The article below first appeared in COMMENTARY VOLUME 27, 2018 SGP 4.0: AN AGENDA.
Full PDF volume can be found here: http://www.nuss.org.sg/publication/1546997087_commentary2018_Vol27%20FINAL.pdf
Singapore 4.0 – the AI (Artificial Intelligence) Lap – will bring about exciting new opportunities but also challenges. Opportunities, which if seized at the right time will lead to great rewards and prosperity for Singapore. Challenges, which if not addressed may break our society and country. The new fourth generation or 4G leaders need to be not only politically savvy, but they also need to be business and technology-savvy. More importantly, they need to take intelligent and calculated risks to grasp the opportunities and challenges of “The Second Machine Age”, which is a term coined by experts on the AI era, Erik Brynjolfsson and Andrew McAfee.
State and Society’s Response to Technological Change
The steam engine brought about the Industrial Revolution, and the Big Data age of the Information Revolution over the past two decades introduced us to companies such as Google, Facebook and Alibaba. But it will be AI that revolutionises the compact between government and its people; between industries and its workers. Singapore has seen this kind of transition herself as factories, which used to hire thousands of workers have shifted their operations to lower-cost countries in the last decade. Globalisation and automation streamlined operations and made them easily portable across countries, often independent of the education level of the workers.
However, in the past, this displacement happened mainly to factory or other blue- collar workers. Not so in the Age of AI where the disruption will affect what we still think of as sophisticated and white-collar jobs. For example, AI systems can read X-rays and MRIs (magnetic resonance images) faster and as accurately as a human radiologist; search, compile and produce legal case summaries in seconds compared to lawyers who may take days; sieve through thousands of documents and transactions to detect fraudulent transactions in audits; compute a client’s risk profile to price an insurance premium in near real-time; advise a bank client on what shares or stocks to trade and/or invest in; and handle voice support calls that respond to customers’ simple, routine, FAQ-type questions.
On the other hand, for each of these challenges, there are new opportunities. For example, the radiologist can now focus on complex cases and reduce error rates; lawyers can now focus on analysing cases and being creative in planning how to win them; auditors can focus on complex human-to-human investigations supported by insights from the AI system; insurance companies can lower their risks, create more interesting products and offer more cost-effective policies based on individual risk profile instead of a generic profile; bank advisors can provide a more personal approach and focus on difficult or premium customers; and call operators will be relieved of responding to routine questions and can focus on difficult questions or those that require human empathy.
Most of the jobs mentioned above entail completing multiple tasks unlike those of production-line workers. In other words, AI will remove the routine, mundane and predictable tasks and allow us to focus on higher-order and higher-value tasks. A good example would be your humble SPAM engine, which is AI-driven and has helped save millions of hours from VIAGRA4U and DEAREST ONE.
We cannot stop the advance of AI, nor should we avoid it. We need to learn how to harness and even excel at AI to maximise the advantage Singapore can have in embracing it. At the same time, we have to try to minimise the negative impact of the displacement of jobs and workers. Some studies have shown that more jobs will be created by AI than jobs lost to it. Some of those jobs will require the worker to have AI skills and hence a higher level of education, but most will not.
The blue, white and grey-collar workers who do jobs that are in-between the first two categories, will need to learn how to race with machines and not against them. They will need to learn how to leverage AI tools to increase their own productivity and value; to position themselves as AI- enabled and AI-ready workers, engineers, executives and managers.
Some examples of companies that use data and AI to power their businesses include Netflix and Youtube. Closer to home, AI Singapore has within the last one year engaged with companies that are keen to undertake an AI project, including building up their own AI talent. Some of these companies are Surbana Jurong, Defence Science and Technology Agency (DSTA), Singtel, Daimler South East Asia and Johnson & Johnson. We expect to support them with up to 200 AI engineers via our AI Apprenticeship Programme initially. If Singaporeans are open enough to embrace technology and ride the wave, there will always be new job and career opportunities.
In that vein, the start-up ecosystem we have in Singapore must be one of those engines for the creation of new industries, businesses and jobs. So, it is imperative that the 4G leadership finds ways to strengthen the start-up ecosystem and encourage the creation of startup companies whether in deep tech or otherwise.
What are some other issues that will emerge with the development of AI as we try to ensure positive outcomes for jobs, wages and people?
Universal Basic Income
There have been calls by various groups globally for governments to implement policies like Universal Basic Income (UBI), which guarantees any adult an income regardless of whether they are employed. Another group of policies are on managing robot or AI taxes.
While it is unlikely that Singapore will adopt UBI anytime soon given the strong anti-welfarist orientation of the Government, we already have policies and government programmes which, if you peel away the acronyms like CITREP, TESA and SkillsFuture, represent a limited form of UBI.
For example, today, in AI Singapore which I am part of, we have the AI Apprenticeship Programme mentioned earlier which is a nine-month programme of self-directed learning and hands-on training in skills to build and operate AI systems. The programme is funded by The Info-communications Media Development Authority’s (IMDA) TESA programme and AI Singapore. The apprentices have their tuition and course fees waived, and are paid a stipend of between S$3,500 and S$5,500 per month. The stipend allows them to focus on learning a new skill and not worry about meeting daily expenses.
This approach of funding training costs, which ranges from a 70 to 100 percent subsidy and the provision of a stipend for some specific programmes is Singapore’s version of UBI. The Singapore approach is measured; we do not freely provide this “UBI” – you get “UBI” only if you agree to upgrade or re-skill yourself.
