Last week I got to spend 3 days at CogX, the world’s biggest “Festival of Artificial Intelligence and Emerging Technology” in Kings Cross, London. The event is billed as the “Glastonbury of Tech” with over 10 stages hosting a full programme of speaker presentations and discussions together with a large Trade Expo full of 150 exhibiting companies. The venues were spread throughout a regenerating Kings Cross in the centre of London – a completely transformed urban landscape with new tech businesses and high rise offices springing up from the ground everywhere, including Google’s huge new £1 billion,1 million square foot UK “Landscraper” HQ.
CogX is the largest gathering of the global AI community, with an estimated 15,000 attendees and 12 topics including healthcare, infrastructure, finance and economy. (Check out the CogX website and also #CogX19 hashtag for videos and tweets over the three days).
AI to Accelerate the UN Sustainable Development Goals
This year’s event was run in partnership with 2030Vision – a new initiative that aims to transform the use of technology through collaborative partnerships and innovative projects in order to support the delivery of the United Nations Sustainable Development Goals (SDGs – 17 Global Goals) and unlock the commercial opportunities they offer.
This key message that AI is a fundamental new tool to help bring about solutions to the world’s major development challenges permeated all of the conversations during the conference. It is much more helpful to have a tangible framework to aim for than the more nebulous “AI For Good” which has been doing the rounds in the last couple of years. As economist Jeffrey Sachs said, the UN SDGs are the only universally agreed objectives we will have in this generation. Time to achieve them, not reinvent them.
The conference lineup was incredibly strong overall with a “who’s who” of global AI speakers presenting: Eminent Berkeley AI professor Stuart Russell, economists Erik Brynjolfsson, Mariana Mazzucato, Benedikt Frey, technology commentator Azeem Azhar, Google X CEO Astro Teller, Bing Xu (co-founder of Chinese AI Unicorn SenseTime) together with Dr Mark Sagar, CEO of New Zealand’s own Soul Machines. As one commentator mentioned, there isn’t another tech event featuring some of the world’s leading economists, top AI researchers and technologists, businesses and ethicists all gathered together – not to mention inspiring the next generation with initiatives such as Teens In AI.
The topics under discussion were equally broad: from hands on demonstrations of the latest machine learning research, conversational tech and human-machine interfaces to discussions on the application of AI to health, financial services, agriculture and national security. Presentations on AI investment strategy and venture capital success stories mixed with discussions on AI and the future of work. Panels covered the increasing sophistication of “Data Trusts” to manage complex data rights issues and there was a significant overlay throughout on AI regulation, safety and ethics.
Here are a few of my highlights from the three days:
Oxford University economist Benedikt Frey – The Forthcoming Economic Singularity. Frey is the co-author of the seminal 2013 paper on automation: The Future of Employment: How susceptible are jobs to computerisation? which – although not uncontroversial – has influenced much of the modern debate on whether robots will take our jobs or not. Debate still rages on whether human intelligence will still have a place to play in a future AI economy – and if not, what should we do about it? (Eg how will the benefits be shared? Should we just have more time off…?). He previewed his forthcoming book The Technology Trap: Capital, Labor, and Power in the Age of Automation which examines how the history of technological revolutions can help us better understand economic and political polarization in the age of automation.
MIT Sloan Economist Erik Brynjolfsson – Productivity from AI. Many applications of machine learning have already reached human or superhuman levels – but Prof Brynjolfsson’s recent research uncovers that while no current occupations can be completely automated by ML, nearly all have some exposure. The implication is that *redesign* of jobs will be key to economies reaping the productivity gains from ML. And this implies a productivity “J-Curve” – productivity improvements will be underestimated during investment phase and the gains only harvested later. However, he sounded a note of warning: “Only those countries which seize the ML opportunity will reap the long term rewards…but currently most countries are focused on protecting the status quo.”
Berkeley AI professor Stuart Russell presenting his latest research on beneficial AI – taking a mathematical approach to “proving” how humans and AI can cooperate beneficially together. He expressed a little bemusement at the presence of an “Ethics” stream at an AI conference – asking whether if it was an engineering conference would that imply that there are engineers building “good” bridges and “bad” bridges? My takeaway is that he sees the question as much more one which can be solved by mathematics and game theory.
