🔮 AI Nationalism; privacy; the limits of neoliberal techno-capitalism; traffic lights, Proxima Centauri & Airpods++ #170
|Jun 17, 2018|
Dept of the near future
💯 “The rapid progress in machine learning will drive the emergence of a new kind of geopolitics… AI Nationalism argues EV reader, Ian Hogarth.
🤖 Chinese retailer, JD.com, opens a warehouse which can handle 200,000 orders a week. All robotics, just four human workers to maintain the robots. (Video.) Is this the future of logistics? (See also, this review of investment bank plans to automate & automation comes to Amazon’s white-collar workers)
🙈 Why the debate about privacy is so confused. "The real issue is liberty", argues Louis Menand in this stunning long read. (See also, EV reader, Chris Yiu, explains how ads got so good at following us around.)
🍎 Ian Bogost: Apple’s Airpods presage changes to our social fabric.
🔎 Why the future of deep learning is tiny. Pete Warden’s manifesto for moving deep learnings to the hundreds of billions of low-power microcontrollers that are pervasive in the world. (See also a video of Pete discussing this below.)
🎭 Pankaj Mishra: The mask it wears. “That curious global conjuncture in which neoliberal capitalism and technological leaps forward guaranteed endless progress, and a tiny elite passed off its interests as universal norms, has passed.” What might replace it?
👍🏽 Viability #5—it works!
I’ve been asking you to help to triple the size of the Exponential View community. A bigger community means better viability - and giving you more of our broad, interdisciplinary insight.
We reckon between 8 and 10% of our readers have taken the time to support us during the past month. Thank you. Here is an excellent example from Sven Meier of German energy giant, EnBw.
And the result of these efforts by readers? Next week’s Exponential View has a piece of proprietary research I’ve commissioned into how people understand Facebook’s use of personal data.
More goodies desired? If you haven’t taken a moment to recommend Exponential View in the past month, please find a couple of minutes. Three effective tactics (do one, two or all three):
Share the URL www.exvw.co with a group of your friends at work, a community of practice that is relevant, a book club, etc with a recommendation.
Write a review on your blog, LinkedIn or in your column.
Tweet a recommendation. You could try this one or the standard one or one of your own.
Dept of artificial intelligence
This week I attended two AI conferences. The remarkable CogX festival in London (7,500+ attendees) where we had run “The Cutting Edge” track. At the other end of the scale, was the intimate and engaging, Transform AI, in Paris. I can recommend both for 2019.
Here are five highlights (all videos) from our stream. Some may be slightly technical, but all are worth watching, even for the non-technical reader.
Stan Boland, the CEO of Five.AI, on how to deliver safe urban autonomous vehicles to European cities.
Simon Knowles, CTO of machine-learning chip developer, Graphcore, on how customers are asking for a million times more computational power than they currently can access – [and how to deliver that](http:// https://www.youtube.com/watch?v=Fm9-W2brkVQ&index=5&list=PLMU7XLs-Lrl9L7Ud-cAg0g5iPXVjvgWD2).
James Wang, Analyst at ArkInvest, on what it will take for AI hardware to bridge the gap between the performance we desire and the performance we get.
Pete Warden, tech lead of the Tensorflow Mobile team, answers: why is the future of machine learning in cheap tiny devices?
Jack Clark, Strategy and Communications Director at OpenAI, on why creating measure infrastructure for AI is essential to securing its progress globally.
We’ve compiled a valuable overview of the first day of The Cutting Edge at CogX, which dealt with algorithms, the limits of deep learning and data. This includes links to all the other excellent talks of the day.
Shadbolt & Schmidhuber: Killer robots will only exist if we are stupid enough to let them exist.
Deepmind releases a paper on “Neural Scene Representation” which infers a 3-d spatial map of a scene from incomplete inputs. The result is akin to the commonsense awareness we have of what a room is likely to look like (even from a glance.)
Here’s a generative adversarial network that can predict what you’ll look like decades in the future.
MIT Researchers show off RF-pose, a system which allows wireless sensors to “see through walls” and predict how people are standing.
Short morsels to make you smart at dinner parties
😲 Bankopalypse now. Britain’s high street banks are shutting at a rate of sixty a week.
Chinese mega-firms have generally avoided competing in Europe. Not so the fintech giant, AliPay, which is launching payments services in 20 European countries.
Bitcoins incredible bull run last year was likely artificially inflated.
🚦Superb. How traffic light phasing disadvantages pedestrians over cars.
🛴 Bird, the alternative mobility firm, has reached a $1bn valuation faster than any startup, as the 2-wheeled mobility gold rush continues.
Might video games soon get paid more than professional sports people?
The geography of inequality (1) Why living in a poor neighbourhood can change your biology: “Perceived discrimination contributes to poorer mental and physical health among ethnic minorities.” (2) Girls outperform boys in both maths and English across the US. Except in white, suburban, wealthy neighbourhoods, where boys do better in maths.
Find your where your address was 240 million years ago.
Travelling to Proxima Centauri? Here is how many crew you need to ensure that a healthy crew arrives.
Could cultivating temporal bandwidth help you survive high-speed society?
Azeem's end note
At Transform.AI in Paris, I got a chance to walk with Jed Kolko, the chief economist of Indeed.com. Jed has access to vast troves of data--Indeed is the number one job site in the world with 200m monthly visitors.
He shared some of his perspectives on workforce changes in the face of the industrial shift, platform work and automation, as we looked for family gifts.
Jed’s observations are worth reading in the context of the automation stories at the top of this week's EV:
The US labour market is incredibly tight right now. There are more job vacancies than there are job seekers. This means the impact of automation may be attenuated. However, transition remains difficult. Workers face barriers to mobility: high housing costs prevent people from moving to many places where jobs are plentiful, and licensing and regulations can make is harder to enter some occupations.
Gigworkers (like uber drivers) have yet to show up in US labor statistics. This is either because of mismeasurement or because, despite the hoopla, they aren’t really that relevant in the economy today. One early conclusion: older workers are the gig economy. They're far more likely than prime-age and young adults to be in such alternative work arrangements.
Lack of competition might reduce workers' bargaining power. The rise of non-compete clauses and the local market dominance of some employers can limit outside opportunities for workers.
Jed is a super-smart guy whose twitter feed is worth following, especially if you are interested in a US perspective of automation’s impact on work.
This week was crazy busy with CogX and travel, normal service will resume next week ;)
Hasta la vista,