🔮🎈Societies & gods; AI’s better lesson; how good is Waymo; what we owe the future; magnetic humans, mammoths and dead products++ #210

Exponential View is 4 years old!

Mar 24, 2019

Azeem Azhar’s Weekly Wondermissive: Future, Tech & Society

Ocean is partnering with private and public sectors to unlock data and power of AI:

See how Ocean Protocol and ConnectedLife are working together to combat Parkinson’s.

You are reading the free edition of Exponential View.
Here’s the last chance to get your annual access to Exponential View Premium for $90 or £68 a year—that is just over $7 or £5 per month. (From Wednesday, we revert to $100 or £76 per year or $10 per month.)

Subscribe here to get instant access to:

Other Exponential View readers. “It's terrific that The Exponential View now has a formal, close-knit community to discuss these ideas.” — Kenn Cukier, Senior Editor at The Economist;
Weekly insights from me and experts in my network. “The value was abundantly there in terms of the quality of the content and its relevance to my professional (as well as personal) life.” — Ronald Ashri, Founder & CEO, GreenShoot Labs;
Regular live briefings which you can access from anywhere in the world. Our next briefing is on April 18th, with Bill Gross, the founder of Idealab & lifelong inventor, among whose creations is the cost-per-click model which became the backbone of Internet monetisation.

We have a special deal for students and academics. Email support@exponentialview.co to learn more.

Dept of the near future

🔱 Do ‘big societies’ need ‘big gods’? The ‘moralizing god’ hypothesis suggests that big societies need omnipresent gods to ensure the large populations felt they were always being watched—and thus behaved well. Now large-scale research looking at 414 polities over 10,000 years suggests that in most cases big, complex societies preceded the arrival of moralizing gods in those societies. Furthermore, in no cases did “big gods” precede the development of large-scale complex societies. Peter Turchin writes, “big Societies are highly fragile when they first appear and they need many such institutions [like big gods] to make them more resilient to internal and external shocks.“ (More extensive academic paper here.)

🔮 What do we owe the future? Roman Krznaric: “It is clear that a movement for the rights and interests of future generations is beginning to emerge on a global scale,” and we may need a new democratic revolution that “empowers future generations and decolonises the future.”

🍏 Apple Show, Apple’s rumoured TV offering, will launch on Monday. With 1.4bn active devices, Apple is well-placed to grow its subscription business. It’ll create headaches for Apple, giving rise to yet more conflicts of interest as the firm runs a platform, an app store and many competing services. This is part of a firm-wide move towards services revenues, which make up just under a quarter of the companies take, although are growing about 19% annually, faster than other product lines.

🧐 What will the world of work be like in 2035? In reality, no-one knows, but the Royal Society of Arts gives it a great shot looking at the UK. Four scenarios: precision economy of hyper-surveillance & optimisation; the big-tech economy dominated by major platforms; the exodus economy struggling under a low-innovation recession, and the empathy economy, where half of us become masseurs for the other half. The RSA also polled UK MPs on key issues 43% of MPs feel they can make sound judgements on technology policy, which seems surprisingly high. Nearly half felt the ramifications of technology are being felt in Brexit policy. Nearly two-thirds were in favour of stricter competition policy, which suggests it may be on the cards. The stand-out for me was that 30% favoured some form of universal basic income—suggesting UBI may actually become part of the policy agenda in coming years.

🚸 How good is Waymo’s self-driving car service? Amir Efrati reviewed 2,500 trip reports and found that nearly 2 in 5 featured a complaint of some kind. Waymo’s digital drivers received a five-star rating 62.6% of the time—a useful benchmark for human drivers is that 94% of Lyft rides are, apparently, five star. Interesting details on how well the robotaxi service is working in the sanitised environment of Phoenix’s wide roads. (Watch a Tesla Model 3 repeatedly try to steer in a lane divider.)

