Is AI a tool, a gimmick, or the death of us all? Part 4

• Part 1
• Part 2
• Part 3
• Part 4
• Part 5

Disclaimer: What you're about to read is a made-up scenario. While I use Apple as the protagonist, I have no clue if Apple is working on this or has even thought about this. I chose Apple as it has the technical know-how and the desire to put users before profits.

AI is starting to get its move on. The speed of development has some worried that it's moving too fast. Elon Musk, among others, is calling for a slowdown. It seems to me that he's calling that because he can't get Autopilot working in his Teslas.

In my last post, I mentioned Apple's been giving us AI for a decade now since the iPhone 5s. It's most likely been longer than that as well. Apple often makes products that have us question why someone hasn't already done it. For this reason I use Apple for this scenario which shows how AI can help us beyond imagination.

Here is my AI powered mapping system of the future. And this is the start of what AI could be doing for us in the future.

Dataset

Let's lay down some ground work for this example. We'll use the following data:

• Car 1: All iPhone users - Heading North
• Car 2: All Android users - Heading North
• Car 3: Mixed iPhone / Android - Heading North
• Car 4: None cellphone or smartphone users - Heading North
• Car 5: All iPhone users - Heading South
• Car 6: All Android users - Heading South
• Car 7: Mixed iPhone / Android - Heading South
• Car 8: None cellphone or smartphone users - Heading South

Each of these cars is full with 4 passengers. The amount of passengers doesn't matter but a people mover could supply a better dataset. We'll go for 4 per car though giving us a total of 32 people.

We'll assume that ALL iPhone and Android users have location services turned on. This isn't real life but given the amount of cars on the road it's plausible we have 8 cars with this setup.

It is obvious that Apple can track all iPhone users in this scenario. As location services are on on all capable phones, Apple can see the Android phones as well.

It's possible to know the locations of non-smartphones, so long as they have Bluetooth on. Given that Bluetooth on non-smartphones sucks battery life, we'll assume BT is off.

That gives us a total of 24 phones providing Apple with data.

In the two mixed device cars, we'll assume the iPhone users are sitting on the diagonal from each other.

Also, as I'm from New Zealand the driver will be on the right hand side of the car. This makes no difference to the mapping system. It's more to not confuse Americans.

In this example all cars are either driving from Wellington to Auckland or vice versa. We're going to use the Bolton Hotel in Wellington and the Auckland Airport. This is roughly a 643km trip by car.

All mapped out

So now we have our dataset, let's see how Apple could use it in an AI powered mapping system.

If you live in NZ then I'm going to make a statement that will rattle people's cages. Google Maps is crap and Apple Maps is the far superior mapping system. I'm going to give you two examples with one being this little story.

When Google Maps introduced driving directions to New Zealand I did what anyone does. I tested a drive from my place of work to my home. What Google gave me was a weird trip through a back road to my house. No matter what variables I tried it always took me through this route. On closer inspection I found the answer. Google showed a paper road from the main road to my house. A paper road is a road that exists only on paper. That section of road Google showed as "paper" had existed as an actual road for 10 years before I moved into my house. In other words, Google Maps was out of date by 10 years on release in New Zealand. That same search when Apple Maps came out sent me the correct way home. I wasn't even living in that house when Apple Maps came to NZ. When AM came to NZ I was contracting to a company I used to work for 10 years earlier. I checked AM to GM for the area I was working in. AM showed the map with the correct bypass and driving directions. GM once again had 10 year old maps. Google Maps is NOT a trustworthy mapping system in NZ from my experience. In fact the only reason I use Google Maps today is when I'm making run lists for my posties. Apple Maps doesn't show house numbers when you zoom in. Google Maps does. But...

midsummer ave stratford - Apple Maps
midsummer ave stratford - Google Maps

Apple Maps is correct, Google Maps is wrong... and that's after having recently updated their maps with this sub-development. Google Maps is still missing a street in this area as you can see from the Apple Maps data.

I do realise that your mileage will vary and it might be the other way around but I don't have any faith in Google Maps.

AI Mapping

So in this scenario Apple is developing a new mapping system. They've been developing this system since it first came out. Until this point they had been using TomTom maps for their data. But as you'll see, Apple's AI system is superior to anything on the planet.

Apple collects the location data from all the iPhones in this scenario. As such it can see where they are all going. Now this is where it can get interesting. Even if each iPhone doesn't have "Find My" setup, Apple can see where each phone is in relation to each other. It has Location Services option which will give it GPS and cellular triangulation. It recognises each phone is going the same way at the same time in the same location. But with Bluetooth, Apple can also pinpoint the exact location in relation to each other. AI takes over and realises that all these phones are travelling in the same vehicle. It can even pinpoint which seat each iPhone is in. It sounds creepy I know but it has a real benefit. It can use that to workout the width of the road from information from Waka Kotahi.

