Singapore businesses not using AI either fear it or don't get it

Singapore

BUSINESSES in Singapore are not using artificial intelligence (AI) enough and that's only because they fear it or don't get it, The Business Times (BT) has learnt.

Tan Poh Choo, operations director at SAS Singapore, a business analytics services company, said: "While we are seeing a growing level of AI adoption, many businesses tend to have a common misconception that intelligent machines replace human activity. There is still a lack of understanding around how AI operates, and more importantly, how it is used."

For instance, less than a third (24 per cent) of sales teams here use AI, going by the third State of Sales report by Salesforce, a San Francisco-based customer relationship management firm.

The figure is lower than that of Germany and the Netherlands (33 per cent), Australia and New Zealand (31 per cent), and the UK (26 per cent) - but higher than that of the US (23 per cent), Hong Kong (13 per cent) and Japan (11 per cent).

Most sales teams here expect AI to "gain ground quickly for a variety of use cases", the report found. Sales leaders also predict that AI adoption will grow by 216 per cent by 2020.

"Despite industry rumblings that AI will displace sales jobs, those using AI continue to expand their teams," said the report, which polled over 2,900 sales professionals worldwide, including Singapore.

Kate Darling, research specialist at MIT Media Lab, said at Pure Accelerate 2018 that AI is going to shape future societies, but "we don't know how yet".

Pure Accelerate, held in San Francisco in May, is an annual conference by US data storage firm Pure Storage, which counts Singapore chipmaker Avago and provider of sustainable urban solutions Ascendas-Singbridge as customers.

Dr Darling, who specialises in robot ethics, said: "In particular, the media is worried that AI might somehow evolve to become smarter than us and take over the world and kill us all. But I'm worried when the media picks up that one thing, because it distracts from the actual problems we need to be concerned about, such as privacy and data security."

Richard Socher, chief scientist at Salesforce, added that such fears also distract from the real, immediate challenges posed by AI, such as biases in data sets that are used to train AI.

Dr Socher told BT in an interview: "Take a supermarket chain that has mostly promoted men to lead the company. If we look at human resource records to make promotions automatically using AI, we might make that kind of bias and miss out on great females.

"In another example, large companies have wrongly classified black people as gorillas. The companies are not malicious - they are just not careful enough or do not have well-distributed population sets. It's not because they program it like that - it's just that their training data is not much."

Roy Kim, director of product marketing for FlashBlade (an AI hardware by Pure), told BT on the sidelines of Pure Accelerate that "companies with the most data are most valuable".

But only 0.5 per cent of data is analysed today; the rest is stored somewhere without being analysed, he noted.

"The problem is not a lack of data per se. Companies need to change to be more 'data centric' and build a business around data, then the equation will flip. But on a high level, companies don't really know what data or AI is good or not good for - and that's where the fear of AI feeds on."

SAS Singapore's Ms Tan said that the core purpose of AI is to play the role of an enabler: to surface relevant information required to solve problems and emulate human tasks and decisions through learning.

"The key takeaway here is that AI machines cannot operate meaningfully without being fed with context from humans," she said.

"Only humans understand what problems need to be solved, while machines help us do it in a more efficient way. Any form of intelligence, including AI algorithms, can be leveraged for good to drive positive impact."

She added that this, however, works both ways. "We have already seen examples of automated decisions causing damaging consequences, increasing the importance of governance and ethics in the use of AI. If an AI machine is making inaccurate or unethical decisions, there is a high possibility that there was a lack of ethical data to train machines."

Singapore recently announced that it is setting up a council on the ethical use of AI and data and a five-year research programme on the governance of AI and data use.

The Singapore Management University will helm the programme, for which it has been awarded a research grant of S$4.5 million from the government.

Rizwi Wun, a partner at RHTLaw Taylor Wessing, told BT that ethics play an important role to remind decision makers to "appreciate and consider" the local context in which AI is employed.

"Firms seriously need to ensure these solutions do not come with unintended costs and consequences that far outweigh the benefits. This is where ethical considerations can play an important role. Ethics with a dose of pragmatic reality guides us to do the right thing, and not simply do things right."