2017 just begun and 34 white-collar workers at Japan’s Fukuoka Mutual Life Insurance have already lost their jobs to artificial intelligence (AI).
The AI in question is IBM’s Watson Explorer which does the work of insurance claims workers by scanning hospital records and other documents to quantify insurance payouts.
Cost savings and increased productivity
The Watson Explorer costs Fukuoka Mutual 200 million yen (S$2.46 million) to install and takes 15 million yen (S$185,000) to maintain annually. Fukuoka Mutual is expected to save 129 million yen (S$1.59 million) on labour costs annually and will recoup its investment in just two years. The company expects productivity to increase by 30 per cent.
Knowledge work perfectly poised to be taken over by AI
In an era where the word “productivity” is bandied around conjuring images of blue-collar workers being replaced by more efficient robots, the fact that knowledge-based workers being replaced by AI should not come as a surprise. But it still does.
For years, school-going children have been told to aspire towards knowledge-based professions that bring not only prestige, but also economic advantage over more mechanical-based jobs — you’d still need doctors, accountants and lawyers in a recession, but not factory workers.
The advent of bots, machine-learning, algorithms and AI seriously challenge this notion.
Construction workers more irreplaceable than lawyers
The construction industry has been in a productivity dearth for many years now. It is because the low-cost foreign construction worker is too versatile and too cheap to give up and invest in machinery for.
For the low cost of $1,000 a month, a construction worker can help someone hold a ladder, manage traffic flow turning those ‘Stop’ and ‘Go’ signs, hose down tipper trucks, buy lunch for your head engineer, carry bricks into enclosed areas, drill things, spade dirt away, and much more.
Why would construction companies invest in machinery which costs far more and are unlikely to replicate more than two of those things mentioned above?
Knowledge-based workers, like lawyers, on the other hand are cost intensive and with a set of highly specialised skills. Wouldn’t profit-seeking companies have more incentives to look for ways to replace them?
“But AI can’t replicate creativity and experience!”
There are two main arguments against AI replacing knowledge-based workers: AI cannot make judgement calls, be creative or tap on experience; and creativity/experience/judgement are essential to these jobs.
Let’s take lawyers as an example. The workflow for lawyers involves receiving a case, gathering the necessary information/documents, looking up precedents, building the case, advising the clients, going to court, etc.
If we break down these tasks, not all of them require the ‘creativity’ or ‘experience’ deemed irreplaceable.
Laws, precedent cases, and other complex legal information can be coded into sets of algorithms that an AI can execute to solve legal problems, draft contracts, predict verdicts or tell you whether you are entitled to alimony.
Sure an AI cannot dumpster dive for evidence like what you see of passionate lawyers on television, but law firms can always hire cheaper labour for that.
The question that remains is whether creativity/experience/judgement in the execution can surpass what human lawyers can achieve.
But an equally pertinent question is whether AI can surpass humans in other aspects of “lawyering” that void the usefulness of, say, a senior counsel’s year’s of experience in law — a computer can recall thousands of cases and precedents, cross-analyse them and present a useful recommendation before the accomplished lawyer even reaches over the table to present you his rates.
Perhaps AI will not eliminate the need for lawyers, but it would radically change the nature of their jobs.
How about parliamentarians?
According to the parliament website, Members of Parliament (MPs) “act as a bridge between the community and the Government by ensuring that the concerns of their constituents are heard in Parliament”.
In other words, MPs are supposed to walk the ground, gather feedback, raise citizens’ concerns in Parliament, and help pass laws/policy to help the nation. And in Singapore’s case, help explain and communicate Government policy to their constituencies.
Following the town council act in 1988, MPs have the additional responsibility of caring for the estate’s development.
Should politician work be replaced by clever coding and computing muscle?
First we have to ask ourselves these questions:
Are views canvassed “from the ground” during MP walkabouts truly reflective of the general populace? Is there a better way to gather sentiment?
Are cases received during Meet-the-People (MPS) sessions being compiled into meaningful data to be studied? Is this data being used to predict problems in a particular estate or on the national level?
Are grassroots personnel truly good “feelers” for MPs or do they run the risk of being sycophants? How do we circumvent human bias when relying on a middleman for feedback?
Can the responsibility of estate development be transferred to the Housing Development Board or to a private company to be managed in a more effective and efficient basis? Can we rely on direct democracy via an app to vote for the type of resources we want? Afterall, we are already voting and deciding for the colour scheme of our HDB blocks.
Are politicians themselves optimised for providing solutions to a nation’s problems? How do we eliminate their personal motivations/affiliations when it comes to policy-making?
Are Singaporeans ready to trust computer algorithms over a politician?
Afterall, most MPs are not full-time MPs.
With AI, he, she or it will be working full-time or 24/7 for you.
Perhaps our sentimentality or “tribalism” is preventing us from voting for computer algorithms as opposed to political parties and candidates, but we must also acknowledge that emotions and sentimentality had a part to play in Brexit and Trump’s victory.