So from those 4 iPhones Apple knows where they're going, how many are in the car, and the width of the lanes.

Beside this car is our mixed car. Once again Apple knows all it needs to make its mapping system accurate. It now knows they are on a multi-lane highway. It already knew that with the TomTom information but it doesn't need that anymore. It's now got realtime information to build the map itself.

Getting the information off the Androids is harder but not out of the realms of possibility. Apple can buy that information off Google. But it only needs to register that they exist so it can do that with Bluetooth, even if they are in a car going the other way. It now has information of how wide this highway is because it has registered cars going the other way.

If you're reading into this deep then you've likely realised what this scenario is showing. Instead of having a mapping system relying on printed maps, you've got a map built realtime. Why is this an advantage?

Ch Ch Ch Changes

Further up the road to Auckland traffic comes to a stand still. There is a crash further up that has blocked off the left lane. Traffic now has to merge into the two other lanes. Cars slow down for the merge. All the iPhone users now get an alert saying there's a crash. But instead of another driver logging that crash, AI has already done it. How?

The police in NZ use iPhones. Not sure about the other emergency departments but I suspect they do as well. Apple knows this and has noted that fire, ambulance, and police are at this location. It's tracked them from their stations to this location. Of course, being emergency services, it's also hidden their locations from others. This is to prevent nasty people tracking them. Apple's AI Maps concludes, because all three of those services are at this location, there must be a crash. Anyone using that mapping system gets alerted. If it can it can reroute drivers according to this information.

The same conclusions can tell about roadworks, tree felling, and the like. Location data is providing Apple realtime information about how our cars are moving. New road layouts get generated adhoc in realtime.

Beyond Maps

Have you heard of Project Titan? Anyone following Apple news and rumours would have. It's supposed to be Apple's secret self-drive car project. I doubt this a lot. I suspect we have already seen the first phase of this project with Carplay 2. I suspect that Apple is working on a set of APIs that will make autodrive more available to all sorts of vehicles.

This would be quite a revolutionary approach because it means Apple wouldn't need to build a car. It would develop the APIs and car manufacturers would use those APIs. This would make their cars self-drive with no effort on their part. All they would be doing is giving Carplay access to the engine components.

The simplest implementation of this would be with cruise control. A car that adjusts cruise control based on the car in front uses lasers to achieve this. With Apple's APIs, they could do this with cheaper Bluetooth transponders at the front and rear of the car.

But give the APIs access to the steering wheel and you can see AI do cool stuff. Remember that in our scenario Apple has developed a map on the go using AI? So now you have a self-driving car with no stored maps, moving with traffic based on a real-time generated map. It's more accurate than any map currently in the system. It's would be capable of sending you along routes based on information from your vehicle. Think about it, no more stuck trucks on narrow streets.

Problem solved

Self-driving cars are going to fix many issues. No more traffic jams because every car will stick to the 2 second rule, or 4 second rule on rainy days.

But our scenario could even cut down on the amount of crashes. Carplay becomes CarOS and gets used in every manufacturer of cars. This allows them to reduce the cost of developing a self-drive car. It would allow them to concentrate on building better cars instead. As each car uses the same OS, each car communicates better with each other. So how can this prevent crashes?

Say a car hits a patch of black ice. We know this ice exists because the other cars have registered the same ice. Only this car has had a sudden drop in tyre pressure - a blow out. CarOS transmits this information to every other car using the same OS. The other cars move out of the way to reduce the chance of collision. They create a safe path for the out of control car.

What about barriers? There will be none as they are not needed. Why would a self-drive car need a barrier when it's not dumb enough to drive into the path of an oncoming car? The above situation is able to happen because there is more than enough room to gain control.

Traffic jams become a thing of the past as the system has all the information about traffic flow. During peak times the traffic will be more one way than the other. For instance. In the morning, traffic will be flowing into the city and vice versa at the end of the day. So traffic travelling out of the city would reduce in lane use and the other lanes fill up. Because all the cars are following correct distances, bunching up can't happen. This allows for a continual flow. Also, merging traffic will merge in the correct way which reduces the crashes and jamming.

Many throw ridiculous scenarios with regards to AI. Scenarios like who will a car hit given the choice between a child, grandma, or a motorcyclist. When raising this they're being contrarian for the sake of it. The AI system I went through above would reduce even that. The relational data gained will put human actions in relation to cars. AI will predict what a pedestrian will do and act to avoid it. No choice needing hard decisions.

In this scenario I've showed how AI could revolutionise travel. This isn't even limited to the roads. AI can replace traffic controllers at airports. They can route ships and trains. Everything will communicate with each other and avoid each other. The world will run smooth.

But will this lead to the end of the world? Will Terminator and the Matrix be our future? Let's look into that next.