🔮 Voice revolution; manipulating media; lessons from Luddites; deep learning predictions; octopuses, buffalo and Canadians++ #95
|Azeem Azhar||Jan 8, 2017|
The voice revolution; how hackers manipulated the media; China’s Red packet revolution; lessons frm Luddites; the impact of autonomous driving; can Uber make the grade; life under authoritarianism; 6-valent carbon; 1-pixel cameras; cephalopods, buffalo and more.
Have great conversations!
Dept of the near future
⭐️ danah boyd: how hackers attacked the attention economy and manipulated the media narrative (See also: microtargeting of low-information voters.)
🔮 What did and did not happen in 2016 (and some guidance for 2017) by EV subscriber, Fred Wilson.
🇨🇳 Red packets: how WeChats money-as-a-message service has propelled China to the largest mobile payments market in the world.
🏭 What the Luddite rebellion can tell us about the coming job market challenges: “[They] weren’t opposed to machinery … if the profits from increased productivity were shared.” (Also important: “why men don’t want jobs mostly done by women”. Many of these jobs are likely to be in growing sectors, despite automation.)
🚘 Frank Chen: on autonomous cars and industry shifts. Superb 30-minute presentation
Dept of voice computing
Many of you who received an Amazon Echo for Christmas. It was one of Amazon’s top-selling products over the holiday period.
There is a double exponential going on with the Echo (and Alexa, the voice assistant embedded in it) right now. It is a clear milestone in the #voicefirst shift, the arrival of voice recognition as a primary interface.
First of all, deployment and usage of Alexa has spiked. By looking at downloads and usage of the Alexa app (available for iOS and Android) we can proxy Echo installations. Echo users need to download the app to configure the device.
The number of monthly users of the official Alexa app in the US alone increased five-fold in 90 days to Christmas. In Q1 last year, Alexa downloads were running at about 80-90k per month; by Q3 this has risen to 150k per month. November saw 500k downloads and December 2.5m downloads. Total US installed base is now estimated at 5m (about 4% of households.)
The second growth is the number of ‘skills’ that Alexa has. A ‘skill’ is essentially a service - like a newspaper, food delivery service or game - that you can access via Alexa. This time last year, Alexa had fewer than 100 skills, that had grown 10x by June 2016, and Alexa ended the year with 7,000 skills.
Alexa is riding the wave of dramatic improvements in automated speech recognition driven by deep learning. This graph, extracted from this paper by Nuance Communications, shows how ASR has gone from unusable in 2009 to pretty damn good in 2015. Since then, of course, there have been pretty impressive improvements all being deployed rapidly to the Google, Baidu and Amazon clouds. (Baidu’s English speech recognition is more accurate than a human typist.)
What makes ‘voice first’ interesting:
It is a very natural way for us get things done. We can dispense with the metaphors of forms, drop downs, menu bars, icons and form filling and replace it with a more natural modality: speech.
It kills the advertising and pay-per-click model that has made Google over the past two decades. This hasn’t escaped Google, says Sridhar Ramaswamy, Google’s SVP of Ads: “one thing that we are all clear about is the days of three top text ads followed by ten organic results is a thing of the past in the voice first world.”
It creates new choke points which businesses will need to navigate in order to reach their customers. Read Ben Thompson on how Alexa creates an operating system-like positioning for Amazon.
It might reduce smartphone distraction. Browsing by voice is not (yet) as easy as browsing on a phone but a voice interface does provide a simple way to handle phone like tasks (weather, reminders, time, ordering) without picking up a device.
Gartner group reckons that 20% of all smartphone interactions will happen by voice by 2020. Early data suggests that homes with an Echo spend 10% more at Amazon during the first six months of ownership. Alexa is also coming to VW and Ford cars.
EV reader Tom Standage has a brilliant leader on the possibilities for voice-computing in this week’s Economist. (Also recommended Lane Greene’s wide voice special in The Economist this week.)
Dept of AI and robots
⭐️ Excellent review of deep learning progress in 2016
Five predictions for AI in 2017 (reinforcement learning, adversarial nets, China, natural language, hype & backlash.)
Amazon’s robotic workforce increased 50% y-on-y
Japanese insurance firm to replace some white-collar employees with IBM Watson
Morsels to appear smart at dinner parties
What the TV wars of the 70s and 80s tell us about trade wars and globalisation
🍁 Is Canada the world’s first post-national country?
Quantum computers are ready to leap out of the lab
∮ Prof Keith Devlin: all the maths I learned at University become obsolete during my lifetime
💎 Six-valent carbon. Go chemistry!
Anti-surveillance clothing (defeats machine vision)
What we can learn from the cybernetic humanities
💊 The drugs we deserve: Patterns of drug use give an accurate insight into cultural history
👮🏼 Everyday authoritarianism is boring and tolerable
Reader special: Shift Forum
I’m helping my buddy John Battelle with his new Shift Forum, a private conference bringing together incumbents, startups, investors and policy leaders. John’s pulling together a really good confab to discuss many of the issues covered in Exponential View and how they may play out in business, government and policy.
The event is invite-only. John and I agreed that EV readers can skip the queue and get a $1000 discount by clicking here. Make sure to use the code Forum17. (The event is in February.)
Have a great week