Bowman: Open the pod bay doors, HAL.
HAL 9000: I am sorry, Dave. I am afraid I can not do that.
-- 2001: A Space Odyssey
Systems Data Wisdom Process Optimization
One movie changed everything
A childhood movie left a deep mark. I am still following its thread. Its protagonist was ‘HAL’, a sentient computer. Yes — the movie was 2001: A Space Odyssey. In it we saw the tug-of-war between humans and machines that can think.
Why do people want intelligent machines?
For a century, our fascination with Jarvis from Iron Man, the Terminator, or KITT from Knight Rider has been a powerful expression of the desire for intelligent machines. Whether machines can truly think like humans matters less than the fact that humanity refuses to give up on the idea.
Machines excel at specialized work
To solve the big problems ahead — curing disease, DNA sequencing, natural disasters, exploring galaxies — we need intelligent systems by our side. Humans need machines for their own sake too. Think about it: most global business transactions run on credit cards. Last year’s AliExpress 11:11 festival sold $1 billion in the first 68 seconds. Records follow. That volume of credit-card data cannot be fraud-checked by thousands of people alone. Intelligent systems have to help.
We depend on machines
I am no exception. My extra dependence on machines for everyday household routine is one expression of this. Assuming “our” work is different, we must delegate routine tasks to machines so we can keep our bigger picture. Of course we want machines that can think a bit like us. When the task is rule-like (somewhat like IFTTT), that works. But when the unknown appears, machines must decide like humans. We already see this in disease diagnosis, self-driving cars, long-haul flights, and on-demand service allocation in data centers — most of the time rule-based machines keep the system running. In short: we want machines as partners at our side.
Huge populations need process optimization
How long it takes today to get a government service — when “thinking” machines work hand-in-hand with humans, processing time for permits, licenses, trade licenses, passports and more will certainly shrink. We marvel that credit cards verify online; those tasks are simpler and largely process-driven. To rise above human bias — who gets service and who does not — we need systems like these.
Government and private systems must work at scale
Our work has to happen at scale. To serve 160 million people, processes must be automated. Traditional software works today but will fall behind. My point is only this: alongside conventional methods, intelligent systems should step in to remove social inconsistency. They will help government service delivery. Cut time. Help doctors with correct diagnostics. Help people.
National policy dashboards
Never doubt that a small group of thoughtful, committed citizens can change the world; indeed, it’s the only thing that ever has.
– Margaret Mead
How does a capable company run? You said it — profit-and-loss spreadsheets. They track every taka of investment. Where profit and loss happen comes from that sheet. Small companies are easy to track; big ones get harder. Hundreds of divisions. Separate books. Knowing that one unit has 2% profit while another has 80% loss on one sheet is genuinely hard. That is why CEOs of multi-million and multi-billion dollar companies keep a ‘dashboard’ — for the high-level view, for informed decisions. A broad perspective sits there; one click reveals the detail.
Take a successful telecom operator. The CEO has a high-level dashboard: company profit and loss, each department’s KPIs, how close or far the company is from the next target — everything lands there. Every tool needed for an informed decision is there. The dashboard also carries the parent company’s goals (for example Telenor’s philosophy). Every company runs on a philosophy; policies that catch that philosophy cascade inside. Everything inside is inter-related. One SIM sale in Bangladesh can show impact on the global dashboard. The pieces may be small, but they are strung on one thread. When government policy changes — say SIM tax rises or falls — the outcome shows there too. Mid-level managers have their own dashboards at their level, where they take informed decisions.
Would a government dashboard work here? Data has great power. It can reveal the inside of many things. Every event on earth connects to another; working with data teaches you that. From a teenage girl’s suicide driven by poverty to road-accident deaths, impact can surface on one dashboard. Even where VGF cards got stuck in field distribution shows up. We may not see the full inside story — village-level data points are already enough. A head of government could learn much of the inside story before events happen. Singapore’s “Smart Nation” or the UAE’s “Smart Gov” dashboards have reached another level; nearly their whole government service-delivery platform is online. We are getting there too. It can start with such a dashboard.
Can a government run this way, like a company? Yes. Many companies now run quasi-government-style Corporate Social Responsibility — duty to people; profit is not always first. As I said, from Singapore onward many countries feel data’s power in their hands. After a regulatory training in Singapore, my head opened: in those countries, data in policy-making saves money — from waste, from wrong decisions. Data helps timely decisions; one timely decision can save a lot of money. Otherwise project costs climb day by day.
What would such a policy data dashboard look like — for the Prime Minister, MPs and officials where applicable? Over 200 data points including ministries, fed from those places. Machine learning over correlated data can reveal people’s demands, where the real problems are, and each MP-area’s gaps or progress. Where work is stuck — what the problem is, the inside inconsistency — all of it lands on the dashboard. Decisions become informed: decided with knowledge. My data says India has launched data-driven governance dashboards in Andhra Pradesh, Odisha, Jharkhand and Maharashtra with help from a private organization.
I have spent nearly 30 years in government. Data use is growing in government, applied in pockets; an Open Data Initiative exists — but not enough. Living with data myself, I understand its weight. A policy dashboard for decision-making keeps turning in my head. My PhD proposal is “National Decision Support System for Bangladeshi Policymakers from Anonymous Large Scale Mobile Phone Data”. Machine learning is a large part. The topic works in three parts: (1) a National Decision Support System can be built, (2) for policymakers, (3) fed from one large data point. I limited it to a framework around mobile-phone data availability; with more data points it works at larger scale. All of it is possible — with artificial intelligence.
Small-scale intelligent machines first
Much of how we think has already been handed to machines. Whether an elevator keeps taking calls until a time, stops at our floor if calls stack above, or if overloaded above returns to external calls floor by floor — or with many elevators, which car by proximity and call load arrives at your floor soonest — is daily routine now.
Software is eating the world
People ask me why learn machine learning or deep learning. I say: this is the field changing the whole world. It is still exploring what conventional software cannot do. Around you, big companies are software companies — bKash, Pathao, Daraz, all of them. But machine learning and deep learning will eat conventional software companies.
“AI will eat software”
Search: “software is eating the world, AI will eat software”. Whatever challenge stands before you, this machine-learning exploration mode can help solve it — I see it around me. The exploration is largely intellectual: I keep meeting new problems whose only solution lives in machine learning.
We must grow our skillsets
Because it is a thinking machine, the intelligence is synthetic — hence our core topic is artificial intelligence. Teaching machines is still foggy ground. Hence these books, videos and conversations — as a series.
I believe one thing: if you build it, they will come. With skilled people, the economic shift arrives here. Abundant skilled people are needed to build intelligent systems. Software now needs intelligence; that needs willpower and brains.