Wiser Technology Advice Blog

HOME
WISER-TECHNOLOGY-ADVICE-BLOG
ARTIFICIAL INTELLIGENCE ON THE EDGE

Artificial Intelligence on the Edge

27 August 2018

Last week I attended the Hewlett Packard Enterprises Re-imagine Summit in Melbourne, where I listened to speakers on a range of topics and met with some very interesting people.

Have you heard of edge computing?

These days we expect our phones and wearable devices to be smart â which means they have massive amounts of memory storage and computational capacity compared to the computers of the past. We're now seeing an increase of other smart devices that make up the internet of things, such as smart TVs, internet fridges, smart sensors, cameras and of course autonomous vehicles (Driverless vehicles - the future is here).

These smart devices are defined as operating at 'the edge'.

Edge computing allows data from the internet of things to be analysed at the edge of the network before being sent to a data centre or cloud computer.

This is starting to solve the problem of âbig data' â the massive volume of information that is gathered by devices in the internet of things, which is too much to be handled by central computer systems.

We need computing at the edge to have local processing of data that makes it valuable in real-time. For example, if you wish to send marketing messages to customers as they're walking past your premises, you need that to happen in real-time. If you have to wait for data to go back to a data centre, be processed then come back out as a marketing message the moment will have passed.

But traditional systems and apps are not enough. These can only cope with traditional text-based data and are not able to understand unstructured data from videos and images.

That's where artificial intelligence comes in.

What is Artificial Intelligence?

Artificial intelligence aims to mimic human behaviour.

Machine learning is the first generation of artificial intelligence, where patterns are matched by algorithms that look through thousands of examples.

Deep learning is a subset of machine learning and is the latest development of artificial intelligence, also referred to as neural networks. These neural networks are trained using massive amounts of sample data, to develop intelligent behaviour. Once trained, deep learning algorithms can infer decisions based on imperfect and incomplete data, making what seems like us to be intuitive leaps.

The opportunities are endless

Combining the latest in artificial intelligence with microprocessors operating at the edge opens up endless opportunities to get value from all the data gathered from the internet of things.

For example, artificial intelligence at the edge can be used for traffic management. Here's how it works.

Firstly, neural networks are trained using masses of data in a huge supercomputer that sits in a data centre, or using a virtual supercomputer of linked computers in the cloud. The purpose of the algorithm in this example is to recognise motor vehicles that are unregistered or otherwise of interest to police, by scanning through a live video data feed.

Once the neural network's algorithm has been thoroughly tested and is ready to be used in the real world, it's sent to microprocessors linked to smart devices in the internet of things. In this example, the microprocessor's installed along with the video camera in the black-box at traffic lights.

Once switched on, the artificial intelligence is able to analyse the unstructured data coming in from the live video data feed. When it detects a number plate of interest, that part of the video is sent to operators for further review.

This is an ideal use of artificial intelligence, combining the power of neural networks to provide relevant information, but leaving the decision about what action to take in the hands of humans.

There are many other ways this artificial intelligence at the edge can be applied. For example, bags left unattended at an airport being detected by artificial intelligence and security guards notified, replacing the need for security guards to constantly monitor videos.

Artificial intelligence at the edge can also be used to pick faces out of the crowd and alert security guards or police to a potential problem. Facial recognition could massively speed up entry of crowds into sporting stadiums or entertainment venues, reducing the need for people to be stuck standing still in a crowd, waiting to get into the venue.

But wait, there are risks to consider…

Deep learning is where artificial intelligence starts to get a little scary. Companies such as Google, Facebook and Tesla don't understand what their artificial intelligence algorithms are doing â because humans can't understand deep learning artificial intelligence algorithms.

And these deep learning algorithms are only as good as the data they're given in training. Human bias can inadvertently be introduced. Many algorithms which are trained to do a human job, such as recruitment, credit scoring, assigning a prison sentence or diagnosing a patient, will have been trained on human generated data, from past occurrences of a person doing the job. If they are trained on data which contains human bias then of course the algorithms will learn it, but furthermore they are likely to amplify it. This is a huge problem, especially if people assume that algorithms are impartial and won't have the same biases as people.

When artificial intelligence evolves to self-awareness, human beings could be in for big trouble. Bill Gates (founder of Microsoft), Elon Musk (Tesla) and Stephen Hawking have all been quoted as being very scared by artificial intelligence.

Imagine a world where a robot has self-awareness and has been given a task to perform. Nothing will stop that robot from completing its task. Humans will not be able to stop the robots. Boston Dynamics have already developed a robotic dog that can't be stopped from opening a door…

You are the product

If you're not aware of this already, you should be. Remember that if it's free, you are the product.

We've enjoyed the benefits of free social media and internet search for years, but these companies make their money by selling your data. All of the free online products we use, such as Facebook, LinkedIn and Google, are funded by them selling advertising space to companies. And that advertising is targeted based on your data that you share.

But were you also aware that free wi-fi is funded by selling your data? Shopping precincts are only able to offer free wi-fi because they're selling data about customer movements to the marketers, who use that information for research or for targeted advertising.

And Bluetooth can also now be used to track customer movements. Bluetooth signals are more accurate than wi-fi, enabling your movement to be tracked to a much more precise degree, for example telling a supermarket which products you're looking at on a shelf.

In conclusion

The future is coming, whether we're ready or not.

It's getting harder and harder to opt out of being part of the data that's captured and sold to marketers. You can opt not to use free wi-fi and turn off your smart phone's Bluetooth, but cameras are still going to be watching and analysing.

Whether you choose to be bothered by that or not is up to you. You might like to concentrate on the problems and risks, or you might like to think about how artificial intelligence can be used to make our world a safer place.

For now, the best use of artificial intelligence is when it's used to enhance human capabilities.

The best outcomes come when recommendations by artificial intelligence are made to human experts who decide on the appropriate action.

Want to talk it through?

If you want to take advantage of these latest developments in technology, get in touch with me today to explore your options!