Zhou Hongyi: If you can’t catch this bus, you will easily be eliminated! Zhang Chaoyang added: If you don’t have the ability, don’t enter randomly!

​ Author | Qianshan

Reviewer | Yun Zhao

A few days ago, Zhou Hongyi expressed his position on the topic of ChatGPT in a live broadcast: "Nowadays, many The company claims to be building ChatGPT, so there must be bubbles in it, and bubbles are not a bad thing." The right strategy is to choose a big track that remains unchanged for a long time, and then adjust tactics according to market changes.

Looking back at the nodes of artificial intelligence development over the years, Zhou Hongyi believes that the reason why ChatGPT became so popular is largely due to the successful implementation of scene exit.

"AI has disappointed everyone before, because although there have been breakthroughs in face recognition, the scenarios are very limited; although the advent of AlphaGo is exciting, it is more like a game. .ChatGPT fully understands human language. If many companies fail to take this bus, they will easily be eliminated.”

Is this really the case? As Zhang Chaoyang later pointed out: The success of ChatGPT is the accumulation of computing power. Companies without the ability should not jump into the ChatGPT trend, as resources will be consumed in vain.

Is ChatGPT a temporary hype effect or does it have long-term development potential? What opportunities and challenges are faced by the giants vying for the top spot? Can it really bring about disruptive changes in the search industry? What are the hidden concerns in the prospects of ChatGPT? When the group frenzy caused by ChatGPT is on the rise, maybe we should calm down and deal with these issues.

1. The curtain has opened and the storm is about to come

The competition triggered by ChatGPT has been very popular these two days. The appearance of the appearance.

Microsoft held a closed-door media conference and officially announced the launch of a new Bing search service integrated with ChatGPT, as well as an Edge browser integrated with AI assistance; Google followed An event was held in Paris to demonstrate the new features of ChatGPT’s competitor Bard.

Since the chatbot ChatGPT launched by OpenAI became popular around the world, Google has been under high pressure. Because judging from the performance of ChatGPT, it is very likely to bring disruptive changes to the search industry.

And Google is indeed anxious. Faced with the rising popularity of ChatGPT and Microsoft’s high-profile bet, Google CEO Sundar Pichai first publicly stated I wrote an article to build momentum for the upcoming conversational AI service Bard, and later held a demonstration event with the intention of making Bard's debut a blockbuster.

Unfortunately, the result was not ideal. Bard made a mistake when answering a question, which raised concerns about its accuracy and made investors worried that it would be at a disadvantage in future competition, causing Google to lose $105.6 billion in market value overnight.

The search industry, which has been silent for a long time, has become undercurrent due to the "intrusion" of ChatGPT. The successive statements of a number of giants made this chaotic war more suspenseful.

The curtain has been drawn, Microsoft has taken the lead, Google has stumbled, the drama is full of tension, and the development is confusing, but everyone inside and outside the venue knows that a storm is coming.

2. Google "gets up early and goes to market late" Microsoft has the ambition to dare to go all in

In the field of artificial intelligence, Google has always been unparalleled Controversial leader. In the field of search, Google is also recognized as the dominant player.

But the emergence of ChatGPT has given this absolute dominance a gap that can be shaken.

ChatGPT, which may appear to be just a chatbot at first glance, relies on a large language model to bring a storm to mainstream Q&A products. Anyone who has used ChatGPT may wonder: "Why can't I use it to search?"

Traditional search results pages are often filled with a large number of advertisements and trackers. This is also the source of Google's high profit margins. But ChatGPT is different. It gives search results or suggestions through dialogue with users and context awareness, making it appear natural, clean, and friendly. When people realize that there is another possibility for search, the traditional search model begins to stand on the precipice.

Pichai pointed out in a public post: "Advanced generative artificial intelligence and large language models are capturing the imagination of people around the world." It can be seen that Google's decision-making level is not Didn't realize this trend.

