The state of AI in 2018

Return to eHealthNews.nz home page

Picture: Ben Reid

NZHIT guest column by Ben Reid, executive director, AI Forum NZ

As 2018 draws to a close, I’ve been reflecting upon what a difference a year makes in the world of AI.

This time last year the concept of artificial intelligence was still largely in the domain of science fiction, with mainstream news articles dominated by Elon Musk’s latest (diversionary?!) pronouncement on general superintelligence and photos of the Terminator accompanying every sensational three-minute bulletin. Rational conversations free of hyperbole were hard to come by.

A year later, around the world and here in New Zealand, we are starting to better understand the very real applications – and implications – of AI across a broad variety of domains. We are seeing early signs of machine learning techniques (ML can be considered a subdiscipline of AI) being applied to vast datasets: in transport, financial services, logistics, environmental monitoring and marketing as well as in health.

AI for social benefit

We are also starting to see the emergence of strong social themes relating to AI. This week I am  attending Google’s AI for Social Good Asia Pacific Summit in Bangkok.

The opportunities to apply AI for beneficial social outcomes have been gaining significant headspace worldwide throughout 2018, and include the following:

• Google.org’s AI for social impact challenge, an open call to organisations around the world to submit their ideas to receive support from Google’s AI experts via Google.org grant funding from a $25 million pool. Applications close 22 January. 
• McKinsey’s recently published Applying AI for Social Good, which analyses around 160 AI social impact use cases.
• Microsoft’s announcement in September of a US$40 million, five-year programme, AI for Humanitarian Action.
• IBM talking to over 60 pioneers who are on the frontlines of applying cognitive technology for social good.
• These examples are well worth exploring further. They illustrate how AI technology is not just a zero-sum game for social media giants, but can also be applied to some of the most pressing global problems to provide solutions that weren’t possible before.

Technology advancement

During 2018 the technology has continued to advance rapidly and widely. AI-based image analysis – where neural networks are trained on huge image sets to detect objects of interest, whether they are cats, licence plates, faces or cancerous moles – has reached maturity, with machines now regularly beating human performance for common image-recognition benchmarks.

The software tools to do this are largely freely downloadable, open source and supported by commodity cloud providers such as AWS and Microsoft Azure, so basically anyone can get started using them.

These techniques are now being used to analyse daily satellite images of the entire Earth’s surface, picking out physical objects in a fraction of the time that was possible beforehand and, combined with other data such as weather, tides, traffic flows, create a ‘digital twin’ of the Earth’s surface – a real-time intelligent data model of the physical world.

Given enough time and investment, the same techniques promise the ability to create a digital twin of our own physical bodies, meaning that our vital signs could be monitored and modelled in near-real time, with more accurate predictions and automated anomaly detection.

Still some barriers

And yet there is still a long way to go. Earlier this year the AI Forum published our major research report, Artificial Intelligence: Shaping a Future New Zealand. In it we identified significant AI-driven opportunities for New Zealand.

However, we also called out clearly that some of the key enablers to seize the AI opportunity are not in place. In particular, the broad availability of large, trusted datasets and a strong enough pipeline of AI and data science talent who will be needed to implement AI solutions in future.

Fundamentally, data – structured, unstructured, petabytes of it – is what feeds AI. If we are not collecting this data at source and then making it available to machine learning applications, with appropriate trust controls, of course, then we are missing out on opportunities that others around the world are already sizing up.

AI in the health sector

AI adoption and uptake in New Zealand’s health system is still at a very early stage.

Local innovators, such as the Precision Driven Health partnership, are pushing the boundaries of what can be achieved using data-driven healthcare in New Zealand. Recently a consortium involving AWS, California Institute of Technology, NASA-JPL, Victoria University and the Ministry of Health has carried out promising research to improve early radiology breast cancer detection.

Anecdotally, in part progress is held back by siloed approaches to data stewardship in the health system, which make it difficult for researchers and developers to obtain enough data to create AI applications that identify patterns in health data and use this to make better predictions.

One solution is what is known as a ‘data trust’ – effectively a large unified ‘data lake’, but with common privacy and trust controls built in rather than fragmented through multiple siloes. Whether it will be a blockchain-type solution, which gives each individual data sovereignty over their personal health records, or a more conventional database with governance controls is still be to be worked out.

Meanwhile, in China tech giants such as Tencent, Alibaba and Baidu are all actively exploring the AI healthcare market at scale, with a very different regime for individual privacy than in the West – but arguably with accelerated health outcomes as a result. Indeed, the struggle for AI supremacy between China and Silicon Valley tech giants has been a major theme of 2018 – Kai Fu Lee’s accessible book AI Superpowers tells the story from the Chinese perspective and is well worth a read.

Looking ahead

Looking forward to 2019, I am optimistic about New Zealand’s capabilities to leverage AI technologies.

The AI Forum’s participation in the international Partnership on AI ensures that we are connected into the fast-moving frontier conversations on responsible, safe, fair and transparent AI deployment, and our rapidly growing New Zealand AI ecosystem clearly shows that there is increased investment into applying the technology.

Finally, I am delighted that the first AI in Healthcare conference Hack Aotearoa, to be hosted by The University of Auckland, has just been announced for January. I also invite you to join New Zealand and international AI thought leaders at our annual AI-DAY conference from 27–28 March.

Ben Reid is the executive director of the AI Forum New Zealand.