🔮 The real AI threat; globalization enters a new phase; universal basic income; the state of the internet; causality, Fortnite, culture wars++ #168
|Azeem Azhar||Jun 3, 2018|
Dept of the near future
📧 Mary Meeker’s compelling annual report on the state of the Internet is out. Two messages stand out: growth rates for mobile and internet are surprisingly tailing off, but still plenty of room for innovation. China is on fire: user growth, payments, m-commerce & the scale of its largest firms.
🌐 Globalization is not in retreat, despite certain shifts in the post-war consensus, argue Laura Tyson and Susan Lund. Rather it is evolving with three distinct characteristics: the rise of China; the relative importance of cities; and the growth of new cross-border digital flows. THOUGHT PROVOKING essay, with a set of themes worth tracking.
🔮 The real threat from AI: A MUST READ essay by EV reader, John Havens. “The greatest threat that humanity faces from artificial intelligence is not killer robots, but rather, our lack of willingness to analyze, name, and live to the values we want society to have today.” (See also, an excellent essay by Josh Dover on the ethics of addictive technology: “technologies wield unprecedented scale and speed of societal change. It's more important than ever to be intentional and conscious of the potential effects of what we build.” Apple may be responding with new tools to manage iPhone usage.)
🎢 Wait, is this the real threat from AI? “We live in a world where ‘digital’ is no longer a glorified marketing department in a company or an economic sector on its own, but a layer over everything”, says Angus Hervey. Move slowly, and don’t break things.
💣 Google’s Pentagon contract has provided more evidence for the emerging overlap between politics and technology. More than 4,000 employees signed a petition complaining, a dozen resigned. Google’s leadership initially felt “that it was better for peace if the world’s militaries were intertwined with international organizations like Google rather than working solely with nationalistic defence contractors.” Google might have subsequently decided not to renew the contract, which was forecast to be worth $250m a year. Presumably, this means that rebellious AI engineers are worth even more than that. (Amazon may not be so squeamish.)
👌 What social media can do for democracy: Ethan Zuckerman suggests a handful of real benefits of social media (including inclusion; mobilisation; deliberation and more). Interesting read, countervailing the current mood.
💸 A good survey of current Universal Basic Income pilots. The evidence is mixed and ultimately may not provide a canonical recommendation: the findings may not scale up to the size of society.
🚀🚀** Viability #3**
Another huge thank you to those of you who recommended Exponential View to your peers last week. I’m asking you to support us tripling the size of the community.
The results of the last fortnight have been meaningful. We added about 1,000 more readers than usual (or about 2% towards our target of tripling readership).
Our models suggest about 4% of our readers have supported us via LinkedIn, Twitter or email. Thank you!
That leaves about 95% of our readers. Please do take a few seconds to help out.
The best way you can do this is to forward this email to seven or eight friends/peers/contacts who you think would enjoy EV. Include a personal recommendation to subscribe and point them to the link → http://www.exvw.co.
Dept of artificial intelligence
The South Korean government will invest about $2bn (or 2.2 trillion won) in national AI efforts. Investments include human capital (targeting close to 6,000 researchers), strengthening its domestic chip industry; and a focus on applications in drug development. (This supports my number 1 prediction for this year - that national AI strategies would emerge rapidly.)
I’m curious about what this means. On a per capita and per GDP level, this is a larger commitment that the UK and France. Although it’s not clear if it as ambitious as China’s plans to build a $150bn annual AI industry by 2030.
South Korea does have some interesting features:
South Korea has by far the highest penetration of industrial robots in the world, at 631 robots per 10,000 employees (more than treble the US rate and eight times that of Europe), and in recent years the rate of robotisation has even accelerated. This might suggest corporates are used to the disciplines of automation--and primed for more advanced forms. AI is a key enabler of what some are calling industry 4.0, the reimagining of manufacturing in an era of sensors and smart machines, something which could be significant in manufacturing-heavy South Korea. (For a quick intro to Industry 4.0, read this guide by EV reader, Robin Dechant.)
The country is the leading producer of semiconductors, mostly DRAM and NAND flash markets globally. While suffering epically from boom-bust cycles, I’m curious as to whether a renewed focus on AI might unlock some of the talent in these industries and create new innovation in AI-oriented silicon. (If you have an informed view on what this means for chip innovation, I’d love to hear it.)
