artificial intelligence behind the scenes

What are Examples of Artificial Intelligence Working Behind the Scenes?

AI is all around us, and whilst some may think that this term refers to autonomous robots like we’ve seen in movies like ‘I, Robot’, ‘Ex Machina’ and ‘Blade Runner’, the truth is that artificial intelligence is visible in so many day-to-day activities and technologies.

In this blog, Turing Technology Founder Amir and Head of Marketing Joseph are going to tell you about some of the most common and most interesting uses of AI that you may not have realised. And, as an AI developer, Amir is going to tell you a little bit about how each of them works, in layman’s terms of course.

Facial recognition on street surveillance cameras

London has more CCTV than any other city on Earth, and it has done for quite some time, so the British are more than accustomed to being watched. Whilst currently it is humans and some softwares that are observing the 4–5 million surveillance cameras in the UK, they are on the verge of implementing AI technology with the help of Israel’s tech industry.

In Russia, this is already a reality. AI technology in Moscow has been applied to 170,000 surveillance cameras in a move that many residents feel is ‘spying’. Human resources had not been capable of processing millions of hours of footage each day, and so AI was seen as necessary to increase the security of the city.

In these kinds of systems, AI is not the base of the technology, but acts as a secondary assistant technology, which helps to increase the probability of successful recognition. You can learn more about how it works here.

image source: Facebook’s new face recognition knows you from the side

License plate surveillance

Automatic Number Plate Registration (ANPR) systems are common on highways throughout the developed world, and use a combination of industrial camera technology and intelligent sensors to read and track license plates. This information gets fed into a central system, where it is processed. The UK Home Office states that ANPR is used by law enforcement agencies in the United Kingdom to help detect, deter and disrupt criminality, including tackling organised crime groups and terrorists.

Same as for facial recognition technologies, AI can be used as secondary technology to improve the recognition results.

image source: ANPR (Automatic Number Plate Recognition)

Virtual Personal Assistants

Siri, Genie and Cortana are artificial intelligence! They use voice recognition and a natural language interface to answer questions, give recommendations and perform actions through your internet services. The more you use them, the more intelligent they seem to get, the better they recognise your voice and dialect, the more helpful the search result and the more individualized their performance. Who needs a robot to make dinner when Siri can order you a pizza?

There are even a few research papers where Deep Learning algorithms were used to synthesise human voice with all its errors and dialect. You can read about that here.

image source: Venture Werks — Blog — Siri — A Prmer

Facebook News Feed

When you react to a post with a like, a wow, a sad face or a love heart, Facebook takes note, and it learns what makes you happy or sad, and will target content to you to inspire different emotions. They performed a study of 689,000 users in conjunction with Cornell University and the University of California to deliberately influence moods based on this AI algorithm.

The findings were interesting: when you see more positive content, you post less positive content, and when you see more negative content, you post less negative content. This scientific experiment produced a huge backlash, but it’s just one way in which the newsfeed applies AI.

Facebook learns who you are, and pushes content that it thinks you will click on and enjoy. It sees who you engage with and shows you more of their content, too. If you don’t like seeing friends’ baby photos, and you never hit like, you’ll probably stop seeing them. The same goes for sports, movies and music — if you don’t engage, you see less. The technology, called FBLearner Flow, makes 6 million predictions per second.

If you are interested in the possibilities which Facebook has, you can install a chrome extension which shows you a tiny bit of what is possible to do with your FB interactions.

Video Games

Think of Call of Duty, story mode. You’re not alone there, you have other soldiers or combatants helping you fight, and they are applying AI algorithms to help you. They comprehend their environment, listen for enemy sounds, take cover, use strategic manoeuvers and much more.

Sports games too — usually you’re just controlling one player, so what about the rest? Yep, they’re AI too, learning your style and adapting to your movements and tactics.

image source: Examining the AI Scripts in Call of Duty 4’s Developer Tools

Amazon recommendations

Sometimes known as ‘purchase prediction’, those pesky emails from retailers like Amazon apply artificial intelligence to recommend items for you, based on previous purchases and browsing history. Predictive analysis is not always popular, but the idea is to anticipate your needs rather than encourage you to buy superfluous items. In theory, it really is a win-win situation.

Of course these systems don’t rely only on Artificial Intelligence, it is data science that is used in most cases and predictions are made based on multiple systems working together, where AI is only one part of it.

image source: Introduction to Artificial Intelligence (AI) at Amazon

Fraudulent transactions

If banks are suspicious that there may be irregular or fraudulent activity on your account, they’re quick to let you know. But it’s not a member of staff at a computer emailing you, it’s their Artificial Intelligence and algorithms at work. As it is artificial intelligence, it only gets stronger and more secure over time

They apply their algorithms to huge data samples and are able to make educated comparisons between fraudulent and non-fraudulent activity, meaning when something irregular happens, then can notify you immediately for confirmation. This technology doesn’t rely only on AI, but it is an important aspect.

image source: Fraud Detection Using Neural Networks and Sentinel Solutions (Smartsoft)

Chatbots on websites

Sarah: ‘Hello, can I help you with anything?’

There is no Sarah, it’s just a name given to an Artificial Intelligence designed and programmed to help you function on their website or online store. These bots are clever, as they combined their FAQ knowledge and programming with the questions that people ask over time, and build up a great library of knowledge. Even when they can’t help, they provide links or contacts to someone who can, meaning they work as a fantastic intermediary.

If this interests you, you should check out this nice project on github or live application on Coinbase support page.

Keywords: #word2vec #seq2seq #nltk #rnn #deeplearning

Smart Home Integration

AI can recognise when you are leaving work, will see how much traffic there is, and can set your heating so that it reaches optimal temperature just as you walk in the door. It’s kind of spooky, but it’s certainly helpful and quite comforting to know your home will look after you if you look after your home. Artificial Intelligence lighting systems will turn lights on and off as you enter and leave rooms, will respond to voice commands, and can even sense moods and will dim or brighten lights for different activities.

There are several ways to apply technology like this, but the most advanced Smart Homes are using AI besides fuzzy logic and statistical methods.

Artificial Intelligence serves us well…

The likelihood is that you’re being served by AI right now. Maybe this answer wasn’t written by an AI expert, but by a machine. Who will know? We will never tell you. Beep boop.

Leave a Reply

Your email address will not be published. Required fields are marked *