VOA Special English
5 Interesting AI Tools Announced in 2022

    2022-12-28

    Throughout the past year, we followed some interesting new inventions in the field of artificial intelligence (AI). Here is a look back at five notable AI developments in 2022.

    Meta's speech-to-speech translation

    Facebook's parent Meta said it had built a technology tool designed to directly translate spoken speech from one language to another.

    This image of an ancient Greek text represents a restoration of the inscription by the Ithaca tool, developed by Alphabet's DeepMind, along with researchers from the University of Oxford, Ca' Foscari University of Venice and Athens University of Economics
    This image of an ancient Greek text represents a restoration of the inscription by the Ithaca tool, developed by Alphabet's DeepMind, along with researchers from the University of Oxford, Ca' Foscari University of Venice and Athens University of Economics

    The AI-powered tool was built to translate between English and the Hokkien language. The company said the project required unusual development methods. This is because Hokkien is mainly a spoken language and does not have a widely used written form.

    Meta said it trained its AI models on written text examples from Mandarin Chinese, which is similar to Hokkien. In addition, developers used an encoding tool designed to compare spoken Hokkien to similar English text. The team also worked closely with Hokkien speakers to ensure correct results.

    Meta said it aims to use the same methods used for Hokkien to create speech-to-speech translation systems for many more languages in the future.

    Method to identify Parkinson's Disease

    Researchers announced a new AI method to identify Parkinson's disease – a neurological disease that harms the brain.

    The system works by measuring a person's breathing patterns during sleep. A study describing the method said it can also measure the severity of disease and record its progression.

    The researchers noted that early signs of Parkinson's disease are not easy to discover. Many signs, or symptoms, only appear years after the disease sets in.

    But the team said its experiments showed that with just one night of sleep, the AI system was able to correctly identify Parkinson's up to 86 percent of the time. With 12 nights of data, the rate went up to 95 percent.

    Tool to interpret pig emotions

    Researchers announced they had created a technology tool that uses pig sounds to interpret different emotions.

    The tool is based on thousands of recordings collected from more than 400 pigs throughout their lives. The scientists developed an AI-driven algorithm to identify a series of emotions the animals could be experiencing.

    The study centered on more than 7,000 audio recordings of pigs in different situations. By training the algorithm to recognize the sounds in both good and bad situations, the team said the tool was able to identify the correct emotion 92 percent of the time.

    Researchers said the tool is expected to lead to further systems farmers can use to improve the productivity and well-being of their animals.

    System to identify and predict smells

    Developers said they had created a technology tool that can identify and re-create different smells.

    The scientists built on past research to produce a brain-like technology model to create a "principal odor map" to identify molecules related to smells. The team said such a system can be used across many different fields, including the healthcare, food and fragrance industries.

    The researchers reported that the tool can be used to help fight one of the world's biggest health problems – diseases spread through insects. They said the system can measure how effective a molecule is at keeping mosquitos away from humans.

    Tool to fill in missing words in ancient writings

    Researchers said they developed an AI system to help fill in missing words in ancient writings. The tool, called Ithaca, is designed to help historians repair the writings and identify when and where they were written.

    Researchers – led by Alphabet's AI company DeepMind – created the tool. They said the system is "the first deep neural network" to identify missing text in damaged writings, or inscriptions.

    The team said that when historians work on their own, the success rate for repairing damaged inscriptions is about 25 percent. But when humans teamed up with Ithaca to assist in their work, the success rate jumped to 72 percent.

    I'm Bryan Lynn.

    Bryan Lynn wrote this story for VOA Learning English.

    ______________________________________________________________

    Words in This Story

    artificial intelligence – n. the development of computer systems with the ability to perform work that normally requires human intelligence

    translate – v. to change from one language to another

    encode – v. to use a word or phrase in a foreign language in the correct way

    interpret v. to explain or decide what something means

    algorithm – n. a set of steps that are followed in order to solve a mathematical problem or to complete a computer process

    odor – n. a smell, often one that is unpleasant

    fragrance – n. a sweet or pleasant smell

    text – n. written words that appear in a document, book, magazine, etc.