Do you think a computer can identify your taste in people? Can the machine pick the right kind of wife or husband for you? Moreover, will you eat food pollinated by an artificial bee? The answers to all these questions lie ahead. But first, we need to talk about something which concerns every single smartphone user on the planet, earphones, and why are they so dumb?
Turn Your Dumb Headphones Smart
Scientists have found a way to turn your “dumb” headphone into smart devices. Xiaoran Fan, a recent Doctoral graduate from Rutgers, has invented a groundbreaking device that can transform hundreds of millions of smartphones into sensing equipment.
The new device is called Head-FI; it is a very affordable piece of equipment that fits into your current pair of headphones with the help of adaptors.
According to Fan, upgrading a smartphone using Head-Fi requires no major hardware changes. It can upgrade existing smartphones into sensing device very easily. Currently working at Samsung Artificial Intelligence Centre, Xiaoran Fan is one of the Head-Fi’s primary inventors.
According to a report, a peer-reviewed scientific paper including all the above-mentioned findings, will be officially published in October at MobiCom 2021. For the uninitiated, Mobicom is a conference that seeks to solve challenges related to “mobile computing, wireless tech and mobile technology.”
A ton of headphones nowadays already include gesture controls that are enabled via accelerometer and gyroscope. Head-Fi seeks to expand on that by upgrading normal headphones with such functionalities on top of heart rate monitoring. Such smartphones will play music and perform sensing tasks at the same time, details the paper. Head-Fi would only need to be paired with a computing device like a smartphone, and they’d seamlessly perform their task.
Using Head-Fi, a normal pair of headphones can sense vital body measurements, including a person’s heart rate. The research team working on Head-Fi had tested their invention on headphones ranging from the cheapest ($2.99) to the most expensive ones ($15,000).
They found 97.2 to 99.5% accuracy on user identification, 96.8-99.2% while measuring heart rate and 97.9-99.3% in gesture recognition.
Indestructible Robot Insects
Surely, the heading sounds something out of a horror movie, but it isn’t. These insects are the brainchild of MIT Assistant Professor Kevin Yufeng Chen. These robots are soft to touch about the size of a large bumblebee but turn into resilient entities when current is applied; this too sounds like something out of a science fiction movie. Trust me; it’s not.
In an interview, Chen detailed said that the main challenge in building a drone this small is power generation. An insect-sized motor will produce no usable power. But Chen found a solution in the form of a soft yet resilient drone made up of rubber cylinders coated in carbon nanotubes. These tubes squeeze and elongate as current passes through them, causing the drone’s wing to beat fast.
The tiny-cassette tape looking drone can flap its wing 500 times per second. According to Chen, the drone can recover from the nastiest of hits. It weighs in at just 0.6 grams which is equivalent to a large bumblebee. The next generation of these drones will look like Dragon Fly, according to Chen.
Maybe in the future, these artificial bumble-bees might pollinate our fields and help us produce our food.
AI Can Make Virtual Spouses
What if I told you a computer could create a perfect partner for you according to your preferences? Would you believe me? Well, you better, because researchers at the University of Helsinki and the University of Copenhagen have trained an AI to create virtual human models with “attractive” features.
The reason I highlighted the word “attractive” is because it is subjective. People in different parts of the world have dissimilar features that they feel are attractive on another human being. Associate Professor Tuukka Ruotsalo and Senior Researcher Docent Michiel Spapé are the ones who taught the computer what different humans think constitutes attractiveness.
Spapé explained that in previous studies, they made a computer identify simple features like hair colour, mood, etc. However, judging whether a black-haired woman is smiling or frowning comes under a pretty objective task. Attractiveness is difficult to judge because “Beauty is in the eye of the beholder.” said Spapé.
The scientists performed the study by showing a picture, one at a time, to a total of 30 people. They were asked to swipe right on who they think looks attractive, similar to the popular dating app Tinder. As each participant looked at the image, their brain response was recorded using Electroencephalography of EEG. It measures brain sensitivity by a pair of electrodes.
The EEG data was fed into a Brain-Computer Interface to be analysed and then into GAN (Generative Adversarial Neural Network) to be transformed into human models.
The images of new human models were again shown to the same participants, and the results accurately matched 80% of their preferences. Indicating that a computer can produce virtual-humans which real humans find attractive.
The new study on attractiveness combines computer science and psychology making it very unique. It’ll also help computers in understanding human emotions a bit better.
“If this is possible in something that is as personal and subjective as attractiveness, we may also be able to look into other cognitive functions such as perception and decision-making. Potentially, we might gear the device towards identifying stereotypes or implicit bias and better understand individual differences,” concluded Spapé.
Doomers always warn me about a future where AI takes over. I always showcase them a tomorrow where machines live together with humans and make their lives better. With every iteration of our “Adventure of AI” series, I become ever more hopeful for the future.