South Korean giant, Samsung, has opened a range of AI research centres in Canada, Russia and the UK. I understand the UK research centre, based in Cambridge, will focus on healthtech AI, and is in close proximity to Apple and Microsoft’s healthtech AI research groups.
Filip Piekniewski reckons there might be an AI winter coming. My take: we’re making the easy gains right now, and many sub-technologies are only now entering into mainstream use. It will take many years for every business to replace their brittle, deterministic planning with stochastic models. Even as we approach limits on current deep neural networks, we’ll see researchers make progress with other methods, particularly those which rely more heavily on prior knowledge or explicit reasoning. That said, it’s hard to believe we are even close to running out of things to do with deep learning.
The investors in China’s SenseTime probably feel deep learning has a way to run. They just pumped another $620m into the private firm which focuses on using deep neural nets for image recognition. (Or said investors are just chasing the crowd, you decide.)
Warning: nerdy. ARM, the semiconductor firm, outlined its machine learning processors. Seemed for a while that ARM had been slow out the gate with supporting the demands of modern machine learning (basically support for matrix operations and more efficient data movement), yielding to NVidia and a slew of new competitors. I suspect they aren’t too late and will be able to play to the advantage of their installed base on mobile devices and the inherent benefits of system level integration across CPU & machine learning processors.
Young Ethiopians turn to AI as the core of future development. (Fascinating take.)
One to watch: Lukas Biewald, a veteran of AI companies, Powerset (where I as an investor) & Crowdflower has launched a new enterprise AI platform, Weights and Measures: "the biggest pain is a lack of basic software and best practices to manage a completely new style of coding." This aligns with trend #4 of my 2018 predictions that we'll see a new paradigm of software development emerge.
Short morsels to appear smart at dinner parties
Fortnite, the bane of many parents, made $300m in April.
👛 Coins.Ph, a Philipines-centric cryptocurrency wallet hits 5m users. (Is this a case of "the future is here, but it just isn't evenly distributed"?)
China overtook the US in healthy life expectancy at birth for the first time.
Shenzhen, China’s tech centre, is overheating. “Incubators are everywhere.”
A vegan diet is the best way to reduce your environmental impact, says new research from the University of Oxford.
🌵 Debate: are plants conscious?
The first Eurasian migrants into North America came by boat.
Transgender brains are more like their desired gender from an early age.
🔥 Another take on the culture wars. EV reader, Shehnaz Suterwalla, on the movements to decolonise culture.
Azeem's end note
I was thinking about national AI strategies. They are, mostly, seemingly a bit ho-hum. When John Kennedy sought to motivate a nation he said: “We choose to go to the Moon in this decade and do the other things, not because they are easy, but because they are hard; because that goal will serve to organize and measure the best of our energies and skills.”
Today’s national AI strategies don’t point to an epic transformational goal, they speak about the number of PhDs that are going to be created, they speak to the how, the process, not the summit we'll conquer.
If I was trying to motivate a group (any group), I’d want to give them a grand goal that is tied to a meaningful purpose. Because motivation is an emotion, not a SWOT analysis.
I can rationally see the benefits of more researchers rather than fewer. But I can’t really get excited about whether it is 1,200 or 1,500 or 2,000.
Perhaps it’s a good thing that these plans seem mildly detumescent. An excitable public policy might lead to monorails. Or perhaps government doesn’t really trust itself enough. Is it still stinging from the Gipper’s rebuke?
But as such brilliant minds as venture capitalist Bill Janeway and economist Mariana Mazzucato have both argued recently the public sector has been a key principal agent (yes, principal, not merely catalyst of enabler) in the innovations upon which our modern life is built, and in Mariana’s words was “the investor of first resort”.
Should boldness rather than temerity be a hallmark of these strategies?
As a reminder, if your firm wants to connect with the fantastic Exponential View readership and sponsor the newsletter, you should drop Marija an email. We have consistently maintained open rates of >50%. Last week our open rate surpassed 60%, every one of them an incredible person... like you.
Hasta la vista,