Google Baidu Officially Announces ChatGPT Competitor! Whether or not the search engine can be subverted is not so important anymore

In November 2022, OpenAI released the chatbot ChatGPT.

Within five days of going online, ChatGPT attracted more than 1 million users, and soon ChatGPT replaced the voice of programmers, text workers, and search engines.

Within two weeks of ChatGPT’s launch, Google sounded a “red alert.” Subsequently, the two co-founders of Google were recalled by the CEO.

Two months after its launch, the number of ChatGPT users reached 100 million.

The time has come to today. After ChatGPT attracted the attention of the whole world, Google and Baidu, the industry giants of search engines, finally launched their respective "ChatGPT".

Google CEO personally fights against ChatGPT

In the early hours of February 7, Beijing time, Google CEO Sundar Pichai announced the launch of an experimental conversational AI service called Bard.

Bard is open to a limited number of developers first, and will be available to the wider public in the coming weeks.

The CEO personally released an open letter, Bard was listed as a "code red" priority project, and the industry leader finally confronted ChatGPT head-on.

In May 2021, Google launched LaMDA, a large language model. According to the official paper in early 2022, the LaMDA model was trained with up to 137B parameters, demonstrating dialogue quality close to human levels.

Bard is based on the LaMDA model, but currently uses a lighter version that requires less computing power, so it can be extended to more users and thus get more feedback.

While ChatGPT is limited to pre-2021 data, Bard will keep pace with the world:

Bard strives to combine extensive world knowledge with the power and creativity of large language models, which leverage information from the web to provide up-to-date, high-quality answers to input questions.

Judging from the official pictures, Bard has a dialog box similar to ChatGPT.

Bard can simplify complex topics, like explaining new discoveries from NASA's Webb telescope to a 9-year-old, or walking you through the best strikers in football right now.

Bard can also help with more specific and subjective tasks, including "plan a friend's baby shower," "compare two Oscar-nominated movies," "get lunch ideas based on what's in the fridge," and more.

However, Google didn't mention integrating Bard into Search , but it does plan to include new AI features in Search.

Google has found that when people use Google Search, they not only need factual answers, such as "How many keys does a piano have", but also increasingly insight, such as "Which is better to learn, the piano or the guitar? How long does it take to practice each?" to master?"

That's where AI comes into play. In the face of questions that do not have standard answers, it can give more comprehensive information, distilling complex information and multiple viewpoints into better-understood answers.

In the future, the Google search of the "AI-enhanced version" will not only find richer opinions, such as blogs of people who play piano and guitar at the same time, but also give you a deeper understanding of a related topic, such as what are the steps for beginners to get started.

In this dimension, AI has changed our relationship with information, as Google CEO Pichai said:

One of the most exciting opportunities is how AI can deepen our understanding of information and turn it into useful knowledge more efficiently, helping people more easily find the core when searching and achieve their goals.

The chatbot race, now

Before releasing Bard, Google had hesitation of one kind or another, whether it was an excuse or sincerity.

Google’s AI chief, Jeff Dean, once told employees that Google has a greater “reputational risk” in providing misinformation and is therefore “more conservative than smaller startups,” CNBC reported.

This is understandable. After all, chatbots learn from the vast amount of text on the Internet, so they cannot distinguish fact from fiction, and may also perpetuate the inherent prejudices of the Internet and human society.

Perhaps a more practical reason is that large language models affect Google Search's current business model — about 81% of Alphabet's 2021 revenue comes from advertising, the majority of which is Google's pay-per-click ads.

The forthcoming Bard shows that Google has changed its stance.

Although Google CEO Pichai emphasized that AI must be developed in a "bold and responsible manner", he did not elaborate on how to avoid harmful content, but said that external feedback should be combined with internal testing to ensure that Bard's answers are of high quality, safety and security. It can be regarded as a response to the "reputational risk" problem.

More than 2 months after ChatGPT sat firmly on the throne, the chatbot wars really started.

Next month, Google will begin connecting independent developers, creators, and enterprises to try out a generative language API based on LaMDA and a series of models.

Microsoft, which has invested billions of dollars in OpenAI, is also scrambling to integrate ChatGPT into its search engine and office software. A new version of Bing that integrates ChatGPT was briefly launched on February 3.

