Although it was expected that May would lose the vote on her Brexit deal, the scale of her defeat wasn't. The reaction from Brussels was predictable, too: basically, big FU from Donald Tusk and Jean Claude Junker. And perhaps that's part of the explanation. Yes, Brussels always had the better hand in the negotiations, but the lesson here is that even if you hold all the cards, you don't have to play them.
By leaving the impression that the EU was going to make it as painful as possible, MPs reacted in an economically irrational but "behavioral-economically rational" way: "If we're going to get screwed either way, deal or no deal, we're not going to give Brussels the satisfaction of getting what it wants".
So now things are going to get interesting. If Britain does look like its going to crash out, it sends exactly the signal the Brussels elite is worried about; to the populists this looks like sticking it to the man, and will embolden other leavers. Brussels should offer a substantially better deal. If it holds the line, Britain will crash out and others will be tempted to follow.
The future of the European project has never looked more uncertain.
Tuesday, January 15, 2019
Friday, January 4, 2019
O2O and IoT
In his book "AI Superpowers: China, Silicon Valley, and the New World Order", Kai-Fu Lee explains the divergence in approaches to AI between China and the US. Chinese startups are extending more directly and deeply into the physical domain, which Lee describes as "O2O" -- online to off-line.
These off-line nodes collect data; they are part of the Internet of Things (IoT). As they become more integrated with our daily lives the data they generate will provide an opportunity to comprehend and "simulate" behavior at the individual level. Much has been made over the last twenty years of the ability to connect disparate data elements--from credit card use in the supermarket to major purchases--to build a virtual picture of behavior. This takes the idea much further, with the potential to build an accurate picture of our routines, when we rise, where and what we eat, where we work, what we do for entertainment, all off-line activities, to compliment browsing data that internet companies already collect.
GE uses a simulation of its aircraft engines, parameterized with actual engine telemetry, to predict when parts are likely to fail. Imagine this kind of modeling applied to individuals. Target already showed us how subtle changes in buying and browsing patterns enable AI engines to make inferences about pregnancy.
In "The Minority Report" Tom Cruise moves though a world in which everyone is constantly surveilled and AI-based marketing presents him with individually tailored offers in real time; but what Lee's description suggests is far more disturbing. Imagine that these simulations know where we are likely to be, for example pre-ordering a coffee at Starbucks so that is ready when we come into the store. Or perhaps it senses departures from observed patterns of behavior and makes inferences from these.
As O2O becomes more pervasive every node in the network has the potential to become an IoT data collection point; Amazon Echo is already collecting audio data. Netatmo's Smart Indoor Camera already collects and uploads video data from which it does facial recognition (to detect home intruders). Lime Scooters in San Jose for example could include cameras that collect audio and video data as well as location and personal data.
We may soon be living in a world in which artificial intelligences know more about us than we do ourselves (and unlike us they never forget). If those AIs are operated by firms, that's concerning; if they are operate by governments, as is likely in China that's terrifying.
These off-line nodes collect data; they are part of the Internet of Things (IoT). As they become more integrated with our daily lives the data they generate will provide an opportunity to comprehend and "simulate" behavior at the individual level. Much has been made over the last twenty years of the ability to connect disparate data elements--from credit card use in the supermarket to major purchases--to build a virtual picture of behavior. This takes the idea much further, with the potential to build an accurate picture of our routines, when we rise, where and what we eat, where we work, what we do for entertainment, all off-line activities, to compliment browsing data that internet companies already collect.
GE uses a simulation of its aircraft engines, parameterized with actual engine telemetry, to predict when parts are likely to fail. Imagine this kind of modeling applied to individuals. Target already showed us how subtle changes in buying and browsing patterns enable AI engines to make inferences about pregnancy.
In "The Minority Report" Tom Cruise moves though a world in which everyone is constantly surveilled and AI-based marketing presents him with individually tailored offers in real time; but what Lee's description suggests is far more disturbing. Imagine that these simulations know where we are likely to be, for example pre-ordering a coffee at Starbucks so that is ready when we come into the store. Or perhaps it senses departures from observed patterns of behavior and makes inferences from these.
As O2O becomes more pervasive every node in the network has the potential to become an IoT data collection point; Amazon Echo is already collecting audio data. Netatmo's Smart Indoor Camera already collects and uploads video data from which it does facial recognition (to detect home intruders). Lime Scooters in San Jose for example could include cameras that collect audio and video data as well as location and personal data.
We may soon be living in a world in which artificial intelligences know more about us than we do ourselves (and unlike us they never forget). If those AIs are operated by firms, that's concerning; if they are operate by governments, as is likely in China that's terrifying.
Subscribe to:
Posts (Atom)