Three ways Singapore can build trust in Artificial Intelligence

Cyber security will be one of the most important areas for the Singapore Government to apply artificial intelligence capabilities, believes Richard Koh, Chief Technology Officer of Microsoft Singapore.

“For the safety of citizens' data, you have to think about employing some of these technologies to help protect it,” Koh says. “And that's paramount over anything else,” he adds.

Koh’s prediction is driven by a trend governments are seeing everywhere now: criminals “have a lot more tools, organisational capabilities and resources available at their fingertips”.

How this looks

Microsoft, for instance, can detect cyber threats and attacks signals from billions of transactions a month across the world. “What we're doing is using machine learning to look for signals of malicious malware and ransomware attachments,” he explains.

But it is crucial for governments to build trust in AI technology for it to be successful, particularly when used in sensitive areas like security.

“I would encourage the people who are developing these technologies to think about the social and moral aspects, more than anything,” Koh says.

There are three key things officials must keep in mind to achieve this. First does it improve our capabilities without destroying our society? “We’re not creating the technologies to beat the humans at games, but to assist and maximise our capabilities without degrading our human dignity,” Koh says.

One good example is computers’ ability to analyse large amounts of data. This is something humans struggle to keep up with, AI is much more proficient at, and governments demand more of. Officials “want to have the ability to collect data, manage it securely, but also reason over it and that requires a fair bit of AI and machine learning capabilities,” Koh says.

Trust

The second point is on the importance of accountability and trust in AI. “There must be strong privacy, security, transparency and compliance mechanisms ingrained in the system,” Koh says and “behind all of that – there must be a human, organisation or team that’s responsible for the decisions”.

Third, and the “biggest challenge”, is guarding against algorithmic biases. This depends on “how do the teams and people who are designing these services guard against unconscious biases,” Koh says.

“It starts from the basic design of our technology, and how people from all walks of life and abilities are able to access the technology,” he adds. As it develops further, AI has to be constantly trained using the right data and rules to ensure it remains inclusive.

Collaboration with Singapore Government

Microsoft is partnering with the Singapore Government’s Cyber Security Agency to share its AI intelligence on security threats.

More broadly, it is collaborating with many government agencies to infuse AI, in a large myriad of scenarios, from bots, to image recognition and broad ranges of its Cognitive Services. More broadly, Koh’s vision is to use AI to make interactions with the government simpler for citizens via chatbots.

“It has come to a point where human conversation, just like you and I are having right now, is probably one of the best interfaces, if you will, to conduct transactions,” Koh explains.

Chatbots can make interactions with government more conversational by connecting and bringing in information from across agencies. If a citizen goes to the wrong agency for information, an AI bot should be able to bring that information in from the relevant ministry - much like an official would ask a colleague for help.

“The vision is how do we grab the ministry’s bot and bring into the conversation in a much more humanistic way?” he says. This will allow officials to reorganise jobs and allow them to focus on more complex requests.

“We can free up frontline staff in the government to be able to tackle some of the higher level service requests where a lot of empathy and emotional aspects of it needs to be deciphered by a human being,” Koh explains.

While cyber security is one of the most promising areas to use AI, empathy, trust and inclusivity are what will make AI a success.

If you would like to learn more about this from Microsoft, fill in the form below to download their guide on the future of digital government. [download_form list_id="78e9f4a610"] Image by the Government Technology Agency of Singapore