At today’s WWDC Apple gave everyone their first look at upcoming versions of their four main software platforms. There was no shortage of intelligent assistance on iOS and macOS to satisfy the tech nerds or messaging bling for seventeen-year-old girls with bionic thumbs. It was an impressive show that alleviated any fears that Apple is giving up in the great software arms race.
I wasn’t surprised one bit by Apple’s emphasis on local intelligent assistance. This is in contrast to Google’s approach which uses server-side logic by supercomputers. Google devices are forever tied to the internet like the Trade Federation’s robot army in Star Wars Episode One. Cut the cord to the mother ship and all that Android N hardware falls limp.
I never understood all the naysayers who thought that Google’s server-side approach was the only way. It’s not the only way and may even be the inferior way. I used to design and implement inventory management systems for an $8-billion-dollar-a-year company. I know what it’s like to setup a software intelligence that is designed to be populated at the local level but is designed at the central level. Let me give you an example.
When designing the system for our new Venezuelan subsidiary we had to use a FIFO (First In First Out) inventory system as opposed to a standard cost system like we use in the United States. This was due to hyperinflation of their currency. I designed and tested the program so that every time inventory was received, a new layer of inventory was created with the new PO price. Every time we sold and shipped a unit, the sales management module would take the most recent layer of inventory and the cost of goods sold was based on the attached PO price.
The cost of the “layers” of inventory were auto-populated variables that no one at the corporate headquarters could see. The local servers operated independently of the corporate headquarters and knew what to do with the information. Yet, they relied on logic that was supplied by the headquarters which told them which cost to use when calculating a gross margin for that sale.
My point is that variables can be set up to be populated locally. Apple can design the intelligence without knowing the variables.
I fully expect that Apple must be doing something similar with iOS on the iPhone. At iOS central in Cupertino they can create the logic that is programmed to learn certain “layers” of our behavior and based on some preprogrammed logic, it would suggest calendar entries, mapped routes, restaurants, etc.
Is it just me or does it seem preposterous that you need vast oceans of data and super computers for your iPhone to figure out that you may want to stop at a Starbucks on your way to work or to have your boarding pass prior to a flight? It’s not like your calculating the speed and angle of entry for a Martian lander. Most of the what the local phone needs to calculate will be fairly mundane and repetitive but no less helpful. And the stuff that can’t be done locally is so far away that even Google’s setup can’t handle it. Real artificial intelligence is so far away that even a company that doesn’t exist could one day own it.
Just for the sake of argument, let’s say that Google’s cloud services give them an edge. They have the equivalent of the IBM mainframe. For a little while, you need a huge centrally located apparatus to calculate simple equations or understand basic sentences. But eventually, as devices get more powerful, this can be outsourced to the local device. I still remember how in the nineties we marveled at the small size of our three inch thick laptops compared to the old mainframes, never mind today’s iPhones or Apple Watches. Eventually Google’s monster apparatus becomes obsolete, rendering its maker obsolete.
If the market swings hard to local intelligent assistance then Google is in big trouble. It’s not that hard to imagine the appeal of local- versus server-side central.
1. You don’t need to be tethered to Wi-Fi or data. In theory, local intelligent assistance allows your phone to continue assisting you whether you’re in an airplane, on a cruise, or just out of range of a cell tower.
2. Data Security – If Apple doesn’t know what you’re doing they can’t surrender the data to the government and hackers who could get into Apple’s system can’t grab it either.
Is Google’s cloud service in danger of going the way of the IBM mainframe? Probably not, but it’s just as valid a question to ask as whether or not Apple is the next Blackberry.
I never understood all the naysayers who thought that Google’s server-side approach was the only way. It’s not the only way and may even be the inferior way. I used to design and implement inventory management systems for an $8-billion-dollar-a-year company. I know what it’s like to setup a software intelligence that is designed to be populated at the local level but is designed at the central level. Let me give you an example.
When designing the system for our new Venezuelan subsidiary we had to use a FIFO (First In First Out) inventory system as opposed to a standard cost system like we use in the United States. This was due to hyperinflation of their currency. I designed and tested the program so that every time inventory was received, a new layer of inventory was created with the new PO price. Every time we sold and shipped a unit, the sales management module would take the most recent layer of inventory and the cost of goods sold was based on the attached PO price.
The cost of the “layers” of inventory were auto-populated variables that no one at the corporate headquarters could see. The local servers operated independently of the corporate headquarters and knew what to do with the information. Yet, they relied on logic that was supplied by the headquarters which told them which cost to use when calculating a gross margin for that sale.
My point is that variables can be set up to be populated locally. Apple can design the intelligence without knowing the variables.
I fully expect that Apple must be doing something similar with iOS on the iPhone. At iOS central in Cupertino they can create the logic that is programmed to learn certain “layers” of our behavior and based on some preprogrammed logic, it would suggest calendar entries, mapped routes, restaurants, etc.
Is it just me or does it seem preposterous that you need vast oceans of data and super computers for your iPhone to figure out that you may want to stop at a Starbucks on your way to work or to have your boarding pass prior to a flight? It’s not like your calculating the speed and angle of entry for a Martian lander. Most of the what the local phone needs to calculate will be fairly mundane and repetitive but no less helpful. And the stuff that can’t be done locally is so far away that even Google’s setup can’t handle it. Real artificial intelligence is so far away that even a company that doesn’t exist could one day own it.
Just for the sake of argument, let’s say that Google’s cloud services give them an edge. They have the equivalent of the IBM mainframe. For a little while, you need a huge centrally located apparatus to calculate simple equations or understand basic sentences. But eventually, as devices get more powerful, this can be outsourced to the local device. I still remember how in the nineties we marveled at the small size of our three inch thick laptops compared to the old mainframes, never mind today’s iPhones or Apple Watches. Eventually Google’s monster apparatus becomes obsolete, rendering its maker obsolete.
If the market swings hard to local intelligent assistance then Google is in big trouble. It’s not that hard to imagine the appeal of local- versus server-side central.
1. You don’t need to be tethered to Wi-Fi or data. In theory, local intelligent assistance allows your phone to continue assisting you whether you’re in an airplane, on a cruise, or just out of range of a cell tower.
2. Data Security – If Apple doesn’t know what you’re doing they can’t surrender the data to the government and hackers who could get into Apple’s system can’t grab it either.
Is Google’s cloud service in danger of going the way of the IBM mainframe? Probably not, but it’s just as valid a question to ask as whether or not Apple is the next Blackberry.
RSS Feed