In 2009, in his memorandum to heads of government agencies, Barack Obama said that government agencies should make information about their decisions “readily available to the public”. Citizens should know what their governments are up to, he believed.
For governments around the world, incorporating AI brings with it the challenge of ensuring transparency and eliminating unintended negative consequences. AI can bring innovation to government sectors like transport and urban planning but must be implemented responsibly.
Richard Jones, Vice President of APAC at Dataiku, shares what governments can do to ensure its algorithms are fair and transparent. He also explains the challenges that face the public sector in the war for AI talent.
How are governments using AI today
Governments are adopting AI across urban planning, transportation and customer service to name just a few. For instance, park managers are using machine learning to analyse the movement of people and crowds, notes Jones.
AI could detect “a massive change in what has become popular from a recreational landscape perspective and create numerous applications, ranging from route guidance optimisation, flows regulation, accident prevention to emergency services optimisation”, he adds.
Ultimately, this could better inform long term urban development plans.
Public agencies have also turned to AI for “service chatbots” to engage with citizens. Governments prefer to use these systems as the AI can interact with “the entire nation from any place at any time” leading to better communication with citizens.
Preserving Singapore’s ocean marine environment is another area where AI can help urban planning. AI can assess the impact of new digging or construction sites in the ocean – something that will become more relevant as the country’s population grows, says Jones.
The nation is using AI in its autonomous vehicles as well. The government plans to make transport “eco-friendly”, and is already trialling autonomous public buses, he shares.
While AI can be used for various uses across government, organisations public or private share the same challenge. They need to build an AI strategy that is scalable and responsible. This means that it should be, governable, sustainable for the future, white box, and free of unintended bias.
Cleaning up dirty data
Data needs to be valuable (high quality, labeled, and organised) to drive machine learning model success. “If you don’t have great data, you’re not going to be able to have great AI”, states Jones. Making sure “inherently dirty” data is ready for machine learning will help the AI to operate as intended.
Creating a standardised data labelling system powered by machine learning will help to make sure data is ready. If a single data labelling system was implemented across government, there would be a standard formula to understand how AI is making decisions, Jones explains.
With time, the algorithm would no longer need human oversight – not even for a simple approval process, Jones says. At that point, AI “can save hundreds, thousands or millions of hours” of manpower, he says.
The war for talent
Governments must also gear up for the war for AI talent, identifies Jones. They need the right skills “to be able to take advantage of the technological shifts that are happening”, he explains.
Having new staff who don’t know how to operate older systems is another challenge. Governments, among other organisations, struggle with “decommissioning and getting rid of legacy technology”, Jones says.
Tomorrow’s tech landscape
For all the benefits that AI can bring as part of the public sector, Jones insists that citizens must feel comfortable “accepting there may be an algorithm that is supporting decisions” that affect their lives.
Governments must provide the ability to “transparently trace back how decisions are made”, and present that in “easily understood” terms for all citizens, says Jones.
Dataiku’s platform provides immense help here, by supporting agility in organisations’ data efforts via collaborative, elastic, and responsible AI, all at enterprise scale.
He identifies blockchain as one potential advancement which could support transparency efforts.
Blockchain creates an “open agenda that everybody can see” through its decentralised system of hosting and securing data, he explains. This will allow citizens to know exactly how algorithms reach their decisions.
Creating “Explainable AI” is a critical task for organisations to recognise. Speaking about Dataiku’s role, Jones says we “enable a layman who has no understanding of AI” to see what sources were used and how it produced that outcome.
“AI isn’t a fad”, says Jones, and governments need to recognise this. They must be ready to implement open, transparent and responsible AI.
“Richard has been a regular speaker and panelist within IT Industry events across APAC over the last 15 years. He is based out of Singapore and has been living in APAC for 17+ years. He has spent the best part of 25+ years working in both vendors and customer landscapes. He has specialized within the Data Domain within customers & disruptive technology domain within vendors his whole career. Has led and scaled multiple companies across the APAC and is very excited to be building and leading the Dataiku team to support international growth. Hobbies are hiking, golf, cooking, snowboarding and deep-sea fishing, and anything that has an engine will be a sure way to get him to engage in activites. He is master training of NLP and fascinated to understand more how the human mind makes decisions and how the impacts of AI support everyday business activities.”
Dataiku provides an inclusive and collaborative data science and machine learning platform that makes data and AI projects a team sport, bringing intelligence to all — from AI builders to AI consumers. The company supports agility in organisations’ data efforts via collaborative, elastic, and responsible AI, all at enterprise scale.