Azeem Azhar’s Weekly Wondermissive: Future, Tech & Society
This issue has been supported by our partner: Ocean Protocol.
Ocean empowers data scientists to drive healthcare innovation in combating pneumonia.
You are reading the free edition of Exponential View
If you’d like to access special perks reserved for Premium members, you can become a member for $100 / £77 per year, or $12 / £9 per month.
Dept of the near future
📳 With 1.6bn sold, the iPhone is the most successful product of all time, argues Horace Dediu. Total sales have exceeded $1tn, generating $131bn in taxes. But the iPhone story is one of dynamic pivots, such as the changes in distribution, financing and the App Store. “Apple, since its inception, has always been oriented around its customers, not its products” argues Dediu, and that such a “priorities-driven company habitually re-designs its processes and its resources.”
⏱️ Rodney Brooks: artificial general intelligence has been delayed. Strongly argued and compelling. Like Brooks, I doubt the optimistic predictions, such as those by Ray Kurzweil, that we’ll see artificial general intelligence in ten years. We will see powerful, wonderful things coming out of the AI domain by then, but not that. A large part of Brooks’s argument touches on the current state of autonomous systems, a topic I dive into in some depth with Professor Missy Cummings 🔈 in this week’s podcast. Listen on Apple Podcasts, Spotify, Stitcher, Overcast, Google.
🗳️ Gillian Tett on pick’n’mix politics: around the world “established parties are losing out as voters and politicians create their own political playlists.” What innovations will we need to adjust if this is the new normal?
Do we need more referendums? Online polls? Entirely fluid parties? Is there a better way, in other words, to engage political voices in a digital, personalised age?
💬 The BBC is trying to develop a “public-service algorithm”, which optimises break filter bubbles on its audio content. Newsfeed and recommendation algorithms need not take us to clickbaity extremes (as they do on other Western platforms like Youtube, Netflix and Facebook, in order to increase engagement and time on site). They can push us towards exploration and, even, shared experience. The BBC’s unique funding allows them to explore in this manner. I am curious to see whether listeners will like it. Of course, regulators (or the law) could insist that recommendation algorithms on major platforms meet some sort of civic hygiene standards rather than allow a race to the bottom of the brain stem.
We have a shorter Dept of the Near Future because I’ve bundled many important points in the Climate Catastrophe section below.
🌡️ 🌎 Dept of climate catastrophe
Each week, we’re going to remind you of the CO2 levels in the atmosphere.
May to June marks the annual high of CO2 readings in the atmosphere. So over the next few weeks we should expect the reported level to drop, although it will be dropping against a strong upward trendline. The key numbers to watch are the delta over the last year and the absolute value, 450ppm being an agreed nasty threshold.
- The latest measurement (as of May 16): 415.39 ppm; 12 months ago: 409ppm; 50 years ago: 326.66ppm; 250 years ago, est: 250ppm.
We exceeded the 415ppm level for the first time in millions of years a few days ago.
👿 Thirty-seven years ago, Exxon Mobil predicted that we would hit 415 ppm in 2019. Yes, in 1982, the oil firms own climate scientists pretty accurately predicted the trajectory of carbon levels.
Some good news, over 2m electric vehicles were sold in 2018 and the rate is accelerating. As predicted in this newsletter in previous years, Bloomberg’s New Energy Finance unit has increased its projections for electric vehicle uptake. By 2040, BNEF reckons 57% of all passenger sales will be EVs and a similar proportion of light commercial vehicles will be electric. Free prediction: they will revise their forecasts upwards AGAIN next year. (One of my favourite analysts and EV reader, Gregor Macdonald, is more sanguine than I am, pointing out that total US EV sales only increased 10% year-on-year.) The reason I am bullish is because of feedback loops: as more people buy EVs, more people experience them and understand their benefits, demand will take off. (Just one advantage—EVs require virtually no maintenance.) My canaries in the coal mine are London taxi drivers, whose increasingly rapid switch to the LEV electric taxi is palpable as you saunter the streets of our capital. Those driving them are avid promoters. Those stuck on diesel are distinctly less averse to switching than they were last year. And charging infrastructure is improving in tandem.
This issue has also been supported by Wired
“Data isn’t king: the model is king. And that’s what principled AI means,” said Vishal Chatrath, CEO of Prowler.ai, an AI platform for decision-making.
Vishal is one of the speakers at the upcoming Wired Pulse event dedicated to the future of artificial intelligence and machine learning.
🌟 EV readers get exclusive 10% off—get your tickets now 🌟
Dept of artificial intelligence
🎭 The City of San Francisco has banned facial recognition technologies. Meanwhile, many police forces are going wild with new facial recognition technologies.
