"Liu Chunhe just has cerebral palsy, not a fool."

After watching "Little Me", I am very pleased that film and television works can face the real situation of people with disabilities. Try to educate the public about their objective situations and needs, instead of blindly labeling them as “tragic” and “inspirational”. I believe that a movie can break these stereotypes because it is the aggregation and condensation of countless voices.

As a science and technology writer, one of the most important and most honorable jobs is to educate everyone about science and technology accessibility. After discussing many technology accessibility projects and interviewing many relevant users, we often feel like this: technology is developing at a rapid pace, and at the same time it is providing support for medical, rehabilitation, and accessibility undertakings at an extremely high speed. However, these achievements are often unknown to friends in need, and the public is even more unfamiliar with them.

We often encounter situations where a problem that is considered to be a lifelong problem for an individual may suddenly have a technological solution. This method may still be in the laboratory in the first half of the year, but it will be applied on a large scale in the second half of the year.

Therefore, it is always necessary to discuss and describe the latest achievements in technological accessibility. This job has no expiration date and must be on active duty for life.

For example, the past two years have been a period of intensive outbreak of large AI models. This technology trend has brought many upgrades to the cause of accessibility. In this article, we will list some of the latest developments in AI+ accessibility.

The theme of these stories is how a small me can have a big AI.

Let’s first talk about the cerebral palsy Liu Chunhe faces.

Cerebral palsy is a disease that must be discovered and treated as early as possible. At the same time, if children gradually discover symptoms of cerebral palsy as they grow up, it will also bring great challenges to parents and school education. Therefore, infants and young children are the best time to observe and identify cerebral palsy. However, the clinical judgment method is often based on observing the subtle movements and reactions of infants and young children. This requires relying on the doctor's observation experience, and there is a huge probability of misdiagnosis and missed diagnosis. At the same time, observing and identifying children with cerebral palsy also requires very rigorous training, which results in very few doctors with relevant qualifications. In areas with relatively weak medical conditions, it is difficult for children and their families to obtain similar support.

But capturing dynamic models and making classification judgments happens to be the specialty of AI. Many years ago, academic circles have begun attempts to use machine vision to identify early-stage cerebral palsy.

Today, with the support of large models, the accuracy and completion efficiency of AI cerebral palsy recognition have been significantly improved. This technology has also moved from the laboratory stage to the clinic. Some tertiary hospitals have begun to purchase early AI prediction assistance systems for cerebral palsy. It is believed that AI cerebral palsy recognition will be popularized in primary hospitals in the near future, solving the key problem of lack of identification ability of children with cerebral palsy in primary medical systems caused by insufficient doctors.

When children come into this world, they should first feel the kindness of technology. Beyond recognition of cerebral palsy, AI technology is promoting the upgrade and iteration of smart children's health care systems from multiple angles, allowing health problems that may plague children throughout their lives to be discovered, corrected and treated during infancy and early childhood.

In the further future, the possibility of using a combination of AI + brain-computer interface to help paralyzed patients represented by cerebral palsy regain their senses and regain their ability to move has also been repeatedly polished in the laboratory.

Health will be the true promised land of AI.

Not long ago, I came across a short video. The hearing-impaired girl in the video used her mobile phone to repair her voice and calmly completed her daily life, including ordering at a coffee shop. How wonderful the changes brought about by technology are perfectly explained at this moment.

In China alone, there are more than 27 million hearing-impaired people and more than 200 million elderly people over the age of 65. Many of them have speech impairment due to hearing impairment that affects their speech function. Devices such as hearing aids can help them hear, but they cannot help them produce sounds accurately. This makes communication still difficult for the hearing-impaired, speech-impaired, and many elderly people.

This year, Huawei launched the Xiaoyi sound repair function in HarmonyOS NEXT. This technology can use Hongmeng’s native intelligence to analyze, repair and optimize the voice input by the user in real time, thereby outputting clear and highly intelligible sound. At the same time, Xiaoyi's sound repair can also preserve the user's tone as much as possible and enhance the user's experience.

With the support of large speech recognition models and large speech synthesis models, Xiaoyi's voice repair function can meet the repair needs of level four and level three speech disorders, and achieves a threshold-free user experience. In order to meet the practicality of this function, the Huawei team invited 3,700 speech-impaired users to participate in the test, and has achieved an improvement of more than 80% in understandability.

Mobile phones are the closest technological products to us, and they will always be the front line of technological accessibility. Today, AI image recognition and screen reading software have become necessities in the lives of the visually impaired. The barrier-free operation of mainstream software is also continuing to evolve. We once discussed accessibility upgrades for mobile phones with a visually impaired developer friend. He mentioned two trends. One is that mobile phone accessibility was once completely dominated by Apple. This also results in many hearing-impaired and visually impaired friends having to buy expensive Apple phones. But now, as domestic mobile phone brands such as Huawei and vivo continue to pay attention to barrier-free experience, Apple's dominance has been completely changed. Even Apple's upgrades in this area have become conservative. Second, a large number of functions that can help people with disabilities are still in specific applications, but the user experience of entering specific applications to call accessibility functions is very poor. Therefore, developers of related functions are now cooperating more with first-party mobile phone brands to integrate related functions into the system's own functions or ports such as negative one screen and quick applications. It is best to use voice assistants, shortcut keys, etc. exhale.

Mobile phone manufacturers, developers, AI technicians, andWith the joint efforts of my speech-impaired friends, the sound of a sentence was finally restored accurately.

Perhaps the next person who needs to use this feature will know that this is the voice that the world has tried its best to help them make.

Everyone knows that the number one job for the visually impaired is as a masseur. But what is the second largest?

Many people don’t know that the answer is programmers. Compared with other jobs, software programming requires less vision and has higher overall benefits and a larger talent gap. Therefore, with the promotion of the country and relevant associations, programming is becoming a new employment option for the visually impaired.

Even so, programming work is still full of hardships for the visually impaired. Screen reading software can help solve some problems, but needs such as checking code and locating information still plague the work of the visually impaired. In addition, screen reading software cannot read the information on the picture, which also brings a lot of inconvenience to visually impaired programmers.

In order to help the visually impaired bridge the gap in their code journey, not long ago, Baidu Intelligent Code Assistant Wenxin Kuai Code joined forces with the Information Accessibility Research Association to launch the "Wenxin Kuai Code Accessible Edition".

This system can convert the code into fluent natural language, thereby helping the visually impaired to quickly locate the code and complete functions such as search and review. In addition, Wenxin Quick Code Accessible Edition can also provide a series of convenient instructions to help users locate the code segments that need to be modified. The AIGC capability provided by Wenxin Big Model can automatically generate executable code after the user inputs a natural language description, thereby greatly improving the programming efficiency of the visually impaired and allowing them to move towards a decent, high-income, and endless world of possibilities. sex jobs.

AIGC code generation is originally one of the core capabilities brought by large models, and this capability has a natural integration point with barrier-free development. Naturally, Wenxin Quick Code has become the first programming tool in China to support barrier-free development, and I believe this is just the beginning.

It is not difficult to see from these stories that large AI models are being deeply integrated with industries, terminals, and development tools, thereby bringing new opportunities to people with disabilities in different fields such as medical rehabilitation, daily life, work and entrepreneurship, etc. Ability pivot.

Big AI may be just a small tool. But it can make an unfair fate a little fairer and make life in the turbulence a little calmer.

It makes us, little ones, finally believe that the world can become a better place.