According to the screenshot shared by designer Owen Yin, the search box of the new version of Bing has become a chat box, and Bing will reply to users' questions in chat bubbles. The new version of Bing encourages users to search using natural language rather than keywords, and is able to adjust search results according to user requirements.

In addition to Internet giants such as Google and Microsoft, many startups such as, Perplexity AI, and Neeva have also joined the competition and launched similar chat-style search engines.

Among them, was founded by language and artificial intelligence expert Richard Socher. It can provide answers through the chat interface, and the reply has a quote to help users track the source of a piece of information. This is a function that ChatGPT does not have.

But this strength is not unique. OpenAI is developing an AI system called WebGPT, which will be able to answer questions more accurately and even explain the source of citations.

This is a process of competition and a process of complementing each other.

In addition, the Baidu version of the ChatGPT project "Wen Xin Yi Yan" (English name ERNIE Bot) has also been officially announced. It will complete internal testing in March and will soon be open to the public.

The official stated that Baidu has been deeply involved in the field of AI for decades, and has an industrial-level knowledge-enhanced Wenxin large model ERNIE, which has cross-modal and cross-language deep semantic understanding and generation capabilities. "ChatGPT-related technologies, Baidu has it."

In an interview with China Business News, several people in the AI ​​industry pointed out that the essence of ChatGPT’s out-of-the-circle is the natural result of the increase in computing power and sample size.

Now each Internet company has handed over its own answers, making a subversion in the search engine industry, which also means that when the big model, big data, and big computing power are accumulated, and the statistical laws in the training data reach the node, we will welcome Here comes the "Cambrian Explosion" of AI.

Search Engines Face a Real Challenge

Google hastened to launch a chatbot against ChatGPT, which to some extent shows that the search engine has really encountered a challenge.

Since the end of the last century, the inherent model of search engines "indexing, retrieval, and sorting" has dominated for more than 20 years. Google makes thousands of changes to the search engine every year, most of which are small and have not undergone fundamental changes.

The core of the search engine is the collection of massive information, rather than the creation of information. You enter keywords in the search box, and the search engine crawls, indexes, and sorts the results that match your query according to the algorithm, and then you see a large number of links, and then find the information you need from them.

ChatGPT belongs to AIGC (Artificial Intelligence Generated Content), which is a new way of content creation. It has been trained by the data set. Through one-on-one dialogue and human-like tone, it can give a single and instant answer. It can also combine context to achieve multiple rounds of dialogue to help you solve more complex and continuous problems. .

Sometimes ChatGPT is very stupid. It only generates text, lacks timeliness, and cannot display the source of information. It does not guarantee accuracy, and even makes mistakes in elementary school arithmetic. Even if it gives wrong information, it still has a confident tone.

But it should be noted that it is still in its infancy, and it has already fundamentally changed the way we interact with knowledge.

As mentioned above, when people search on Google, they're increasingly looking for insights rather than just factual answers. Even if we just use search engines and chatbots to complement each other, it means that our fundamental need for "search" has changed.

To make ChatGPT smarter, humans first need to learn how to communicate with it, which is already an essential skill for humans.

The game blogger @祝佳音 taught ChatGPT to speak in an old Beijing tone. UC Riverside researchers let ChatGPT familiarize itself with the works of American philosopher Daniel Dennett, and then let it pretend to be the philosopher himself to answer questions. The primary "digital life" is realized on ChatGPT.

For another example, you can guide the rules step by step, let it design games and other products, or give it a program, let it check for bugs, and you can also give it a demonstration case, so that it can draw inferences about other cases from one instance. The more complex the interaction, the richer the capabilities of ChatGPT will be.

There are also many scenarios around the world that actively ask AI to do it for you, from term papers, creative planning, to real estate introductions, and legal documents.

We can criticize AI for its accuracy today, but we should also see its creativity. It doesn't really understand the world, but it can actually communicate in natural language, communicate the needs of its interlocutors, and process massive amounts of information efficiently. Now it is already eye-catching, and it will be able to do better and better in the future.

It is as beneficial as autumn frost, and it can eliminate evil disasters. Work email: [email protected]

#Welcome to pay attention to Aifaner's official WeChat public account: Aifaner (WeChat ID: ifanr), more exciting content will be presented to you as soon as possible.

Ai Faner | Original Link · View Comments · Sina Weibo