AI for traditional industries

GIC's annual thought leadership event, GIC Insights, was held here on 15 September 2017. It saw 110 prominent global business leaders deliberate long-term issues relevant to the international business and investment community. The theme was Asia's Evolving Role in an Uncertain World, and topics included Asia's Challenges and Prospects over the Next Decade, Artificial Intelligence for Traditional Industries and A Long-Term Future in an Uncertain World.

Below is the second of two excerpts from GIC on the topic AI for traditional industries:

Artificial intelligence (AI) is significantly changing and disrupting a wide range of industries, including industrials, agriculture, finance, and more. From self-driving trucks and autonomous factories, to Ebola cures and hunter-killer drones, AI can drive new solutions at a far lower cost than traditional companies like 3M and Boeing. New entrants have even beaten the initial wave of disruptors at their own game. AI will transform several markets into winner-take- all environments. Companies need AI to rise up to the challenge of AI, but they need to realise that understanding and adopting AI requires a fundamental rethink of their mindset and approach.

To understand why this is happening, we first need to understand how AI works. AI apes human reasoning, and can beat any set of human beings by puppeteering their moves to optimise its end-state. It is changing the paradigm by altering how current systems work. Existing systems are programmed by humans, working on a minimal subset of rules that can be applied and scaled for profit. But as tasks get more complex, more rules are required, driving costs higher. AI solves this problem, and the last barriers preventing AI from understanding the world we live in - vision, speech, natural language processing, etc. - are disappearing. With these capabilities and access to new datasets, AI can address complex, previously unsolvable problems. There are many compelling examples of these capabilities in the small, AI-focused companies operating today. For example the technology powering most satellites is remarkably limited. They are unable to image the surface of Earth in areas where it is cloudy or dark-- which is about three-quarters of the planet at any given time. Capella Space established only in 2016, has solved this by building tiny satellites that use synthetic aperture radar (SAR), a technology that can capture images in any light or weather condition. Capella's satellites are roughly the size of a backpack and a fraction of the weight of the few competitors in the SAR space. They cost less to build and are able to take photos much more frequently. SAR opens the door to many new applications, from tracking soil moisture to assess the health of crops to more-accurate mapping for self-driving cars. And it is not just traditional companies that are facing disruption by AI. Older companies who had once been viewed as disruptors in their own fields are now being outdone by younger competitors. Embark, a company started by a bunch of college drop outs, has developed a system of autonomous trucks at much lower cost and ahead of its more established competitors.

The good news is that AI provides a suite of capabilities that can be used to mitigate or anticipate upcoming disruptions, helping companies to manage risks, optimise their functions, and stay ahead of the curve. For example, there is Merlon Intelligence, the world's first AI enabled KYC /AML compliance solution. It largely automates the jobs of today's multi-thousand person compliance departments better and helps banks ensure that they do not transact with criminals and terrorists, cost-effectively.

Furthermore, large organisations can leverage their existing position and assets to their advantage, and use AI to reinvent their business models to capture large portions of the market. To do this, however, they need an enabler. There are companies like Element AI that serve this role by helping companies transit to AI, providing access to smart technology and new datasets that they cannot build themselves, thus enabling them to train robust AI capabilities.

Traditional companies need to change the way they work. They have a wide array of decisions to make, and those decisions require more data and examples to create an accurate simulation of the world around us. Only by doing so can AI help generate novel solutions to the business problems we face. This means, however, that AI cannot be "wrapped" around existing systems and data. The data that organisations have today is a mirror of their past operations, including previous biases. They need new datasets to replace systems trained on assumptions that are no longer accurate. Companies should invest in simulations and "games" that may not have immediate relevance to revenue, but are useful for gathering information on consumers and their lifestyle preferences, to train AI engines to monetise that information for their business model. Doing this, however, requires a fundamental rethink of the company's role. AI is a formidable capability, but it cannot learn on its own; it requires human guidance. Organisations must transit from a command-and- control role to teaching AI systems, improving and iterating on the algorithms to ensure the whole system creates value. And to truly capitalise on AI's potential, organisations will need to look past their old corporate partners, and access Silicon Valley founders and talents that may not necessarily move in the same circles. Many of these talents will be in Silicon Valley, but there are several hubs that are home to innovative companies that are linked up with universities and labs, including Montreal, Beijing, Tokyo, and Singapore.