- The World Economic Forum’s “Annual Meeting of the New Champions” was held last week in Tianjin, China, bringing together scientists, entrepreneurs, and global industry executives.
- Artificial intelligence, machine learning, manufacturing automation and recognition technologies are far beyond conceptual ideas: they are increasingly well-developed and in widespread use across a range of industries.
- Recognition technology is advancing rapidly and has implications for not just surveillance but also retail and product recommendations across industries.
- There are economic benefits, according to some modelling, but also a lot of risks.
Thursday evening was spent near the banks of the River Hai at a bad German restaurant in the Italian Style Quarter of Tianjin. We shared banana splits and there was a conversation about memetics among the group, which included a direct descendant of John Jacob Astor and a Kiwi physicist whose company sells AI-generated reports to investment banks and intelligence agencies. Yes, the big ones.
Tianjin – a feeder city for Beijing with a population of 15 million – was the location for the WEF’s Annual Meeting of the New Champions, also known as “Summer Davos”. As one person explained it, Davos is for those who sit over the top and provide the capital and leadership, while this event is for the doers.
There wasn’t a clear fulcrum of conversation for the event as there was in Davos back in January, when the investment and corporate worlds had their minds focused by a warning from BlackRock CEO Larry Fink about the need for companies to have a “social purpose” if they were to secure part of the $US6.2 trillion in funds he manages.
Escalating economic tensions between the US and China were naturally on the agenda, with Premier Li Keqiang delivering an official opening speech on Wednesday. Somewhat bizarrely he didn’t mention the tariff exchanges, but did promise China wouldn’t allow the yuan to continue depreciating, a declaration that moved global currency markets.
He also promised to ensure foreign companies have a level playing field in China – an ongoing bugbear for firms setting up in the world’s second-largest economy – and directly invited people to register complaints about unfair treatment with the government, while threatening severe consequences for government officials found to be causing trouble for overseas firms.
And while he didn’t talk directly about tariffs, he did say China would continue to reduce taxes and fees for business. “This is not empty talk… we will make sure these are materialised with concrete actions,” he said. The underlying message: while the US goes down the protectionist route, China will continue to open up.
One striking feature of the three-day event was the advanced nature of the discussion around artificial intelligence and machine learning. The conversation has moved a long way beyond the conceptualisation of the technologies and the question of whether they may be of benefit to industry or have the potential to disrupt. It is deep into its applications into every industry sector and business model.
Clearly, there’s a rapidly expanding class of established companies and well-funded startups that are far advanced in their use of these new technologies to solve existing problems and build new businesses. Events like this where the C-suite executives and entrepreneurs involved are all under the one roof underline the scale of progress.
It’s hard to escape the sense that companies not actively working to automate process and find AI applications that support the evolution of their businesses are going to be overwhelmed by this exponentially expanding cohort of leaner, smarter, globally-focussed companies that are already well advanced on the path to capitalising on what the World Economic Forum calls the Fourth Industrial Revolution. This is the idea that automation and the connection of vast, newly created datasets being analysed and processed by machines will create new businesses, destroy some existing ones, and reshape how people live.
The technology adoption cycle, it seems, is becoming shorter. At least to the delegates, AI and machine learning are no longer shiny new things. Perhaps this is because, nowadays, when some new digital idea comes along – think about Web 2.0, blockchain, the idea of the Internet of Things — we’ve all seen the movie before. The technology becomes rapidly integrated into industry practice – again, consider how IoT is now meaningfully integrated in the transport sector providing real-time updates on public transport networks – instead of there being a debate about “whether this is the right thing to do”.
China is certainly building a huge amount of capacity in artificial intelligence, not least in visual recognition.
One CTO from a multinational food company recounted visiting a startup in Shenzhen China’s technology hub on the mainland not far from Hong Kong. The product they were demonstrating was for textile recognition: you could take a picture of a random fabric and the platform could identify it and then provide suggestions for getting a suit or garment made from that particular material.
Cute idea. But there was much more to this company.
The firm, the executive recounted, had been given the photos from the identity cards of all the residents of Shenzhen. They are also given access to every photo taken at every street corner by the city’s surveillance cameras each day.
You don’t need much imagination to come to grips with the nature of their other core business. This company that does clever things with figuring out how to find you a suit made from a fabric that takes your fancy is also engaged in mass public surveillance.
(On the streets of Tianjin, there are also heavy-duty, highly visible cameras on every corner.)
The widespread use of this type of recognition engine will of course have huge implications for society and the boundaries of privacy. But the horizontal transfer from surveillance into textiles is a powerful example of how a technological development in one area creates opportunities for disruption or improvement in another.
