View profile

Twill #5 - Personal heating bots, pre-crime and Facebook's AI reads Peter Pan

Revue
 
A personalised heating bot, the many ways we favour right over left, what cities to avoid once robots
 

Twill by yiibu

March 13 · Issue #5 · View online
A weekly dispatch on our evolving relationship with technology. Topics: Near-future design, autonomous everything, design ethics, machine learning and the human impact of embedding technology into everyday life.

A personalised heating bot, the many ways we favour right over left, what cities to avoid once robots come to take our jobs, and a look at NASA before Powerpoint.

This week...
0.0 The fascinating idea that heating and cooling humans might in the future be more practical and efficient than heating and cooling physical spaces.
“…high-tech undergarments, fan-equipped office chairs, and even a personal attendant robot that follows you around blowing hot air
1.0 An exhaustive study on right vs left handedness and the many conscious and unconscious ways we favour one hand (or even simply the concept of ‘right’) over the other. 
where people saw pairs of alien creatures—one on the left side and one on the right side of the page—and we asked which alien in each pair looked more honest, or less intelligent, or more attractive. On average, righties attributed more positive qualities to the alien they happened to see on the right, while lefties preferred the creatures they saw on the left.”
2.0 A short piece outlining three potential ways an ecosystem of bots might evolve. Ultimately, a bot is just an app, and the past ten years has taught us that most people can’t be bothered to interact with much more than a handful of them a month. What might cause this to change this for bots? (PS - If you’re into AI and robotics you may want to subscribe to the newsletter this story is from). 
3.0 A list of US cities that are most at risk to lose jobs to automation. This based on the general thesis that knowledge and creative work will be hardest to automate. Maybe someone should triangulate this list of cities with those most at risk from climate change? (Source PDF report from Oxford University including more global data.)
4.0 An entirely unmanned grocery store in Sweden. Open the door using your app. Scan purchases. Walk out. Interestingly, the store is targeting older demographics who live in small villages whose services are slowly dying due to de-population. Calls to mind the “shop on the subway with QR codes” concept first trialled in Korea by Tesco (and now available at Gatwick airport). that was aimed at busy commuters. 
5.0 A first stab at creating a VR code of ethics
Unlike other forms of media, VR can create a situation in which the user’s entire environment is determined by the creators of the virtual world…[This] introduces opportunities for new and especially powerful forms of both mental and behavioral manipulation, especially when commercial, political, religious, or governmental interests are behind the creation and maintenance of the virtual worlds.
6.0 A study (PDF) to determine how users could easily protect themselves from automatic image-location tech (similar to this recent Google project). 
GLE research should strive towards the vision of the geo-privacy aware camera. Such a camera would use visual information retrieval techniques to alert users in real time when they take a picture that potentially leaks their geo-location via its visual content.” 
Not sure if the proposed contingency would really help in the end. You can imagine a future where we all start taking slightly odd photos…which the AI then learns to disambiguate…which causes us to take even weirder photos… 
7.0 This isn’t first pre-crime’ initiative, but may well be the largest. The Chinese Communist Party has directed one of the country’s largest state-run defence contractors to develop software to collate data on jobs, hobbies, consumption habits, and other behavior of ordinary citizens to predict terrorist acts before they occur.
We don’t call it a big data platform but a united information environment
Right…
8.0 Facebook’s AI is learning by reading lots of children’s books. Training its AI this way is part of what Facebook calls the Goldilocks Principle (PDF) which tests the role of memory and context in language processing and understanding.
Not sure how reading The Jungle Book, Peter Pan and Little Women will really help a modern AI. If anything, it may breed some uncomfortable and outdated assumptions about society. But hey…the books are all Public Domain, so if the AI gets really smart, at least Facebook won’t risk a 'derivative work’ lawsuit.
9.0 Long but interesting and well researched read: “Is advertising morally justifiable?
10.0 Facebook Lite, a super lean (<1MB) application aimed at users on 2G networks now has 100 million users. Here’s how they built it. Sad thing is, much of what they discuss could/should be best practice. There’s no reason most apps couldn’t be built this way.
Because Steph was briefly in Hong Kong this week...
(If you’ve never been to Hong Kong, and want a glimpse of what a future of 5-7 billion urbanites will look like (we’re ~3.5 billion so far)…here’s some awesome photos).
Thing from the past. Thing from the future
The future: In the future, all the kool kids will be wearing a Bluetooth-enabled, 250-earbud SongWig?
The past: NASA before PowerPoint (~1960). (Source unknown but appears from the watermark to be the now defunct Life magazine)
That is all...
Our best wishes for a productive week.
If this is not of interest, feel free to unsubscribe. If you think a friend or colleague would benefit from what we share, please pass this on!
Did you enjoy this issue?
If you don't want these updates anymore, please unsubscribe here
If you were forwarded this newsletter and you like it, you can subscribe here
Powered by Revue