This has proven to be a powerful policy and it has allowed our precious tax dollars to be spent on nudging as many as possible towards acquiring new skills and knowledge.
Software Intellectual Property
In taxpayer-funded research, we need to review research and development (R&D) policies specifically where the intellectual property (IP) generated is software, and question policies where universities and research institutes hold on to taxpayer- funded research while making Singapore companies pay royalties for the licence to use them. It feels like these companies are being taxed twice.
Our IP policies, at least with respect to software R&D, are still based on the legacy model of closed source development where the source code is not released to anyone except under a fee-paying agreement, that is, through royalties or other commercial agreements.
However, the open source model of development and innovation have changed the economics of IP exploitation. In the open source model, the source code is shared freely with anyone under a friendly licence such as BSD, MIT or Apache 2.0. In particular, the Apache 2.0 licence is commercially friendly and has gained wide adoption.
The Apache 2.0 licence allows anyone to freely use, modify, distribute and sell a software licensed under the Apache Licence without worrying about the use of software, including patents. This is because the licence explicitly grants developers the copyright and patent of the derivative software. The rights given are perpetual, worldwide, irrevocable, but also non-exclusive.
Just look at examples such as Apache Spark, developed by UC Berkeley under a friendly open source licence, and subsequently commercialised by some of the students and original developers. Spark is now the de facto standard for big data processing and storage, and has created hundreds of thousands of jobs worldwide.
Or at Google’s Tensorflow, the most popular AI framework today which can be used for free. It has created numerous start-ups and the economic benefits that accrue are enjoyed not just by Google, but also by the whole ecosystem.
The advantage of the open source model is clear in the two examples above – people everywhere like free and good software. If the software is good, it will be adopted as the standard. Companies can then commercialise the adoption, and often the companies that succeed in commercialisation do include the original developers. After all, it is very difficult for an external party to come in and “own” the software if they are not part of the original development team or community. This community safeguard is stronger than any licence.
Back to Singapore – how much of our taxpayer-funded research has been exploited and commercialised to the likes of Spark and Tensorflow? Are our researchers in the best position to exploit the IP that has been created? Would it not be better if taxpayer-funded IP were in the hands of entrepreneurs via friendly open source licences? It will allow them to commercialise the IP in a faster and bolder fashion. It is not that our technology is not world-class. On the contrary, many are at the top across several fields. However, it is our legacy IP policies that have prevented researchers from being able to see the adoption of their IP by a critical mass of users. Instead, these languish in folders in the legal departments of some offices.
Be Agile, Be Bold
Singapore admittedly has lost its technological edge in a few areas like e- payments and digital courts. In contrast, China’s Supreme Courts already recognise blockchain-based evidence as being legally-binding. It is only recently that the Government here has embraced Agile IT development practices and open source technology. This, after the industry has been pushing for the adoption of Agile and open source technology since 2000. While it is understandable to want stable and proven systems, there are many areas, especially in information technology, where some risks need to be taken and where we innovate fast and fail faster.
We need the 4G leadership to allow agencies like GovTech and its partners to experiment, to fail and to learn from failure and iterate again; to allow universities to experiment with innovative ways of training the next generation of engineers, developers, management, thinkers and makers.
We need the 4G leadership to allow government agencies to be bold and experiment; to take risks with Singapore start-ups. The IMDA Accreditation scheme, which validates and accredits our local start-ups and SMEs, provides a green lane for them to access to government projects. This is the right move, but we can be bolder and allow agencies wider leeway to work with our Singapore local start-up companies.
What will the situation look like when that transformation is complete? Would we be able to view “SGP 4.0” as an upgrade, an improvement, or a paradigm shift altogether? Will it be achieving something that is long overdue?
A transformation is complete when the butterfly transforms from egg to larva to pupa and finally to adult butterfly. However, for a country like Singapore, there can never be a complete transformation. Just like we encourage Singaporeans to adopt a mindset of lifelong learning, the 4G leadership must adopt a mindset of lifelong evolution and transformation for Singapore. Our transformation will never be complete since technology never stops evolving. It will mean that we must continue to evolve, transform and adapt too.
About the author
Laurence LIEW is the Director for AI Industry Innovation at AI Singapore and is driving the adoption of artificial intelligence (AI) by the Singapore ecosystem through the 100 Experiments and AI Apprenticeship programmes. A visionary and serial technopreneur, he identified and introduced Singapore’s enterprises to Linux and open source in 1999 as the first RED HAT partner in Asia Pacific; High Performance Computing (HPC) Cluster and Grid computing from 2001 by deploying most of the initial HPC clusters in A*STAR, National University of Singapore (NUS), Nanyang Technological University and Singapore Management University, and architecting and operating Singapore’s first Grid platform – IDA’s National Grid Pilot Platform; and Open Source Analytics in 2011 with Revolution Analytics. Revolution Analytics was acquired by Microsoft in 2015.
He is Chairman, AI Standards Technical Committee, ITSC, Advisor to SGTECH AI+HPC Chapter; and member of the Technical Workgroup for IMDA National ICM Technology Roadmap – Track T4 (AI and Data, and Blockchain). Between 2013 and 2015, he was a Working Group Member of the National Infocomm Media Masterplan 2025.
He graduated from the National University of Singapore with First Class Honours in Engineering, and also holds a Masters in Knowledge Engineering from the same university.