⚗️ How China is doubling down on materials science. China now publishes more high-impact research papers than any other country in 23 fields with clear technological applications, including batteries, semiconductors, new materials and biotechnology. Grants have increased four-fold since 2009. (Azeem’s comment: the global materials industry is ripe for smart innovation, as so much of it is dependent on brute-forcing chemistry through the expensive application of temperature & pressure and petro-derived feedstock.)

🔥🌏 Burning planet: 410.13ppm

Each week, I’m going to remind us of our level of the CO2 in the atmosphere. We must avoid a level of 450 parts per million for a chance to keep global warming below 2°C.

This week’s level: 410.13ppm (12 months ago: 408ppm; 250 years ago, est: 250ppm; likely to exceed the target in 10-15 years.)

One bit of optimism: I grew up in the oil era where we worried about peak oil supply. It turns out, we may be close to peak oil demand--a topic I’ll return to in my upcoming podcast season.

By the end of this year, a cumulative 270,000 barrels a day of diesel demand will have been displaced by electric buses, most of it in China.

In the past six months, very rare extreme weather, likely worsened by climate change effects, has whacked the USAF jet inventories more than any military adversary since the 1991 Gulf War (and, arguably, in terms of the dollar cost of lost material, more than anything since the Vietnam War.) Hurricanes destroyed nearly 10% of the US F-22 Raptor fighter fleet (cost $210m per craft), and floods disabled Offutt Air Force Base, the home Strategic Air Command.

Dept of artificial intelligence

Rich Sutton, a pioneer in reinforcement learning, reflects on the “bitter lesson” that AI methods which “leverage computation are ultimately the most effective.” While it is tempting (and satisfying) for researchers to codify prior knowledge into their systems, eventually the exponential increases in computation make favour approaches involving search and learning, he argues.

Rodney Brooks counters Sutton with his “better lesson” that “we have to take into account the total cost of any solution, and that so far they have all required substantial amounts of human ingenuity.” Brooks’ argument is elegant, and worth reading. Take this point:

Computer architects are now trying to compensate for these problems by building special purpose chips for runtime use of trained networks. But they need to lock in the hardware to a particular network structure and capitalize on human analysis of what tricks can be played without changing the results of the computation, but with greatly reduced power budgets. This has two drawbacks. First it locks down hardware specific to particular solutions, so every time we have a new ML problem we will need to design new hardware. And second, it once again is simply shifting where human intelligence needs to be applied to make ML practical, not eliminating the need for humans to be involved in the design at all.

On the topic of massive computation, here is what it takes to build an exaflop supercomputer.

⚔️ Another wrinkle added to the attack surface: new research suggests that medical AI systems could be vulnerable to adversarial attacks. (Very interesting. In our future state of hybrid war, you could imagine a foreign adversary overwhelming parts of a health system by inveigling adversarial data into machine learning diagnostics.)

Nvidia’s GauGAN turns doodles into realistic paintings.

Dept of people

As the structure of our economy changes, so too will the nature of the safety net. EV reader, Nicolas Colin, reckons that entrepreneurs will be part of creating this new welfare net. (FWIW, I concur that startups will form part of this new social infrastructure.)

👻 Debunking the capitalist cowboy. Nan Enstad on how the popular image of the innovative entrepreneur is based more on marketing than reality. Companies often succeeded through leveraging their power and manipulating corporate law. (Anton Howes has some great examples of this, dating back to the 16th century.)

No sleep, no sex, no life. On the ruthless conditions amongst China’s startup workers.

My former colleague, James Stanier, argues that, similarly, Silicon Valley’s do-or-die culture is costing us:

Growth-or-die culture, hustle culture — whatever you wish to call it — is beginning to leak beyond the office and into the daily lives of a whole generation of workers...many young founders have only been exposed to one way of doing business: grow at all costs until you exit or self-destruct. They can sometimes apply that same logic to their own lives.

👍 In praise of the older software developer:

Making the software industry more welcoming to coders past their thirties and creating roles suited for very experienced programmers will make companies more effective and fairer. These changes will also benefit the rest of us [...] older coders [need] to stay in the industry long enough to pass their knowledge to their successor

🇮🇳 Insights into the large-scale data annotation businesses growing in India. These are the people-powered businesses who build the data on which to train AI systems.

