The Wall Street Journal wrote today that the concept of artificial intelligence has been heated up, but most of the artificial intelligence at this stage is a "pre-trained" system, which is only equivalent to the jellyfish in the evolution stage of bio-intelligence, far from being comparable. The level of brain wisdom. Artificial intelligence still has a long way to go.
The following is the main content of the article:
The era of artificial intelligence has arrived, but creating systems that can be used to solve multiple problems can take decades.
Startups that claim to use artificial intelligence (AI) are attracting record levels of investment. Large-scale technology companies have gone all out, and the talents of the entire department of colleges and universities have been drained. Since 2011, nearly 140 artificial intelligence companies have been acquired, including 40 this year.
AI is entering our daily lives, such as speech recognition in smart devices, image recognition in Facebook and Google accounts.
Now, Google's parent company, Alphabet Inc., has teamed up with companies such as Amazon and Microsoft to lend some of their smarter offerings to other companies. Want your app or gadget to respond to voice commands and answer questions with your own "sound"? These services can do it. Need a transcription dialogue for analysis? This type of new service can do this, as well as many other things, from facial recognition to identifying the content in an image.
However, it is not easy to extract quantifiable functions from these new artificial intelligence toys. “Everyone thinks that AI Spring is coming, will AI's summer be far behind? But I think that's 10 years later,†said Angela Bassa, who works on energy intelligence software company EnerNOC. Scientific team leader.
Before switching to the new role, Ms. Barcelona led a team of EnerNOC to use artificial intelligence techniques such as machine learning and deep learning to turn large amounts of data into computer programs to "train" them. But the company found customers more interested in analytics than the incremental value that complex AI-driven algorithms can provide.
Barcelona said that artificial intelligence requires three things, but most companies can't meet the demand. One is sufficient data, like Facebook, Amazon, Alphabet, General Electric and other companies are collecting large amounts of data, but this is not the case.
The second problem is that a small difference can prove that creating an artificial intelligence system is worth the money. If an artificial intelligence system can increase the fraud detection rate of a credit card company by 1%, it could be worth tens of millions of dollars. However, for a medium-sized manufacturer that produces a different product, a 1% increase in productivity on a particular production line may not be as cost-effective as hiring six high-paying engineers.
This led to a third scarcity: insufficient staff to build the system. The talent struggle of artificial intelligence is driving the rise in costs. “Perhaps only 5,000 people in the world can create a machine learning system that helps people save money, provided that they increase their investment,†Ms. Barcelona said.
This does not mean that artificial intelligence is useless to the enterprise. But it does show that the concept of artificial intelligence has been overheated. Using artificial intelligence to create systems that can be used to solve a variety of problems, rather than just putting them into a narrow application area, can take decades. The system must be built and trained, just like educating children, this is a lot of work.
Most of the artificial intelligence devices available today are "pre-trained" systems built by large companies like Google, Amazon and Microsoft, reflecting the data of these companies. These companies have billions of images that provide commercial image recognition services to others. Amazon has edited a lot of spoken language through Alexa's personal assistants, so it can provide services for others to process spoken and replies to conversations.
Some startups are beginning to offer a wider range of artificial intelligence systems that do not require machine learning experts or pre-trained systems like Google. Israel's n-Join sells a small box to the manufacturer, collects data from the machine on the line, and then uses machine learning to locate the abnormal deviation that could lead to failure.
Strauss Group Ltd. is one of Israel's largest dairy manufacturers and an early customer of n-Join, the company's senior technician Guy Tsur said, n-Join The key to the device is that it does not need to know the type of pipeline being connected, or which sensor is being used to obtain measurement data. It is just looking for correlations that indicate that the manufacturing process is different from normal. Then remind the human supervisor that the latter uses his own experience and judgment to identify a particular problem.
One thing that a savvy reader has pointed out is that, to date, none of the above-mentioned artificial intelligence can be compared to the world of science fiction machines. As a neuroscientist who studies invertebrates, recalling his short experience, I want to say that today's artificial intelligence is only equivalent to the jellyfish in the evolution stage of bio-intelligence. The real brain and even the real intelligence of the future are still out of reach, far beyond any reasonable prediction.
Or, as Andrew Ng, the chief scientist of China's search giant Baidu, and artificial intelligence expert, has raised the fear that artificial intelligence is controlled by some evil intelligent creature, it is equivalent to worrying about overpopulation on Mars, and the two are equally unreasonable. (Small)
new smart watch,smart watch bulk, smart watch wholesale,smart watch with logo
Shenzhen Konchang Electronic Technology Co.,Ltd , https://www.konchang.com