Making web pages, writing copy, pattern design, code development... Can GPT-4 replace me at work?｜Base Construction

In the Iron Man movie series, there is an impressive intelligent butler named Jarvis. It has its own thoughts and can handle almost any affairs, which impressed many fans.

Stills from the movie "Avengers: Age of Ultron", the yellow one is the smart butler "Jarvis" (program)

Source: Movie screenshots

The term "artificial intelligence" was first coined by John McCarthy in 1956, with the assumption that every aspect of (human) learning or any other feature of intelligence can in principle be described so precisely that it can be simulated by a machine.

In November 2022, an American artificial intelligence research company launched Chat GPT. Once it was released, it had 100 million monthly active users within three months of its launch, an unprecedented scale of user development. On March 15, the "upgraded version" GPT-4 was launched, and users who have come into contact with it feel that the artificial intelligence in the movie has come to reality. Why is it so highly rated?

Source: Internet

1

What is GPT-4?

GTP-4 is a super-multimodal model based on artificial intelligence technology. It can input text or images and output text. It learns and trains based on a large amount of data, and acquires the ability to generate human-like reactions or answers to user questions. The biggest feeling of users is that they feel like they are chatting with a "human".

Compared with previous products, GPT-4 can perform visual analysis on images , and the database content has been updated and expanded, with significantly fewer errors or contradictory feedback. It can also analyze and create more large, complex, and abstract content, and is applicable to a wider range of scenarios.

GPT-4 can analyze images

Source: Internet

GPT-3.5 and GPT-4 answer the same question, and it can be considered that it can think

Source: Internet

GPT-4's knowledge covers almost all known subjects, supports the generation and processing of up to 32,768 tags (about 25,000 text words), can analyze or create long texts, and even rank among the top in difficult tests such as the judicial examination, Olympic competition, and SAT examinations. It is at the level of being admitted to Harvard.

Source: openai.com

Many netizens joked that they were about to lose their jobs because GPT-4 was too powerful.

Source: Weibo

We already have AI voice intelligent audio products around us, such as Xiaodu, Tmall Genie, Siri, etc. They often grab keywords from the Internet for feedback based on user questions, and may fail to recognize or answer inaccurately; and more functions are reflected in the intelligent control of furniture and appliances.

Domestic intelligent voice assistant

Source: Internet

GPT-4 first learns and trains on massive amounts of data (perhaps including all the knowledge currently known to humans), and then analyzes, integrates, and gives feedback on the problem; it is even constantly learning new knowledge, upgrading and iterating, and expanding its "brain capacity." Therefore, it can follow complex instructions in natural language and generate technical or creative content. This allows GPT-4 to better identify content than other products, reduce misreading and misjudgment, and provide accurate responses.

This is the question AI researcher Adjunct Professor Oren Etzioni asked when he first tried GPT-4.

Source: Internet

2

The development of artificial intelligence is not smooth.

Artificial intelligence seems to have achieved great results, but its development path has been bumpy.

After the concept of "artificial intelligence" was proposed, it ushered in the first major development. In 1959, Engelberger developed the first robot in human history, named "Unimate". The following year, the first artificial intelligence programming language was also born. After that, the intelligent manipulator, the robot Shakey, and the first humanoid robot based on intelligent software, Wabot1, were all born one after another.

Robot wabot1

Source: Internet

After 1970, the development of artificial intelligence entered a trough period because the development of hardware and technology could not keep up with the vision of scientists.

From 1980 to 1986, the emergence of the XCON expert system, Hopfield network, and BP network significantly improved the hardware computing power, so the artificial intelligence industry entered a development boom again. During this period , the intelligent robot wWabot2 that can communicate with people appeared. However, since hardware can only solve part of the development problems, the development of artificial intelligence entered a trough after this small boom.

This trough was broken by the improved version of the supercomputer "Deep Blue" released by IBM in 1997. It defeated Garry Kasparov, the world's number one chess player at the time. This victory proved the potential of artificial intelligence to the world.

Supercomputer "Deep Blue"

Source: Baidu Encyclopedia

As artificial intelligence has developed to this day, we have to mention scientist Geoffrey Hinton, who is known as the godfather of AI. He once said, "Everything you can do, neural networks can eventually do. All industries will be affected by it." He has been studying the methods of using neural networks for machine learning, memory, perception, and symbol processing for more than 30 years, and has published more than 200 papers and dozens of patent applications in these fields, bringing the development of artificial intelligence to its current height.

Geoffrey Hinton

Source: China Intellectual Property News

In 2016, we witnessed “AlphaGo” defeating Go world champion Ke Jie.

Ke Jie plays against AlphaGo

Source: Internet

In 2017, Sophia, a robot developed by Hong Kong Hanson Robotics, became the first robot to acquire human identity. It is a female humanoid robot with rubber skin and can make more than 60 expressions.

Sophia the Robot

Source: Baidu Encyclopedia

The release of GPT-4 in 2023 is the result of scientists' hard work and efforts. As the most powerful AI system currently, its success comes from trillions of data inputs and learning, and the leapfrog development of natural language processing.

3

What will the future look like if it develops further?

After the release of GPT-4, Microsoft immediately announced that its Microsoft Bing had already integrated this technology, and would also integrate it into Office in the future. As soon as the news came out, many users began to look forward to the convenience and changes that the upgraded Office could bring to their work.

For GPT-4, making web pages, PPTs, writing copy, pattern design, and code development can all be done in seconds; intelligent robotic arms have become popular in the manufacturing industry, and only two or three people are needed to control the entire production line; controlling smart home appliances is even easier.

SAIC-GM Global3+ robot for assembly line

Source: Internet

Judging from its current development, GPT-4 already has several notable features and may achieve greater development in the future, especially in education, medical care, content creation, industry services, etc.

> > > > GPT-4 has a massive data reserve and is a huge resource library. It may develop into a new search method, which may have an impact on various current search engines and education and training industries in the future.

> > > >Because GPT-4 can analyze and judge information, better integrate and feedback information, and improve problem-solving ability and efficiency, it has strong advantages in content creation and analysis.

GPT-4’s ability to analyze copywriting

Source: Internet

> > > > GPT-4 can judge users' emotions through conversations, and in the future may be able to provide emotional value to humans, even conduct psychological counseling, and assist the development of the medical industry .

> > > > By combining with various other electronic products, it provides new possibilities for intelligent services in manufacturing and service industries.

Of course, the development of artificial intelligence is also a double-edged sword. Less than a month after the release of GPT-4, more than a thousand scientific and technological workers publicly called for a "pause of giant GPT training." They have become wary of the rapid development of artificial intelligence.

Musk tweets

Source: See watermark

The famous Turing test is designed to determine whether a computer can think. So far, only one machine has passed it. A computer software named "Eugene Guzman" developed by a Russian team was put through the "Turing test" in the UK in 2014, and 33% of people thought it was a 13-year-old boy.

Hawking believes that "humans cannot compete with machines due to the limitations of slow biological evolution and will be replaced. The development of full artificial intelligence may lead to the end of mankind." Schmidt, former chairman of Google, believes that "robots are our friends." The development of artificial intelligence has infinite possibilities. Is a higher level of intelligence a new starting point for the progress of human civilization, or a dilemma for the evolution of human brain and physical strength? Perhaps we will see it in the near future.