ABCD: not as easy as you might think

TL:DR: ‘ABCD’—artificial intelligence, blockchain, cloud and data—could be considered the alphabet of the future. But amidst the excitement surrounding these technologies, it’s also important to acknowledge that they aren’t a silver bullet. Ultimately, for technology to make an impact, it must be designed and implemented with care, said panellists at GovTech’s STACK 2018 Developer Conference.

Along with learning how to spell, children in classrooms of the future might have to contend with another category of ‘ABCD’—artificial intelligence (AI), blockchain, cloud and data. But even as organisations the world over rush to integrate these technologies into their business models, getting the most out of them is nowhere near as easy as reciting the English alphabet. “That’s the challenge with ABCD—there’s a lot of hype, and it’s very often sold as a silver bullet. But in the end, the devil is in the details and in trying to iron out the actual production issues,” said Mr Parimal Aswani, director of government digital services at the Government Technology Agency of Singapore (GovTech), who was moderating the ‘Building Apps For Tomorrow: The Future with ABCD’ panel discussion at GovTech’s inaugural STACK 2018 Developer Conference.

Design dilemmas

One major decision companies face is whether to develop AI applications using off-the-shelf application programming interfaces (APIs) for machine learning offered by the likes of Google and Facebook, or to build their own customised neural networks. Each option has its own pros and cons, said Ms Annabelle Kwok, founder of vision analytics company NeuralBay.

Off-the-shelf APIs are cheaper and easier to prototype with, but are often too generic to fulfil specialised requirements; custom networks, on the other hand, are suitable for secure applications and can be deployed according to one’s needs, but are complex to develop and maintain.

“There are many differences, but I would really recommend that you consider building your own customised neural network, because it’s worth the investment of time and money, especially when you think about the security implications,” said Ms Kwok, adding that companies should think twice before entrusting their business-critical data and analytics to a third party.

Alongside AI, companies are also turning to cloud-native approaches to develop applications faster and more nimbly. Here, design is also paramount, said Mr Kief Morris, head of cloud transformative practice at tech consultancy ThoughtWorks.

“The whole idea of cloud is it enables you to move quickly to come up with a minimum viable product, and to make small changes very rapidly,” said Mr Morris. “It’s not enough just to adopt the technology; you also have to think about the architecture of your applications—a monolithic application is going to be very difficult to even run on the cloud, much less allow you to take advantage of being able to move quickly.”

Focus on the fundamentals

Of the ABCD technologies, blockchain—strongly associated with cryptocurrencies—is perhaps the most hyped. “The biggest misconception is that [blockchain] is all about money, tokens and the current price of Bitcoin. But what I think is most interesting are the other applications being built around distributed ledgers, whether it’s smart contracts or alternative blockchains,” said Mr Hunter Nield, co-founder and CTO of cloud technology company Acaleph and an alumnus of GovTech’s Smart Nation Fellowship Programme, adding that digital identity is, for him, one of the most important blockchain applications.

Energy is another, said Mr Chang Sau Sheong, managing director, digital technology at SP Group, who outlined his organisation’s efforts to create a marketplace where homeowners and small players can trade renewable energy certificates. “Energy marketplaces today only cater to big buyers and big sellers—for example, corporates buying from large solar farms. What we’ve done is really to extend and democratise the entire process using blockchain.”

In today’s climate of dizzyingly fast technological change, it’s important not to get swept up in hype cycles; ultimately, a strong grounding in the basics is what matters the most, agreed all four panellists.

“Spend a lot of time on the fundamentals—the basic building blocks and concepts in computer science,” Mr Chang summed up. “Don’t focus too much on new-fangled technology frameworks, because they sometimes go out of fashion very quickly… but computer science—our roots—won’t go out of fashion.”