In fact, shortly after the release of ChatGPT, Pichai became aware of the threat and issued a "Code Red" warning internally, mobilizing researchers to find ways to deal with it. In addition, Google founders Larry Page and Sergey Brin, who had left the company, also returned to the company urgently and held multiple executive meetings to emphasize the urgency of dealing with ChatGPT.

In addition, as early as the Google I/O 2021 Developer Conference, Google released the large language model LaMDA and the multi-task unified model MUM for conversational applications. Moreover, these two models demonstrated their powerful capabilities in tasks such as multi-turn dialogue and cross-modal search in official demonstrations. This is much earlier than the node where ChatGPT appeared, but Google completely missed the opportunity.

The competing product of ChatGPT, Bard powered by LaMDA, has only recently appeared in the public eye. At this time, ChatGPT has achieved the success of exceeding 1 million users in 5 days and 100 million active users in 2 months, making it the fastest growing consumer application in history.

What’s even more amazing is that Microsoft quickly signed an additional billion-dollar investment agreement with OpenAI and announced in a high-profile manner that it would integrate ChatGPT into its entire product suite. In other words, in addition to Bing, ChatGPT may be integrated into related functions of Office, Github, and Windows in the future.

In contrast, Google currently announced that Bard is still in the testing phase. Although it has been demonstrated externally, the launch time of the official version is still a mystery. And although Pichai mentioned that he will apply the latest artificial intelligence technology to search, the specific form is not yet known.

3. I wanted to give everyone the whole big job but was "stolen"

Before this game started, no one expected It seems that in the field of AI and search engines, Microsoft, which is on a narrow road with Google, will come out on top.

But if you try to sort it out, you can find that in the face of the "radical" Microsoft, the "conservative" Google is not unfair at all.

The reason why Google is so conservative can be mainly attributed to the following points:

First, try to keep a low profile and avoid AI ethical controversies. In recent years, Google's AI ethics farce has been frequent. Topics such as algorithmic bias, dismissal of minority female researchers, and AI awareness awakening incidents have repeatedly caused controversy. It needs to be cautious in order to rebuild credibility.

Secondly, the gains outweigh the losses, which is contrary to the main business model. Google's main revenue comes from advertising, but if the chatbot includes the display of advertising pages, it will undoubtedly be extremely "drama" and extremely bad for the user experience. How to find new business models is a huge challenge for Google.

Thirdly, the cost is huge and it is difficult to find a balance. The combination of search and artificial intelligence is not new, but this combination is actually not deep. Because the search itself requires a lot of computing power support, and the cost of a single round of answers by a model like ChatGPT may not be high, but if it is superimposed on the hundreds of millions of searches per day by search engines, the cost of answering can be imagined. More importantly, this model cannot yet produce direct economic benefits.

In short, although Google has never been willing to spend money on investment in generative AI and large language models, it has too many worries and foresight, and has not found feasible technical advantages. landing path. Google, the dominant player in the search industry, may be planning a big job, but it didn't expect that it would be "stolen" by the sudden emergence of ChatGPT.

As a giant, why does Microsoft dare to take the lead in promoting in-depth cooperation between Bing and ChatGPT?

According to speculation, the key lies in the significant gap between the market shares of Bing and Google search. Statistics show that Bing’s market share in the global search market currently does not exceed 10%, while Google’s market share is conservatively estimated to be no less than 80%. Because of this difference in size, Bing's revenue during the same period was far lower than Google's search business. In other words, due to the low profit margin, Bing does not need to bear too much revenue pressure.

Although a big ship can withstand pressure, a small ship can turn around better. Microsoft Bing dares to make such more breakthrough attempts because it can better afford the cost of trial and error. And no one can predict what chemical reactions will occur after chatbots like ChatGPT enter large-scale commercial use. The gamble has begun and the outcome is uncertain, but if a certain explosive point is achieved, then you can experience the comfort of sitting firmly on the Diaoyutai at the dawn of a new era.

4. The other side of ChatGPT is a cheating weapon, a rumor spreader, and a waste-making machine...

The popularity brought by ChatGPT is rising steadily. Never before has an AI technology been able to achieve this kind of circle-breaking effect.

