Advances in technology are changing the computer industry and everyday life
Just five seconds after Abu was asked to draw a picture of a cow returning to the farm, the image was provided with the answer, “This is it.”
Abu is not a human being, but an artificial intelligence (AI) based robot, so the artwork was really impressive.
In addition to painting, the robots can write reports, make video clips, and perform any task that a human can do on a computer, as directed by a member of the user group.
Thanks to advances in computer algorithms, the development of AI has accelerated rapidly in recent years, acquiring skills previously thought to be the only human capabilities.
Zhou Xiao, an assistant professor at the Gaoling College of Artificial Intelligence at Renmin University of China, gave several examples of AI that can be used in everyday life, such as competing with humans in games such as Go.
The game was developed by AI company DeepMind in October 2015 and features search algorithms, deep neural networks, and reinforcement learning (the science of decision making).
AI can also recognize a car’s registration plate number to help police search for suspects. For example, in 2016, AI could correctly recognize 94 percent of car license plates, but still required a police officer to interact with a computer screen. By 2020, AI will be able to accurately recognize him in 99% of car numbers, obviating the need for humans. Police have also benefited from the use of robotic dogs.
A recently announced AI-related technology, ChatGPT (Generative Pretrained Transformer), uses natural language transformation algorithms to enable it to “understand” human language.
“The way this algorithm works is like finding a point in space,” said Wang Jianshuo, founder and CEO of Baixing AI, a company that builds the basic infrastructure of a world where bots talk to bots.
“In human language, we may need thousands of parameters to describe an object. For example, to define an apple, we need ‘edible’, ‘fruit’, ‘green’, or ‘red’, etc. parameters are required and grown from it. “It’s a tree, usually less than 10 centimeters in diameter. The more parameters you define, the more accurately the AI can find the right point.”
Wang cited apples and bananas as two examples. Neither humans nor AI could distinguish between parameters such as “edible”, “sweet” and “fruit”.
But adding parameters that refer to shapes such as ‘long’ or ’round’ achieves such a distinction. By using parameters related to colors such as “red”, “green”, and “yellow”, it is possible to make more reliable judgments.
“That’s also how GPT works. Most of the time, AI needs thousands of parameters to define an object like humans do. We just don’t realize it,” Wang said. .
ChatGPT and related technologies are based on GPT-3 technology, which is characterized by excellent understanding of human language.
Unlike other forms of AI that can only receive commands from certain types of sentences, GPT-3 analyzes sentences to get their true meaning as humans do, and commands it to speak to a friend. to be able to run
However, this also means that the AI may misunderstand if given instructions without sufficient clarity and information.
The reporter twice asked the domestic app BaixingAI to draw an image titled “Crane”. But the app initially drew birds flying through mountains and rivers before drawing machines working on construction sites.
In an academic paper on ChatGPT’s potential, prospects and limitations, Zhang Junping, professor of computer technology at Shanghai’s Fudan University, emphasized that the amount of data used by language models is very important for AI such as GPT.
“ChatGPT generates multiple variations of responses that match the user’s intent. ,” Zhang said in his essay.
AI has revolutionized not only the computer industry, but also everyday life. Over the last five years, the gaming industry has made great technological advances, allowing people to hone their skills by playing games like chess and Go against smart AI.
Image recognition AI has evolved, making it easier to unlock doors and make payments. GPT is also evolving rapidly, and as a result could change the way people interact with computers.
Advances in GPT and the emergence of GPT4 have fundamentally changed the way humans and computers interact, said Chen Jing, a researcher at the Fengyun Institute for Science and Technology Strategy in Beijing.
Computers typically receive instructions from humans primarily through the keyboard or mouse, with users clicking icons or typing commands. But as speech recognition and facial recognition AI technology became more mature in the 2010s, Chen said computers were able to receive new forms of input, such as voice.
The GPT4 technology, announced by research firm OpenAI on March 15, takes this process a step further. The technology can recognize images more accurately and reliably, so it can understand humans more efficiently.
Chen said the technology could allow humans to command machines through gestures. For example, smart home appliances are already on the market that can give verbal instructions. In the future, a human will be able to give such instructions simply by waving his arm, and the movement will be captured by a camera. Such technology exists today, but is not yet fully developed.
“Imagine leaving home and going to work, and you have to turn off the lights and lock the door,” Chen said. “With GPT-4, you just wave at the front door camera and the AI will understand you, turn off all unnecessary appliances, close and lock the door.
“When you get home, all you have to do is smile at the camera and everything will ‘wake up’. is able to understand your gestures” GPT-4. ”
Wang said one of the biggest changes AI could bring to humans is the way we live.
“In the past, it was said that AI could only do repetitive tasks like copy and paste, but as AI advances, such things are rapidly becoming obsolete. It’s creative as you can see in the images it draws and it also helps us humans save time and energy and be more creative.”
Not everyone is happy about the progress made by AI, and such misfortune seems to be related to one’s profession.
41-year-old painter Xi Li (pseudonym) said, “Paintings made by AI have no soul. This technology can help humans, but it will never replace human hands.” Told.
“We don’t call the work created by the AI a drawing. The AI is just collecting material online and putting it together.”
One science fiction writer, who requested anonymity, said AI could never write as well as professional writers. “There’s no metric to measure the quality of an essay, but I don’t think AI-written essays are true,” he says.
Qu Xiaobo, a professor at the Institute of Population and Labor Economics and deputy director of the Center for Talents at the Chinese Academy of Social Sciences, said some people are worried AI will take their jobs.
“It makes sense. Even we professors are worried that they will be replaced by AI professors,” he said.
Q remains positive about the prospects for employment in the AI era. “While AI will reduce jobs, it will also create new jobs,” he says.
“The general impact on the market will depend on how many jobs will be lost and how many new ones will be created. Clearly, AI will improve the total factor productivity (TFP) of society. This means, in theory, that AI will create more jobs than it replaces.”
If you search for “AI” on multiple job sites in Japan, you will find new jobs related to AI such as “consultant”, “engineer”, and “tutor”. In Beijing, these jobs span pages and offer salaries ranging from RMB 15,000 to RMB 30,000 ($2,182 to $4,365). Last year, the average monthly salary for such jobs was about 15,000 yuan.
Qu said these emerging jobs are mostly concerned with working with AI, or what he calls “work for human-AI interaction.”
The rapid development of AI will create new jobs, Wang said, adding that every major advance in technology creates new job opportunities.
“The invention of the automobile gave rise to the driver, and the invention of the airplane gave rise to the pilot. Along with progress such as productivity improvements, the standard of living has generally risen. The faster AI progresses, the easier life will be. will be,’ said he.
However, maintaining a job remains a challenge for many who do not have the skills needed for their new profession.
Qu said there are “structural inconsistencies” in the job market, adding that this is not unique to AI but has been observed throughout human history as technology continues to advance. The difficulties graduates currently face in finding employment are part of such “structural disparities,” Qu added.
Solving this problem requires individuals to keep learning new things and gaining experience and skills. Ku also emphasized the role of the state in providing better training to employees so they can keep up with the changing job market.
Fudan University’s Zhang said AI is still far from being developed enough to replace humans.
When it comes to the future of AI, everyone interviewed for this article sees hope.
Zhou said GPT-4 did not perform well in all tests, but it revolutionized the AI industry by making it able to understand human language.
“In essence, AI was invented by humans to help humans,” added Zhou.
Yang Xiaoheng contributed to this story.