
Categorized: Government
Share this post:
This morning in Brussels, the AI4Belgium coalition presents its recommendations for an overarching Belgian Artificial Intelligence Strategy. IBM is proud to have been a driving force in developing recommendations in 5 specific areas, together with a multi-disciplinary group of experts from the technology industry, trade organizations, universities and research institutes, governmental organizations and many more.
We strongly believe that the 5 pillars identified, if implemented, could enable Belgium to be a leader in Europe on AI.
Develop a responsible data strategy
For IBM a key factor for AI’s success is ensuring that it is trustworthy, which is part of pillar 2 on “developing a responsible data strategy”. Businesses are now seeking to use AI broadly across their organizations to improve customer service. Develop new products. Lower costs. Drive revenues and manage risk. In order to be able to move ahead with the adoption of AI, trust is crucial. We at IBM believe, that only by embedding ethical principles into AI applications and processes, can we build systems that people can trust.
The Belgian Government should promote the AI Ethics guidelines currently being developed by the European Commission’s High Level Expert Group on AI, of which IBM is also a member, within the Belgian business community and provide guidance on how these can be implemented.
Responsibly and transparently
For some, AI is like a black box. Data goes in and recommendations or insights come out. Therefore, we should develop and deploy ethics-embedded AI that is transparent, fair, explainable and secure. I believe that every organization that develops or uses AI, or hosts or processes data, must do so responsibly and transparently. People have the right to ask how an intelligent system suggests certain decisions and not others. This is why we should make AI systems transparent and trustworthy by starting at the foundation with the algorithms – from how they interpreted the task, to how they reached the output.
We should also state very clearly the major sources of data that we use to train AI systems. Transparency alone is not enough. Explainability is key. We need to shine a light on decision-making in ways that are clear to different audiences: an explanation must be tailored to the user who needs it. For example, an explanation of how a healthcare treatment was decided should be different if talking to a patient than talking to a doctor. We can provide this information through Fact Sheets, another concept that IBM is pioneering.
Government to lead by example
Another area where we believe AI has lots of potential is in government. We believe it’s important for the government to lead by example. This is recognized by the AI4Belgium coalition in Pillar 5, because AI offers the opportunity to develop better, faster and more affordable public services. It also allows an easier engagement with citizens, digesting vast amounts of data and improved decision-making.
We look forward to working with the current and the next government on developing and implementing Belgium’s AI Strategy. We are pleased that Ministers De Croo and De Backer have reacted so positively to the work of AI4Belgium and we look forward to the government implementing these recommendations as soon as possible.