The amount of data CERN produces is stagggggering (spelling mistake intentional) and without AI it would be humanly impossible to get through. When the electrons are spinning round that 27 km tunnel at speeds approaching the speed of light, the production of data on collision reaches 300 gigabytes per second. Only AI could have reduced this to a ‘manageable’ 300 megabytes per second of what they call ‘useful data’. The proof is in the pudding, as they say, and the pudding in this case: the recent discovery of the fabled Higgs boson particle.
RBS realized that though they have much more data on customers than, for example, Amazon, they do little with it. They decided to utilize AI (irony alert) to bring a more personal touch back to banking. Customers who contact the bank on their birthday get a ‘Happy birthday!’ Customers who are paying twice for a product like insurance can be informed of that. So if someone calls to tell you it’s time to get your holiday jabs – it could be your bank manager!
Milton Keynes is one of the UK’s ‘new towns’, built in the 1960s with a target population of 250,000. Estimates for the next decade expect it to reach 350,000, and the local authority has to make the expanding city work. Enter AI and the Internet of Things (IoT). Using sensors and monitors, bins can be emptied only when completely full, lights turned on only when there is footfall in that area and traffic lights adjusted to prevailing traffic conditions, making the town fitter, leaner and with great abs. All software used is open source, cutting down on implementation costs.
GE Power supplies 30% of the planet’s electrical power and has been using AI to predict when maintenance will be needed, when and where power will be in demand and, ultimately, to create a completely networked model for power (before, presumably, retiring to Mordor). Early results: 25% cut in operating and repair costs; 5% cut in unscheduled downtime. Multiply that with 30% of ‘all the electricity’ and you get the kind of figures accountants swoon over.
Fitbit sold 14m units last year, and that’s a lot of heartbeats. As individuals track their vital statistics for longer, they become less a fitness tracker and more a rich source of medical data. Microsoft has developed HealthVault, for example, to allow patients to give their docs access to their Fitbit bits, and John Hancock will give you a discount on health insurance if you share your data with them – and is shifting to selling ONLY ‘interactive health insurance’, as it’s called.
Shell’s exploratory drills cost big bucks, but have been a bit hit and miss in the past. They used to take around 1000 readings before deciding if a site was worth test-drilling, but with their new AI team, they can do a million or more and analyse that data in no time, increasing the probability of success massively. They’ve used Hadoop on Amazon Web Service to crunch some 46 petabytes of data so far – with no indigestion reported.
Peter Glenn is a US outdoorsy goods chain, something like the one in Last Man Standing. When they hired an AI company and unleashed it upon their customer database, it discovered that over 80% of the people on the database hadn’t walked into one of its locations for years. They started re-engaging and then targetting customers based on needs, likes and locations. On the way, they boosted their AOV (Average Order Value) by 30%.