How can artificial intelligence improve the lives of people with disabilities | Science and health

In Brazil, according to the 2010 census of the Brazilian Institute of Geography and Statistics (IBGE), the country had 45.6 million people with disabilities (or 23.9% of the population). ;time).

Luciane is blind and uses various applications on her smartphone to recognize images, product labels, payment cards and access printed documents.

With these accessibility solutions, she was able to discover the Wi-Fi pbadword alone at home and also when, for example, her computer was the object of. an update. Indeed, the current screen reading software for the blind does not have access to screens such as the Windows update.

The professor also presents technological resources to her students. "With each novelty that I present, their life becomes easier," he says.

Recently, a student from Luciane gained more autonomy through an actual reading app. "She was eager to be alone at home and able to pay for the kettle without the help of other people," she says.

Most of these solutions have been developed using "data-intensive" "learning" systems. From this information, they are able to identify patterns and make decisions with the least possible human intervention. This is what is called "machine learning".

However, speech-to-text transcription and real-time computing capabilities are just a few examples of artificial intelligence applications.

"The more we use [os recursos de inteligência artificial para pessoas com deficiência] the better they are," explains the professor. Among its applications, it uses TapTapSee to read product labels and Seeing AI to organize documents and diplomas in folders and recognize the texts of photos you receive via email or via social networks such as Facebook.

The teacher even developed a wooden stand to support books and printed materials to make photographs at right angles and with good lighting.

Seeing AI is a Microsoft initiative on artificial intelligence for blind and partially sighted people. The application uses computer vision and neural networks to identify the objects, colors, texts, scenes and even the physical characteristics and facial expressions of a person. For the moment, it is only available on iOS, the operating system used by Apple devices.

The application is only one of the accessibility actions developed by Microsoft. In May 2018, the company's general manager, Satya Nadella, announced an investment of $ 25 million (over $ 93 million) over five years in artificial intelligence projects for people with disabilities.

Developers, NGOs, academics, researchers, and inventors can submit AI-based accessibility projects for the AI ​​For Accessibility program until February 1, 2019. Those who have been approved will receive a contribution from the company for another level Applications are accepted at all times and to register, you must complete a form.

Facebook is another technology giant investing in artificial intelligence projects for people with disabilities. The social network uses algorithms to generate an automatic image description for the blind. The functionality is not perfect and does not replace a human description, but it works for a more accessible environment.

There are several methods of machine learning. What they have in common is that everyone needs people to teach them how to learn a particular task and do what they want. On Facebook, image recognition systems are supervised by people who look at the photos and tell the system what they contain.

According to Matthew King, an accessibility engineer on Facebook, although artificial intelligence has been in development for decades, the systems we have are still new. "Although they have improved rapidly in recent years, they still have many limitations," he said. Nevertheless, they represent a big step forward for the full participation of the blind in the online environment, according to the expert.

Luciane Molina, a university professor and badistive technology technician, agrees. She says that before, the social network did not interest her so much, because of the number of messages containing images with which she could not interact, because they did not know what it was .

Once one of his friends posted a picture and wrote in the caption: "Look at my new bracelet". But in reality, the image was referring to the bandage placed on the arm of the friend after a bruise. Luciane only noticed this fact by reading all the comments.

She recognizes that the descriptions are basic and do not replace an audio description, but she values ​​the resource. "The fact that Facebook's artificial intelligence has aggregated these image recognition features has represented an incredible gain in affordability," he says.

Recently, the company has also released the automatic description feature and written a manual description also on Instagram. However, some blind people blame the social network for not being easy to use. Asked by the BBC Brazil report, King, the Facebook engineer, said it was an ongoing improvement process.

"Adding photo descriptions was a very important step for Instagram because photos are the most popular type of social networking content, but we expect a lot more."

New Opportunities for Alternative Communication

Speech recognition and transcription of human language in real time is one of the most traditional features of artificial intelligence and is very useful for people suffering from certain types of disabilities. Alex Garcia is deafblind and was the first Brazilian with this disability to attend a university.

He has 1% vision and does not listen. Deafblindness is defined by the non-compensation of the senses. There are those who can hear and see with great difficulty.

Today, Alex is dedicated to the teaching and management of the Gaucho Association of Parents and Friends of the Deaf and Multidisciplinary (Agapasm). One of the applications he uses is called Once Touch Communicator (CTO), a communication tool for deafblind people with multiple resources.

Among them lies the saying. Although the main form of communication occurs when people write in your hand, sometimes this is not possible. In this case, Alex brings the microphone closer to the person's mouth so that she can talk. Then he can get the text on the screen in enlarged letters via the application.

"The cell phone helps me when the person does not know how to write.The application is therefore a valuable alternative, because it gives me peace of mind.If a support does not work, the other will work, "he says. He says being deafblind implies an extremely delicate and vulnerable social situation.

Although artificial intelligence resources are extremely prevalent in smartphones, initiatives have also been taken to bring this reality to the physical world. This is the case of the Wheelie, a device developed by a team of Brazilians who promise to order wheelchairs through artificial intelligence. The technology is currently being developed in partnership with Intel.

In the United States, the equipment is tested by 60 people with disabilities. According to Paulo Pinheiro, CEO of Hoobox Robotics, the company that develops this technology, the next step is to enter the Chinese market in 2019. In Brazil, the product is expected to arrive only in 2020.

Orcam My Eyes is another way to increase the independence of people with disabilities. He is able to read accurately the printed documents and is available in the 54 municipal libraries of the city of São Paulo.

Evolution of badistive technologies

According to Lúcia Miyake, a research and badistive technology specialist, badistive technologies for people with disabilities have always existed and have practically accompanied revolutions industrial. "There has been a breakthrough when the era of steam engines has moved into the era of electricity, during which electronic equipment has emerged, such as the electric wheelchair and the # 39; recorder ".

But the revolution has even come with the possibility of programming electronic equipment. "Together, came the importance of data and information, which I consider the beginning of artificial technology," he said. We are now entering a new phase called Industry 4.0. In this document, the data is in the cloud and commands are sent remotely.

"The most interesting thing is that the user can count on help without knowing where the orders come from.In this respect, the disabled person can work or study remotely with any access to information. , in addition to letting the controls of the electronic devices programmed a coffee, open the curtain, etc.), "says Miyake.

The main limitations of the sector are the lack of researchers and resources or investments in the field of research in badistive technologies. In addition, if the artificial intelligence product is unique for people with disabilities, the cost will be higher because of the low demand for manufacturing.