December 15, 2021

10 Everyday AI Applications That You’re Not Even Aware of

By Charlie Isaksson

Not long ago the hype of artificial intelligence (AI) exploded like a Mike Tyson punch. It spread across the globe like a wildfire. However, that generation of AI has already started to be phased out by new-gen AI. 

While all new-gen is based on the original AI, the new-gen differs by relying entirely on algorithms that teach themselves by monitoring humans. 

AI is already intertwined with our lives from the moment you wake up until you go to bed. Let us take you on a familiar journey that explores 10 everyday AI applications. 

#1 Smart Assistant Orders More Coffee

It is 7:00 AM, and your alarm clock gently wakes you up. Your smart home recognizes movement and starts adjusting the temperature to your liking. Alexa, your virtual assistant developed by Amazon, lets you know that it started your coffee maker. You ask Alexa to order more Lavazza coffee beans. It all happens seamlessly. Alexa uses Automatic Speech Recognition (ASR) to take the audio stream over the cloud and turn it into a text string that is sent to the Natural Language Understanding (NLU) system.

The NLU interprets the recognition result and produces an intent. In our case, the intent is to order more coffee. And the entity is Lavazza beans. The response system takes what was produced by the skill (some skills are interactive and will ask follow-on questions that require answers) and uses text-to-speech (TTS) to generate the audio speech file to stream the audio to your device. Alexa is designed to get smarter every day.

#2 Spam & Phishing Blocker

While you are relaxing over a cup of coffee, your smartphone starts letting you know of your emails. You are relaxed because you know Spam and Phishing emails are being blocked by advanced AI models that utilize Natural Language Processing (NLP) and Deep Neural Network models to understand the intent behind emails. In cases deemed suspicious, the email is swiftly moved to the spam folder.

#3 Traffic Jam Detection

You are now headed to work. Google is warning you on your smartphone that there are massive traffic jams on your usual route and recommends you to take an alternative path. Google Maps utilizes AI to generate predictions based on both historical traffic patterns and live traffic conditions. Google partnered with DeepMind to improve the accuracy of traffic prediction capabilities by using Graph Neural Network methods.

#4 Code Automation

It is 8:00 AM and you are at work. Your boss is already standing there to let you know that you have to code a module by the end of the day. Normally, you would be stressed because you know it would take yet another late evening at work. However, you are at ease because you know the GitHub Copilot will auto-generate code for you. 

Copilot is a new service from GitHub and OpenAI that is powered by a deep neural network language model called Codex, which was trained on public code repositories on GitHub. You simply start coding and the copilot will complete the rest for you.

#5 Spell Check Supercharged

It is 5:00 PM and you are smiling as your code and test is ready and committed to Git. However, you stay a little bit later to finish your blog post. Even though your writing is great, Grammarly helps you fix basic grammatical mistakes to avoid overused words and keeps your writing concise. 

Grammarly’s AI system combines machine learning with a variety of NLP approaches. The algorithm learns by being presented with correct and incorrect sentences to recognize the mistakes and fix them. 

#6 Take Quality Pictures of the Night Sky

It is 6:00 PM and you are headed home, but you stop to take a photo with your colleague. Although it is dark outside, you are glad that you have a Google Pixel because, with the latest update, Night Sight can snap a photo in very low light. The night sight process is very complex. The phone utilizes both hardware and AI to capture images or videos where human eyes are unable to see. 

The camera already starts buffering frames even before you press the button. Based on gyroscope measurement, an AI model is used to choose the best burst capture settings to balance between noise and motion blur. All these captured frames are aligned before merging. Depending on your phone version, HDR+, or Super Res Zoom for Pixel 3 and above, is used for merging images. To provide accurate coloring balance, a Fast Fourier Color Constancy model is used. The final step is to determine ideal light-to-dark tone mapping. This is to make sure the photograph conveys a dark scene. 

A picture of a person standing against a backdrop of stars

#7 Pizza Delivery by Autonomous Car

It is 7:00 PM and you are back home. If you live in Houston, Texas, you order Domino’s Pizza and get it delivered with its self-driving car. The autonomous cars are packed with AI models that need to handle real-time inputs from LiDAR, video camera, GPS, rear camera, ultrasonic sensor, radar sensor, odometry sensor, etc. 

This is not news: Tesla introduced this feature some time ago. However, the latest innovation in autonomous cars is the ability for the AI model to learn by themselves by observing human drivers.

#8 Netflix’s Personalized Recommendations

You are exhausted and ready to sit down and watch a movie. You have no idea what to choose from. Luckily, you are not overwhelmed because you know Netflix has your back. Netflix not only gives recommendations based on star ratings but also has vast amounts of data to give each customer personalized recommendations. The Netflix recommender system consists of a variety of AI models that compiles information for each member based on various factors including the device, time of day, day of the week, intensity of watching, etc. 

Netflix will also rank top picks based on a few personalized recommendations from their entire catalog. It will also consider what is trending, newly added videos, video-video similarity, continue watching, and many other options.

#9 Self Cleaning with Roomba

While you are enjoying your pizza and watching a movie, your Roomba vacuum packed with AI models is efficiently cleaning every square inch of your home. The earlier models of Roomba are composed of a multi-step AI modeling system. The first step is to extract visual features from cameras. Then, based on the image, a model is used to get the current position. 

Finally, a model is used to move in that direction. The current version of Roomba uses deep neural networks to allow Roomba to navigate in the real world by only using the camera without relying on any other sensors such as LiDAR, Radar, IR, GPS, or IMU.    

A picture of a circular robot vacuum cleaner with a stereo on it's top.

#10 Robo Dog

Because you work hard, having a dog is not an option. No worries, Boston Dynamics delivers a robotic dog named Spot. The dog can dance, open doors, and can even work with other Spot dog packs. You can take Spot anywhere imaginable. Spot is 83 cm tall, weighs 70 lbs, can carry up to 14 kg, and can run with a top speed of 3 MPH. 

Spot can go without food (battery) for an average runtime of 90 min. Your robot dog also has an arm named the spot arm that can pick up your clothes, dig holes for your plants, and open or close your water tap. Best of all, your Robo-dog is fully autonomous with AI beyond imagination. The dog has a LiDAR to measure distances and map out its 3D environment.

Is AI Friend or Foe?

Just like many aspects of human behavior and social judgments, AI needs to be consistent with our ethical judgments. AI needs to exist with a purpose and its behavior needs to be justified within our ethical frameworks. Steven Hawking gave us a clear warning about the rise of artificial intelligence: “It will either be the best thing that’s ever happened to us, or it will be the worst thing. If we’re not careful, it may very well be the last thing.”

One such concern about AI technology already in use is the ability to distort reality. One clear example is the DeepFake technique based on deep learning techniques. The algorithm can create images and videos that are almost impossible for humans to distinguish from the original. With very little data and time, you can create a video imitating someone’s face and voice.

The algorithm uses an autoencoder-decoder. The encoder detects the face in the image or video, and a decoder transforms it with the target face.

As excited as we are about human advancement in AI, at phData, we view all aspects of AI from a bird’s eye view and take any preventative action necessary. 

Harness the Power of AI with phData

The future is coming quickly, and artificial intelligence will certainly be a part of it. That is why organizations should be thinking about how it fits into their organization. Choosing the right solutions for development and enabling technologies requires a diligent approach —  and we’re here to help!

We have experience solving tough machine learning problems and putting robust solutions into production. If you’d like to leverage our insights in your AI initiatives, don’t hesitate to reach out!

Data Coach is our premium analytics training program with one-on-one coaching from renowned experts.

Accelerate and automate your data projects with the phData Toolkit