Short morsels to appear smart at dinner parties

🧭 Magnetic-field-sensing humans.

💯 A brief introduction to Wright’s Law. (You’ll be hearing more about how Wright’s Law how it describes exponential learning curves in the coming years. It predates Moore’s Law by decades.)

Cemetery of Google’s products and why they died.

Scientists reawaken cells from a 28,000-year-old woolly mammoth, but no cloning in sight just yet.

🌳🌳 Using drones and satellites to build a marketplace for forest carbon offsets.

Gorgeous Jupiter, with all its storms! 😍

Warner Music signs a deal to release 20 algorithmically-generated albums this year.

End note

In March 2015, I sent a short email to about 25 friends. After seven years of running a company, and selling it to the good folk at Brandwatch, I raised my head from the operational tempo of being a founder, looked around and noticed something.

With 20 years in the industry, I felt there was something really interesting about the moment we had got to. Less eye-opening than the early dawning of the potential of cyberspace that gave rise to the Internet bubble, but certainly more profound.

There were so many fascinating technologies accelerating and combining. It was accelerated by a global digital infrastructure that stretched from factories across supply chains and into the supercomputers gripped in our hands. And finally, it was grinding, mercilessly against our political, social and economic systems. Observing this, and trying to understand it, is the mission of Exponential View. (I came up with the name by issue four.)

I wanted to bring my training in philosophy, economics & politics together with a three-decade interest in the computer industry to give an interdisciplinary assessment of the world and the near future. I wanted to bring my experience as an investor, in startups and public equities; as a founder; as a regulator; as a corporate executive; and as a journalist, together to help us understand more about the dynamics of these changes. And I wanted to do it with emoji.

Four years on, Exponential View reaches almost 40,000 people. These 40,000 are building the future. It counts founders, investors, politicians, novelists, scientists, artists, academics, researchers, policy wonks and parents across dozens of countries. And I’m honoured that you read it. And delighted that I learn so much from you.

It is more than a list of links. The context, narrative, critique and open questions surrounding the content matters. In any given month, I’m lucky to speak with a dozen or more founders, many investors, policymakers, researchers at the top of their game, and mainstream business leaders. The community that clusters around Exponential View is in large part building this future. And the newsletter and premium member’s service is a way of sharing how my thinking is developing based on what I read and the conversations I have.

Our podcast has now reached its third season. This new season kicks off in April and will examine new paradigms, what they are and how they affect us. Please look out for it.

🎁 This week marks the last week where we’ll be offering discounted access to Exponential View’s Premium service. So if you’d like to support the endeavour and access the additional benefits, please do so by signing up here. There is a special rate to celebrate our fourth birthday.

And if you’d like to wish EV happy birthday, you can do so here.

Thanks for reading,
Azeem

This week’s Exponential View has been supported by our partner: Ocean Protocol.

Ocean Protocol is kickstarting a data economy that is global and borderless

Here are 4 reasons why you should dive into Ocean.

What you’re up to — notes from EV readers:

Sarah Gold’s IF team published research on explainability in personalised machine learning, in the context of federated machine learning.

Alix Dunn launches Agile Ethics, a company devoted to building organisations for an inclusive future.

Ian Ernzer’s World’s Greatest Internship is a unique opportunity for emerging entrepreneurial talent from around the world. Deadline: August 27.

Hampus Jakobsson on founders, stress and burnout: “Founders are not fuel to be burnt on the altar of innovation”

Ali Jiwani and Upile Chasowa are launching For Inquisitive Minds, a monthly event featuring cutting edge research from top PhD students. The first event will be held on March 27 and feature two topics: the cost of energy storage, and the evolution of quantum computing. EV readers can use code "EV50" for 50% off the price.

Azeem was appointed to the board of the Ada Lovelace Institute, an independent research and deliberative body with a mission to ensure data and artificial intelligence (AI) work for people and society.

Got something to share? Email marija@exponentialview.co

Exponential View