Author ∠ Enhao
Released ∠ Numerical Intelligence Sharp Angle
Guide
To participate in the formulation of the rules, you must first go to the card table before you can have a voice and have a ticket to the global competition.
At the Zhongguancun Forum on May 26, Baidu founder Robin Li delivered a keynote speech entitled "Big Model Changes the World".
In Li Yanhong's view, big models have changed artificial intelligence, and big models are about to change the world; In 10 years, 50% of the world's jobs will be prompt word projects; Every product in the era of big models deserves to be redone, and Baidu wants to be the first company to redo all products.
Robin Li: Big models change the world
Looking back at the release of Baidu Wenxin's words in March, Li Yanhong once said, "When a machine has the ability to understand, express, and reason, it is very similar to a "human", but the machine's memory and calculation ability are far stronger than that of a human. With this ability, machines will indeed change every industry. The rapid evolution of artificial intelligence technology will bring us a very different world. So what Baidu really believes in is to use technology to make complex worlds simpler. ”
At the Zhongguancun Forum, Li Yanhong further said, "We are at a new starting point, this is a new era of artificial intelligence with big models as the core, big models change artificial intelligence, big models will change the world." ”
> big model redefines artificial intelligence
First, the big model redefines human-computer interaction. The emergence of large models allows us to interact with computers in natural language, and natural language human-computer interaction will bring a revolution in prompt words, that is, future applications are achieved by mobilizing native AI applications through natural language prompt words. This means that your future compensation level will depend on how well your prompts are written, not whether your code is written well. Robin Li predicts that in 10 years, 50% of the world's work will be prompt word projects.
Second, the big model will redefine marketing and customer service. Each customer can also have a dedicated assistant who knows everything 24/7, ×and whoever has the best way to communicate with the customer will have the customer.
Third, large models will give birth to AI native applications. DoNotPay, for example, is an app that uses AI to help people fight lawsuits and write legal documents, and AI helps you get back money you shouldn't have paid. Jasper is an app that helps businesses and individuals write marketing promotion copy through Al. Baidu is also reimagining all of its products, services and workflows with AI-native thinking.
> big models change artificial intelligence
Big computing power, big models, and big data have led to the "emergence of intelligence". In the past, artificial intelligence was that if you want the machine to learn any skills, you will teach it what skills, it is possible to teach, but it will not be taught, and the emergence of intelligence is that if you have not taught skills, you may also be able to.
The development direction of artificial intelligence has also changed from discriminative to generative, search engine is a typical discriminatory, generative is AI-based content creation, such as writing reports, drawing posters, etc., generative AI, will greatly improve work efficiency. Some research institutions believe that in the next 10 years, the efficiency of knowledge workers can be increased by 4 times.
From redefining artificial intelligence to changing artificial intelligence, the value of big models lies in the ability to bring more advanced decision-making and execution capabilities to the industry, improve work efficiency from the perspective of quantity and quality, reshape the life and work of human society, and ultimately change the world.
Big models change the rules of the game, so you have to sit at the poker table first
After the explosion of Open AI's ChatGPT, the industry came to the consensus that large models are miracles, which is why participants are crazy to buy GPUs and continuously increase the amount of data. However, as everyone understands the big model more and more, it is found that the premise of the GPT series violent aesthetics is a strong engineering architecture capability, that is, the coordination ability of the four-layer architecture such as chip layer, framework layer, model layer, and application layer, which is the most perfect participant form at present.
Baidu is one of the participants in this form, Baidu has a first-mover advantage, but the advantage is not obvious, Baidu uses large models to reconstruct the business at the same time, other manufacturers can also use large models to impact Baidu's business, the first is search.
> large models leverage business boundaries
According to data from website research institute StatCounter, in April 2023, Microsoft's Bing Search share in the Chinese mainland desktop search market reached a record high of 37.4%, replacing Baidu as the largest desktop search engine in China. Baidu Search's market share fell to 27.01%, a 10 percentage point gap with Bing.
Although the new version of the Bing search engine and Edge browser integrated with ChatGPT is sought after by users, this share data should not be accurate and quite different from the data of other channels. However, we should be able to see the impact of large models on search applications, and as the gap between different large models gradually widens, its impact on search applications will become more and more obvious.
In addition to Microsoft's Bing search, other domestic search players are also integrating large models into search. For example, 360's large model 360 intelligent brain, internal testing shows that the answer to "the current employment environment of Chinese college graduates, what structural problems exist, how to solve these problems" is remarkable, the content expression, understanding of the context, etc. are acceptable.
While Wen Xin empowers Baidu's search business, the big model also empowers other search applications. From another perspective, the large model can strengthen the business boundary while leveraging the business boundary. It's clear that big models are a game changer.
> on top of search, outside of search, Baidu needs to sit at the card table
Since the release of Wenxin ERNIE 1.0 in March 2019, almost all Baidu products have been using the flying paddle Wenxin model. At the Baidu World Conference in July 2022, Robin Li said, "AIGC is a new way of content production after PGC and UGC. It will not only improve the efficiency of content production, but also create content with unique value and independent perspective. ”
In addition to applying the large model to the landing and verification of its own products, it is also used in external scenarios in various industries. Wenxin model has been empowered by Feijiao open source platform and Baidu intelligent cloud to various industries such as industry, energy, finance, communications, media, and education. For example, in the application of the insurance field, the intelligent parsing ability of the Wenxin large model can increase the efficiency of text processing by 30 times; In the application of human resources, the Wenxin model can realize the intelligent classification of candidate information, and the model recognition accuracy reaches 99%; In the medical field, the Wenxin model has shortened the examination time of each medical record from 30 minutes to seconds.
In May 2022, Baidu and State Grid released a knowledge-enhanced power industry large model, which can achieve good results in a shorter training time with less computing power, and the comprehensive index of the model reached 91.18%, an increase of 11.38% compared with the traditional method.
Therefore, for enterprises participating in the big model competition, it is natural to reconstruct their own business, and the icing on the cake is the space outside their own business.
brief summary
At present, the scale of large models is getting bigger and bigger, the knowledge compression ability of large models is getting stronger and stronger, and the capability boundary of large models is constantly expanding. For individuals, the big model will make some jobs disappear and create more new jobs, for enterprises, the big model can empower the business, make the original business go to a higher level, and also be eliminated by the market.
In the process of rapid development of artificial intelligence technology, there may indeed be a direction that is unfavorable to human beings. To prevent loss of control, countries with advanced AI technology need to work together to formulate rules from the perspective of a community with a shared future for mankind. To participate in the formulation of the rules, you must first go to the card table before you can have a voice and have a ticket to the global competition.
———— Follow Us SmartCompanion ————