No government by experts in which the masses do not have the chance to inform the experts as to their needs can be anything but an oligarchy managed in the interest of the few. – John Dewey, “The Public and its Problems”
In 1927 Dewey probably had a different image of threats to democracy. What this passage does echo is a concern that democracy becomes captured, contorted to the will of a few.
Technological development isn’t nazism. It is, however, an incredibly powerful tool by which a small number of people (scientists, writers, coders etc.) can have disproportionate impacts on the world and society.
When technology amplifies the impact of a small number of people the flip-side is obscuring the others – the impacted. This is politically important. Rule by the few isn’t a core democratic principle.
Dewey worried about governance by ‘experts’ or authority figures.
Would he also worry about Smart Cities? If our cities are festooned in smart connected devices, empowered by A.I. to respond in real-time to events, are they experts? If their intent is not clear, if the line of authority they answer to opaque, would he call it oligarchy?
I think Johnny Soraker illuminated this point last week (facebook live video) when he asked if we can accept the political cost of a ‘good’ A.I.’s 15% false positive rate. 15 out of 100 times you won’t get what you might be entitled to – a loan, a parking space, a public good.
If we give A.I. (and an internet of things through which it acts) the status of experts, Dewey would probably worry.
IoT: Automatic Government
A focus of local government is to distribute public goods and maintain a social security net. A professionalised state helped – with varying degrees of success – to deliver public goods to citizens (health care, law enforcement, public lighting and water).
We all know the stereotype of politicians and lobbyists carving up the pie. Yet when faced with egregious policy we still hold the power to protest and ultimately kick them out.
Why the Internet of Things could change politics | World Economic Forum — www.weforum.org
With so much data, policy can be automated.
As you can read above, the promise of the internet of things is that it can help to automate policy. Inputs picked up by connected devices (water meters, cameras, occupancy sensors) can immediately translate into policy and action.
Synchronisation of data flows and decision making might result in the automatic selection of the “best” possible policies.
This model bypasses conventional public controls like civil servants and politicians. Public scepticism of these institutions has never been higher but they are better than nothing. Replacing Sir Humphrey with Sir Watson make policy more remote not less.
The article above is only one model but it is an intriguing view of what connected devices and decision-making A.I. means for politics and government.
The policy “production” process might be utterly redesigned. Data collected by devices we use on a daily basis (such as vehicles, domestic appliances and wearable sensors) will provide precious evidence about the drivers of personal voting choices, or the impact of government decisions.
…think tanks, lobbyists, advocates and campaigners – will have to find new ways to use and explain this data. They will still have a role proposing and communicating alternative policy scenarios, but they will also have to consider algorithm-generated policy options. In this context we will probably see changes in political communications.
For professionals, technology means an idealised process where raw data is processed by A.I. and algorithms to produce policy decisions. This means two things:
- Completely takes ‘the people’ out of the picture
- Reinforces the biases built into A.I. or machine learning and builds it onto surveillance technology.
Smart Cities: Civic Tech Innovation and The Internet of Things — www.gothamgazette.com
Reports from a recent Smart Cities conference in New York.
Plenty of cities are starting to cast the net on data gathering. Don’t get me wrong, there are incredibly powerful and positive ways these devices can make lives better.
The way that happens is as important as the outcomes.
The Internet of Things is like a pair of brackets. The opening bracket provides real-life inputs from the world and cities around us. The black box in the middle (A.I.) does its work and the closing bracket implements those decisions in the world.
Just one example:
Nvidia wants to mine metadata from surveillance cameras in real time as a smart city solution.
Automatic image processing can do certain good things: it can control traffic flows to ease congestion or alert emergency services to an accident.
It can also do some other things – like alert law enforcement to ‘suspicious looking people’ or suggest areas that only need trash collection half of the current time because it is run-down.
There are dangers in the technology, which won’t stop rolling forward. There are dangers in the way policies and actions get processed and implemented. The farther removed from people, their representatives and their input, the more dangerous.
Automatic governance wouldn’t be my ideal. Government does not develop its own ML or AI. It will be reliant on private providers like IBM and Microsoft for processing.
How are ethics and democratic input accounted for in agreements and rollouts? They might want to talk to emerging companies in this space.
I can’t answer the rhetorical “what would Dewey do?” but I suspect his response would seek democratic control and accountability for the ‘smart’ administration of our public spaces.
Automatic for the people is not the same as automatic by the people.
Ads are coming to Amazon Echo skills — www.engadget.com
If the Beauty and the Beast and Burger King debacles were anything to go by, owners of smart speakers like the Google Home aren’t big fans of ads. But that won’…
I recently discussed the Echo Look – a nuclear submarine in Amazon’s push to digitise our physical, biological lives. I suspected it was going to propel 1) Amazon’s ad targeting and 2) Amazon’s AI offering through AWS.
So perhaps no surprise to see Amazon’s partners attempt to develop ad platforms for the connected assistants. The ads are limited by Amazon’s TOS for now but I expect the boundaries to fall soon enough.
New ad technology can track people’s mood and gender — www.rte.ie
Advertisers are piloting new technology which can track people’s eye movements and determine their age, gender, mood and facial features as they pass digital ad displays on Dublin’s streets.
An Irish startup wants to use screens with sentiment-analysis to target advertising. This is not dissimilar to our Russian friends in Issue 1.
Essentially a camera mounted to the screen interprets your image for things like age, sex, mood and displays ads for you from its inventory based on the insights. The aim is to provide real-time ad exchange based on real-time demographic information fed from the board.
An odd wording suggests this is perfectly fine with EU data laws.
Because no images are actually recorded, the system is said to comply with Europe wide data protection regulations.
Over on Twitter knowledgable folks, any myself, wondered at how this kind of application fits with GDPR, data protection and privacy rights.
PPS: Security = Surveillance ?
I will just leave this here.
A home assistant/security device with always on camera, 3D sensor (to see where people are in the room/house) and facial recognition.
Lighthouse is an Andy Rubin-backed smart security camera that identifies people and pets – The Verge — www.theverge.com
The team at Lighthouse, a startup out of Android co-founder Andy Rubin’s Playground accelerator, doesn’t see its new hardware product as a home security camera. Instead, they see it as an…
Ultimately, it is up to us to decide if the purpose of these devices is the same as the USP. If we are not certain, or if companies are not clear, then we have to demand more information on what is going to happen to the digitised life-data these IoT devices send to servers.
How are companies using or selling it? Am I ok with that?
If we are to go down that route, a label which focuses solely on security is insufficient. There is a strong case that whether you, your behaviour and your data is monetised, how, and by whom should also be pretty clear.