Jamus has been thinking about AI.
1 d ·
I’ve been thinking quite a bit about the future of AI in recent months, and had the occasion to share some of these thoughts in response to a bill about the National Productivity Fund (NPF) a few months back. First, some context: the NPF is not new (it was set up in 2010), and is but the latest in a long string of attempts by the government to raise Singaporean productivity (these include bodies like the National Productivity Board, the Productivity and Standards Board, SPRING, and EnterpriseSG). By and large, these efforts have had limited success. Total factor productivity—the broadest measure of efficiency and technological progress—has hovered in the low-single digits for much of our nation’s economic history, even (and especially) during our high-growth decades.
So, in principle, upping financial support to boost productivity is a laudable goal. But as the disappointing track record suggests, it simply isn’t enough to open the cash spigots and hope that productivity will magically appear (no matter how generous that spigot may be). One thing that bothered me about the Bill in particular was that it appeared to use relatively expansive language (basically, using the term “promotion of [not just] investment [but] economic activities”), which opens the door for non-productivity-enhancing spending. After all, some recent work by industry economists suggests that Taylor Swift concerts generate an economically-measurable impact. Would TayTay concerts then qualify for the NPF? Would F1 or Great Singapore Sale advertising? (Thankfully, the Ministry clarified that this was not the case, which is important to get on record. I still feel that the language could be better refined, but then again, I’m not a lawyer).
But more importantly, the dough should be (in my view) directed toward expenditures to boost productivity by encouraging development and adoption of the latest general purpose technology: artificial intelligence (AI) and the robotics revolution. AI and robotics are general purpose technologies in the sense that it can be applied to a great many areas, from advanced manufacturing to after-sales support to customer customization to knowledge work (and many more we haven’t yet conceived). Much like electricity, steam power, computers, and the Internet, we are only scraping the full potential of the technology, and AI’s effects are likely to diffuse into all areas of the economy over the next few decades.
Given how pervasive its effects are likely to be, it’s hard to imagine what won’t be affected by AI and robotics. The fear now is that, during the transition, a large number of jobs may become displaced or, at best, significantly altered. This could turn out very badly, if not well managed. Most general purpose revolutions have affected the lower end of the skills distribution, allowing older workers to retrain and upskill, and for younger ones to develop their human capital for the needs of the future. But given the speed of AI advances—Twitter took 2 years, Facebook a little less than a year, and Instagram 2.5 months to get to a million users, but ChatGPT took all of 5 days—we don’t yet know what sort of jobs and skills we need for the future AI-driven economy.
This leaves significant economic insecurity, led by fears of redundancy and income loss. We should keep our heads up and trust that new functions will eventually emerge (as it always has in the past in response to new general purpose technologies), but a safety net is also needed now. This safety net should include a comprehensive redundancy-retraining-redeployment pipeline for our workers, including a promise of reemployment for at least a probationary period, conditional on acquiring skills needed in the new position (with interim financial support during training).
Companies will also have to adapt. Of course, they typically have the incentives to do so, to remain competitive. But make no mistake: some firms that do so better than others will thrive, while those who don’t will be left behind, and subject to failure or takeover. So it comes down to speedy and successful adoption and deployment, if Singapore is to reap the benefits of the AI-robotics revolution. This is where our local companies have lagged behind; in applied research—the “D”evelopment part of R&D.
Governments can only do so much, but pushing our A*STAR entities to foster stronger pipelines with industry (or to be able to more easily spin off development and commercialization bodies) can help raise the currently below-par levels of private R&D. But R&D credits—offered by the government, and directed specifically toward the adaptation, adoption, and/or scaling-up of AI-robotics tech by firms based here—can also usher the process along. Importantly, officers in government funding bodies like the NRF and EnterpriseSG should be freed from excessive focus on meeting short-term KPIs, but instead adopt a more venture mindset, where (big) payoffs may only be realized after sufficient runway into the future.
Analogously, research and training opportunities should be provided for our workers, both those in technical fields, as well as those who aren’t. The natural channel now is SkillsFuture, which lists some 887 AI-related courses:
https://bit.Iy/3SbkdEG. We do not yet know the full scope of how AI will change the way we live, work, and play. But we know that it will, and the imperative is on us, as a nation, to embrace the promise and potential that this general purpose technology offers.
#makingyourvotecount