Thomas Reardon of Ctrl Labs – Human-Machine Interfaces: The fundamental challenge with human cognition is the limitations of outbound bandwidth – currently a maximum of 100 bits/second no matter how fast you can speak or type – this new technology promises Kb/sec – which would imply a huge increase in personal productivity if it comes about. He demonstrated the amazing advances in myocontrol (muscle-based) human-computer interfaces that CtrlLabs are building, promising increasingly sophisticated “telepathic” control of devices via non-invasive techniques. (Incidentally Thomas started out his career on the Microsoft Internet Explorer team and is featured in the very funny new docuseries Valley of the Boom about the late-90s Dotcom era which I enjoyed on the long journey home!)
MIT – Cesar Hidalgo – his team at the MIT Collective Learning Group have been working to crunch the numbers and data on “economic complexity” across the world – and have published work that economic complexity is correlated with lower levels of inequality. They have analysed an amazing amount of data and made this available accessibly online – check out these visualizations at The Observatory Of Economic Complexity.
Finally the panel discussion Governance and Accountability in an AI World between AI ethicists De Kai (MIT) and Joanna Bryson (Bath University) and facilitated by technology futurist Azeem Azhar was a thoughtful, intelligent discussion touching on how concepts such as human rights and privacy are changing rapidly as technology evolves. De Kai also made the very relevant point that current AI Ethics conversations are dominated by Western values – for example the recent IEEE Ethics in Action standards development effort only begins to consider how non-western traditions such as Buddhism, Confucianism can be reconciled together in the design of automated systems.
A note on UK Government AI Investment
CogX was held alongside London Techweek – a week-long series of events focussed on promoting technology sector in the UK – very similar to New Zealand Techweek which was held here in May. London Techweek is supported by the UK government who – despite the constant political diversion of Brexit over the last three years, have been quick to acknowledge the importance of AI technologies to Britain’s future prosperity and invest in strategic national AI institutions. This includes the AI Sector Deal headed by a National Office for Artificial Intelligence and a large research endowment to grow the national Alan Turing Institute. During CogX there were a number of new AI policy announcements including:
- £100 million for 1,000 new AI PhDs
- A new prestigious fellowship scheme for top AI researchers.
- Creating up to 2,500 places in AI and data masters conversion courses around the country. These courses will help people who have originally trained in other degree disciplines to gain the skills to contribute to the ongoing AI revolution.
- As part of this, funded up to 1,000 scholarships to ensure these opportunities are open to everyone, no matter what their background.
In particular the AI Sector Deal forms a core pillar of the UK’s industrial strategy and shows a number of key policy levers which have the potential to “tip the scales” in the UK’s favour.
After the whirlwind of attending back to back mind-expanding conference sessions for three days I had plenty of time on the long plane ride home to reflect on what this all means for us back in New Zealand.
I have distilled my four key takeaways from CogX:
- CogX contained the cutting edge of international conversations on AI (at least in the West): I came away with the knowledge that New Zealand’s relatively small AI community is actually well progressed in many of these conversations – in particular we have internationally strong thinking in Open Data, AI Ethics, Synthetic Media and Deepfakes, the Future of Work and AI regulation. However we fall down in our capabilities for AI investment – both private (Venture Capital) and public (strategic Government investment), together with talent development and attraction.
- There is a deepening international AI talent crunch and New Zealand needs to intervene quickly and credibly to attract and retain AI researchers and developers here – and right now we don’t really have a plan to do this. Countries like the UK are showing the way for how to do this.
- The emergence of “Data Trusts” as a working model for resolving complex data rights issues appears to have critical mass and should be actively explored here in New Zealand.
- Right now there are two “AI hemispheres” – “The West” and China. And what was clear from CogX is that the rules and ethical values underpinning the Chinese AI sector – particularly around individual privacy and human rights – are directly inconsistent with those held in the West. However this “ethical unencumbrance” is arguably enabling Chinese AI innovators to progress their research faster and deploy AI innovation at much larger scale. In our outward-facing multilateral tradition, New Zealand has an opportunity to play a key role by facilitating dialogue among these two traditions and establishing the common international ground.
Attending CogX was incredibly valuable to inform the AI Forum’s ongoing programme – in particular our next research report series – Towards Our Intelligent Future – which is due for release near the end of July. The first reports will present an updated landscape view of AI adoption in New Zealand and a strategic policy framework to address the opportunities – and impacts – of AI on our society, economy and overall wellbeing. The global context provided by CogX helps to ensure that our thinking about AI in New Zealand works on an international level, not just domestically.