In the past few months, about 100 million people have played ChatGPT. Students use it to write homework and check book lists; creative workers use it to write plans and develop projects. Brainstorming; programmers use it to check information and write code; even some companies find it surprisingly powerful after trying it out, and use it to generate copywriting and produce manuscripts...

From the perspective of functionality Judging from the above, ChatGPT is undoubtedly powerful. It can be both a highly practical production tool and an entertaining and creative toy. Bill Gates commented: "The emergence of ChatGPT is of great historical significance, no less than the birth of the Internet and personal computers."

This screen-swiping "carnival" seems to have arrived The boiling point. When Google launched Bard, people also had high hopes. However, Bard made mistakes at the demonstration site, which was like a drop of cold water in the head, making many people suspicious.

ChatGPT has also previously been caught providing inaccurate or outdated responses, but there hasn’t been much mention of this in the frenzy. Its power and novelty are enough to make people ignore this "flaw". Anyway, many artificial intelligences are like artificial retards, and people tend not to "expect too much" from machines.

No one can deny the role of ChatGPT, but things are far from simple. The other side of it is not so nice, and even a little scary. After all, it is just a tool, and it can not only be abused, but also used by people with impure motives.

For example, academic fraud. The emergence of ChatGPT has impacted the existing academic integrity system.

Not long ago, Antony Aumann, a professor of philosophy at Northern Michigan University, discovered when grading courses that the top paper in the class was actually written by a student using ChatGPT. The professor then decided that students must be supervised when writing the first draft of the paper and write in a browser that restricts access to the Internet; if there are changes to the draft, the student must give a reasonable explanation.

There are also many journals that prohibit ChatGPT from being listed as a co-author of papers. Science explicitly prohibits listing ChatGPT as a co-author and does not allow the use of text produced by ChatGPT in papers.

Another example is the spread of rumors. Mira Mulati, chief technology officer of OpenAI, once said in an interview with Time magazine that ChatGPT may "make up facts", which is no different from other artificial intelligence tools based on language models.

Recently, NewsGuard, an American news credibility assessment and research organization, tested ChatGPT. Researchers asked ChatGPT questions full of conspiracy theories and misleading narratives, and found that it can Adapt information in seconds to produce a wealth of convincing but unsourced content. They said that there is currently no way to effectively solve this problem, which will cause ChatGPT to become the most powerful tool for spreading online rumors.

And the most terrifying thing is the production of "waste". In the past, searches would give us a list of search results, but ChatGPT will integrate information, sort out clues, and provide a unique answer. This looks beautiful, but it further degrades our sense of touch.

"The popularization of the Internet has made the collection of information more and more convenient and fast. However, the spirit of self-breakthrough and continuous innovation has quietly disappeared. Since all knowledge has become readily available, On the contrary, we have lost a lot of opportunities to think hard and think independently."

This view of the well-known designer Yohji Yamamoto also applies to the present, just like the emergence of social media has disrupted people to a certain extent. The innovation of search engines has also weakened people’s ability to collect information and get closer to the truth to a certain extent.

When we are used to "using it as it comes", when facts and fiction are seamlessly mixed in the flow of information, and the only truth disappears in the huge noise, we also become Domesticated "waste".

Of course, it is not advisable to give up eating because of choking. After all, ChatGPT has achieved real scene implementation. Its emergence is a victory of AI technology. Although it is not perfect, it has unlimited potential. We just need to be vigilant, don’t let it become a carrier of evil thoughts, don’t let the monsters of technology come out of the cage of order.

Reference link:

https://www.theregister.com/2023/02/08/ai_battle_microsoft_google/?td=rt-3a

https://blog.google/technology/ai/bard-google-ai-search-updates/

https://www.tmtpost.com/6403782.html

The above is the detailed content of Zhou Hongyi: If you can’t catch this bus, you will easily be eliminated! Zhang Chaoyang added: If you don’t have the ability, don’t enter randomly!. For more information, please follow other related articles on the PHP Chinese website!