💯 EV reader, Stephanie Hare, speaks on the BBC about the real problems with facial recognition. The full report on how badly British police are rolling this out is available to people in the UK via BBC iPlayer. Start at 17min 51 seconds to get the full report.
A Chinese court case details how a man stole 10,000 yuan from his sleeping roommate by using face recognition to unlock his buddy’s phone and transfer the cash. (Apparently, that particular face recognition system will work on a sleeping face 30% of the time.)
Residents of Toronto rebel as they realise Google’s plans for a smart city there resembled more of a surveillance city.
📦 Amazon is rolling out machines that can pack 600 boxes an hour. Each machine can replace 12 human packers.
Asset manager, Invesco, reckons that automated systems could replace bond fund managers. Their analysis of returns by bond funds suggests that easily-described factors, rather than manager skill, are responsible for fund outperformance.
Somewhat speculative but a good intellectual exercise: What will be the expectations of the first generation borne into an age of artificial intelligence?
A Chinese investor is suing an “AI-powered” hedge fund for $23m in losses.
- Paper by Joshua Cowls et al, on seven critical factors to consider when designing AI systems for social good. Academic, but accessible and interesting;
- Important research on measuring how resilient a neural network may be to adversarial attacks, an important element of building trustworthy systems;
- Teaching self-driving cars fear;
- Paradoxes in machine learning. Super-fun essay.
Dept of funding the new economy
This section looks at current VC funding trends, places them in historical context, and features data from our friends at Dealroom. It is available to Premium subscribers.
Short morsels to appear smart at dinner parties
🧐 The Chinese bike-sharing market has been a bloodbath with high-profile bankruptcies, such as that of Ofo, which burnt through $1.5bn of capital. HelloBike is bucking the trend, relying on novel use of data, as a way of improving the fundamentals of the business. The company, I am told, now handles 25m rides a day for about 15m daily riders. Excellent profile here.
⚠️ Just 376 people hold a third of Ether, a crypto-asset, equating to about $8bn.
Job offer acceptance rates at Facebook have collapsed in the wake of the Cambridge Analytica scandal.
Some facts about abortion in America.
🌊 Researchers found a plastic bag in the deepest natural trench, the Mariana Trench.
Archaeologists found a coin off the coast in Australia which might date from a 14th century Africa.
Danielle Steele has written 179 books. She works 20-22 hours a day. Merely reading this profile may exhaust you. 😮
🐀 The Hanoi Rat Massacre was a great example of unintended consequences of technology and poor incentive design.
😍 Descartes vs. the ferret. The French philosopher thought animals were soulless mechanistic devices. This mother ferret proudly showing her babies to her human owner is no machine.
Fascinating: designing proteins which can move in predictable and tunable ways.
Earlier this week, I spent a day or so with the most senior execs of some of the world’s biggest companies. They represented the top execs from market leaders in finance, apparel, manufacturing, logistics, media and many more sectors. I had several discussions with them about AI and how their firms are using it. I came away with three strong conclusions.
The first is that these firms are really amping up their capabilities; I could feel the difference in the level and class of investment to three or four years ago. The second is that they have barely scratched the surface with applying machine learning and AI to their core businesses, or whatever they will become. However, and thirdly, in many cases these businesses still lack the ability to hire the very best machine learning talent and are stuck wading through the mud of innovating in a legacy enterprise.
I’m looking for someone who is active on Hacker News, Product Hunt and the reasonable parts of Reddit who can help us get the word out about our podcasts. If this is you, could you drop Marija (firstname.lastname@example.org) a line?
P.S. Do scroll down to read what other readers are up to.
This issue has been supported by our partner: Ocean Protocol.
See how different signaling mechanics as a means of data curation can facilitate data discovery. Here is part 1 of A Generic Guide to Data Curation.
What you are up to—notes from EV readers
Tom Glocer, former CEO of Thomson Reuters, says regulators should order Facebook to delete its entire database of user information.
Rachel Coldicutt at DotEveryone releases a fascinating report on British tech worker’s attitudes. Read the executive summary. More than a quarter of tech workers have seen decisions made about a technology that they felt could have negative consequences for society.
Dawn Duhaney on why ethnic and gender diversity alone won’t be a solution to bias in AI.
Daniel Murray’s fantastic TedX talk: laughter builds resilience.
Thomas Jestin considers alternatives to GDP in the age of automation and why they’re necessary.
Marija Gavrilov on her concerns over China’s investments in surveillance technology in her home country, Serbia.
A number of our readers at NewLab are bringing this series of events on existential medicine to life. Next up is a session on computational medicine on June 11.
Want to share your news with the community? Email email@example.com.
Sign in or become a Exponential View member to join the conversation.