Innovation has contempt for sectoral siloes. Today’s Airtasker is tomorrow’s reinvention of capital expenditure models for agriculture.
(I did speak to someone who sells farm machinery and they are starting to see the “renting” of tractors and other equipment in China between farmers through digital platforms. Not every farmer needs a tractor now.)
The recognition revolution
Various types of recognition technology – from visual to speech to asset prices — were being showcased and discussed at the forum. I interviewed Jason Crain, Amazon’s entrepreneur-in-residence, who sold his recognition startup, Partpic, to Jeff Bezos’s company and now works there helping build the capability for the e-commerce giant.
Partpic allowed people to take photos of hardware parts and order them from major US hardware chains. It involved building a platform that could distinguish between different nails and screws and bolts from each other, just from a photo taken on a smartphone. They started with the small items like fasteners because as Crain explained, the AI recognises things a bit like the human eye and brain – it takes effort to look at small things and recognise them, but gets easier as you work up into larger items like machine parts.
Crain, who used to work at the global leader in music recognition, Shazam, revealed Amazon has iterated on his technology and is using it to help catalogue the tens of millions of items that it sells in a visual database. There is a future not far away where you will be able to point a camera at almost anything and order it from Amazon.
The idea for Partpic came from Crain’s co-founder who worked in customer service at a hardware retailer. Their call centres were consistently struggling to help customers identify broken parts and precise hardware requirements. People would call saying a widget had broken; it was simply impossible to identify the exact part over the phone, and meanwhile their machine had broken down and they were losing thousands of dollars.
Often, after an agonising process of trying to describe the faulty part, the hardware company would ship out an item that would turn out to be not what was required anyway.
Partpic solved this problem for hardware stores. But Amazon, naturally, has a bigger idea in mind: the Shazam for everyday objects.
How many jobs do you think there are in the world where recognising what a customer is after is a skilled role that needs knowledge and understanding of a large inventory of supplies?
This is work and expertise that is currently in not just at the end of the phone in customer support call centres, but in front-line retail jobs all over the world.
Automated product recognition and recommendation is going to entirely replace that particular knowledge element of those jobs. Playing the tape forward, it extends into product recommendation in all sorts of other categories such as finance or agricultural machinery when there is sufficiently rich data on a customer’s needs.
This of course will increasingly be the case as various aspects of industry performance and our own individual consumption preferences are captured by platforms in the data age.
How public policy responds to the potential job disruption from automation was also a topic of discussion. The World Economic Forum works hard to ensure questions around social impact and inequality are woven into the conversation about innovation and technological advances.
One of the more confronting scenarios here is that proposed by Oxford professor of globalisation Ian Goldin, whom I interviewed last year and was speaking at a range of events in Tianjin. He argues that emerging economies which have benefited from decades of manufacturing offshoring by advanced nations – lifting billions out of poverty in the process – are potentially at risk of a catastrophic collapse in demand for their services as robots replace the millions upon millions of jobs that have been created in countries like China and India.
But there are more upbeat appraisals.
PwC’s UK chief economist John Hawksworth has built a global macroeconomic model for assessing the impacts of automation. He admitted being surprised by the results it has been returning, including a base case that new industry has the potential to add more than 20 per cent to China’s GDP over the coming two decades, if the policy settings are right in areas like competition policy, education and healthcare (and as long as wages don’t rise so fast that they increase the incentives for companies to automate more and more work).
Hawksworth’s basic premise is that while automation will dislocate some existing jobs, China “will become the world leader in manufacturing all the robots, and the drones, and driverless vehicles. While that process will also be automated to a degree, there will still be some jobs,” he said.
But that’s just the start. The necessity of retraining people whose jobs are automated will increase demand for “human touch” jobs in education through the need for more teachers and teachers’ assistants, and in the aged care sector to cater for China’s rapidly changing population.
Hawksworth also has an interesting perspective on the idea of a universal basic income, or UBI, the policy tool that has seen increasing attention. He argues a UBI in China is impractical because it would simply be unaffordable to make it not just truly universal but also meaningful to 1.4 billion people.
Instead, he sees potential for a special treatment of those “human touch” jobs that improve social outcomes and, over the medium-term, improve cohesion, productivity, and a society’s economic potential.
It all makes any time spent thinking about the self-indulgent tendency of political classes to become obsessed with their own issues doubly despairing.
The scale and pace of change going on around the world in technology and industry are enormous, with major implications for investment flows, regulation, and society because of the likelihood of job dislocations and changing skills requirements.
This presents major challenges for public policy and it’s impossible not to notice how far removed political leaders are from driving a real conversation about how they are going to manage it.
Business Insider Emails & Alerts
Site highlights each day to your inbox.