Salesforce has set up its first international artificial intelligence (AI) research facility in Singapore, where it is aiming to train up to 100 postgraduate students over the next three years, and where the government hopes to help lead the development of AI standards and adoption across Asean. The US software vendor will pool the students from Singapore Management University (SMU), National University of Singapore (NUS), and Nanyang Technological University (NTU), and focus their training in deep learning, machine learning, and natural language processing (NLP), amongst other fields in AI.
The first batch of students will begin their training from August 2019 at the lab, which will be helmed by Salesforce Research Asia's managing director Steven Hoi.
The local facility also will work closely with the vendor's main research centre in Palo Alto, collaborating on projects that include NLP, computer vision, and speech recognition, as well as initiate research specific to the region, such as language-based capabilities in chatbots, said Salesforce's chief scientist Richard Socher, who was in Singapore for the launch on Wednesday.
Speaking to ZDNet on the sidelines of the event, Socher declined to reveal how much Salesforce was investing in the facility or how big the local team would be. Asked if there were plans to open more labs in the region or elsewhere, he did not dismiss the possibility but said his team's focus currently was on ensuring the Singapore centre was well integrated within the organisation.
He explained that the vendor's research unit focused on developing AI algorithms and models, while the wider Salesforce organisation provided the data needed to power the AI initiatives. Both would have to collaborate to ensure the research would eventually lead to practical applications.
At a panel discussion held at the launch event, Singapore's Minister for Communications and Information S. Iswaran said the government had been building a foundation for AI and ensuring there was clarity in terms of what AI could do and how it should be positioned within the societal framework. He underscored the need for governments to "set the right tone" so citizens were not "spooked" by negative perceptions of such technologies.
"In our context, we look at it as a frontier technology," the minister said. He noted that AI was an important application that the nation should learn to use, but there also was a need to balance this with considerations for ethical issues associated with AI.
Iswaran said: "The economics and value proposition aside, we must also address the ethical and societal dimension of AI, partly because it is still not completely well understood. The earlier we address it, the more transparent we are, the greater the potential.
"Governments need to support technological development and, at the same time, manage the equilibrium in society with accountability in managing ethical and public interests. We currently do this through frameworks that help inform how the public and private sector can use these technologies," he said.
In January, the Singapore government released a Model AI Governance Framework that outlined key ethical principles and practices in AI deployment, including in customer relationship management and risk management in autonomous decision-making. The document offered guidelines to help organisations ensure decisions made or supported by AI were "explainable, transparent, and fair to consumers" and that AI applications were "human-centric".
An advisory council also was established to assess the ethical and legal use of AI. The move aimed to build "a trusted ecosystem" and ensure consumer confidence, as the country continued to develop its digital economy and new business models emerged. Chaired by Singapore's former attorney-general VK Rajah, the Advisory Council on the Ethical Use of AI and Data comprised 10 other members including representatives from Google Asia-Pacific, Temasek International, Alibaba Group, and DBS Bank.
Iswaran said: "We need to ensure AI isn't a black box, but a transparent box. Allow people to understand what the technology is because that would foster greater trust and confidence, and diminish resistance to these technologies."
The Singapore government recently announced plans to establish an "inter-agency taskforce" later this year to asses how the country should develop AI as a strategic capability and be a global testbed for the deployment of AI applications.
One area it could look at was the need for more standardisation to ease the development of AI-driven applications.
Call for more standardised regulations across Asean
For instance, governments in the Asean region could work towards a similar set of regulations to reduce the complexity in tapping data to power AI initiatives, said Pan Yaozhang, head of data science at Shopee. The Singapore-based e-commerce platform uses AI tools to trawl its data and identify abnormal consumer behaviour, as part of its efforts to detect and combat fraud.
Speaking on the panel discussion, Pan noted that her company operated in seven markets across Southeast Asia and Taiwan, each of which had its own regulations around the use of data. "And we're struggling to deal with this," she said, adding that her team had to be mindful that what they could do with data in one country might not necessarily be permitted in another.
She urged governments in this region to adopt an "open mind", recognise the value of tapping data, and unify regulations governing the use of data. Such efforts, she said, would be impactful for Shopee.
In response, Iswaran acknowledged that access to data was an important element in AI since a larger dataset would result in a more effective application of the technology. He said the Singapore government was working to help create a more seamless environment for the flow of data within Asean and the wider region, pointing to the APEC Cross-Border Privacy Rules (CBPR) as an example of such efforts.
The region-wide initiative outlines certification mechanisms that ensure certified organisations have implemented data protection policies that are consistent with the APEC Privacy Framework. CBPR applies to data controllers, including organisations that control the collection, storing, processing, and use of data.
Iswaran, though, noted that there were legitimate concerns amongst governments about the data of their citizens as it moved beyond local shores, the protection that should be applied to such data, and the access of the data--whether it was to support enforcement activities or otherwise.
He concurred there was a need for a clear set of standards, similar to how a global set of standards had been established and widely adopted to ensure the quality of food products.
The minister also called on private organisations to work alongside and foster dialogues with governments, so AI policies could be more effective. On its part, he said the Singapore government was putting in efforts to lead such conversations in Asean and initiate the development of standards around AI, such as the classifying of sensor data.
Socher echoed the call for checks and regulations to allay any public concerns about the use, or potential abuse, of AI across the various industries. Self-driving cars, for instance, should go through mandatory safety checks, while medical applications should use algorithms that were trained across a broad set of data types, such as different genders and races.
He also suggested that the Singapore's Advisory Council on the Ethical Use of AI and Data could comprise academics, researchers, and philosophers to bring more varied perspectives in this area.
A specialist in NLP, the scientist believed AI had the potential eventually to replicate anything the human brain was capable of performing, but questioned the need and relevance of doing so.
Asked which concept in pop culture best depicted AI, he pointed to Her, which was a 2013 film about a man who grew close to an AI-powered virtual assistant. He noted that the popularity of apps such as Replika, for instance, demonstrated some people's affinity towards personal chatbots.
Singapore government announces plan to establish an "inter-agency taskforce" to asses how the country should develop artificial intelligence (AI) as a strategic capability and be a global testbed for the deployment of AI applications.
Singapore government is setting up an advisory council to look at the "ethical use of artificial intelligence and data" as well as recommend codes of practice and governance.
Country's government has introduced initiatives to train 12,000 people in artificial intelligence skillsets, including industry professionals and secondary school students.
Philosophers and psychologists may be needed in future corporate workspaces as businesses will need skills that machines and artificial intelligence cannot easily replace.
Government unveils a new scheme, investing up to S$150 million over five years, to use artificial intelligence to resolve challenges affecting society and sets up data science consortium to drive the sector.