Artificial intelligence and the fridge

Milk-buying robots could mark end of the world
I bank, therefore I am?: a still from the science fiction film ‘I, Robot’

Listen to this article

00:00
00:00

Could an internet-connected thing — a smart fridge, a thermostat or a home-help robot — become a millionaire? This is not as ridiculous a question as it may seem.

If we do indeed move toward a world in which devices are connected to the internet and performing mundane chores, it is likely that many of them will be connected to some kind of bank account.

For maximum efficiency, the smart fridge that orders your milk should also be able to also handle the payments to the supermarket. You may not hook the fridge directly to a current account — worried perhaps a software glitch might cause the fridge to accidentally put you in debt by ordering £20,000 worth of dairy products in a single day. But you might set up a pre-paid account with a set amount of milk money that the fridge can access.

What happens if the fridge is able to buy your milk for 1p a litre cheaper than you have budgeted? Perhaps it is able to join with other networked fridges to buy in bulk, lowering the price. Or its ability to search hundreds of thousands of prices allows it to do some bargain hunting. The milk account grows a surplus. But is this money yours?

As the owner of the fridge you would assume so. But is this always clear? What if you do not own the smart fridge, but rent it? Or if it has been supplied to you, free of charge, by its manufacturers or by the dairy? Who owns the surplus then? Would it be like the deals some solar panel companies offer, where they install the equipment for free on the understanding that they keep the money from any electricity generated? At the very least, you will need to read the ownership contract carefully.

And what happens when the artificial intelligence capabilities of appliances increases markedly and the fridge is capable of making more complex decisions? What if, rather than keeping the milk funds in the prepaid account, it connects to an online stock market trading site, invests the money and makes a profit?

It still buys the milk, but now there is an extra £20 in the prepaid account. Does that belong to you? Will there be anything to stop an advanced, AI-enabled fridge opening a online bank account and depositing the money there without your knowledge? Should banks have a “humans-only” policy to stop this happening?

And what would a rich machine buy? A group of Swiss artists have experimented with this idea, creating the Random Darknet Shopper, an automated online shopping bot that was given an allowance of $100 a week in bitcoin and instructed to make one random purchase a week from an online marketplace. The items were displayed in an exhibition at the Kunst Halle in St Gallen.

To the consternation of the performance artists — and the Swiss police — the purchases included 10 ecstasy pills and a false Hungarian passport. This prompted debate over who was liable when a robot broke the law on its own initiative, a debate the police sought to clarify by seizing the contraband goods.

The idea of the shopping robot adds another dimension to the discussion about what a robot-filled future will be like. Last month, an open letter, signed by scientists and entrepreneurs including Stephen Hawking and Elon Musk, urged us to begin considering some of the ethical dilemmas and potential dangers posed by artificial intelligence. Who is liable if a robot has an accident, for example? How will a self-driving car choose what to do if the only way to avoid a head-on collision is to veer on to a crowded pavement?

Scientists at places such as the Future of Life Institute based in the US, the Centre for the Study of Existential Risk at Cambridge university and the Machine Intelligence Research Institute in California (yes, these really exist) are also considering what might happen if we are eventually able to create machines far more intelligent than humans — superintelligences that may break out of the bounds of their programming and begin to disregard human controls.

In science fiction, this scenario — called “singularity” or “transcendence” — usually leads to robot versus human war and a contest for world domination.

But what if, rather than a physical battle, it was an economic one, with robots siphoning off our money or destroying the global economy with out-of-control algorithmic trading programmes? Perhaps it will not make for a great movie, but it seems the more likely outcome.

After all, the financial crisis of the past decade has shown us how easily it could be done, even by dumb humans.

Copyright The Financial Times Limited 2017. All rights reserved. You may share using our article tools. Please don't copy articles from FT.com and redistribute by email or post to the web.