Google is scared! ChatGPT smashes search engine jobs, CEO Pichai calls a meeting to sound "red alert"

This month, OpenAI’s ChatGPT came out and immediately caused a big storm in the AI ​​circle.

Clear, intelligent, and step-by-step, and can even handle papers and codes. At that time, many people predicted that Google might be a pill.

This is not empty talk, now, Google is really afraid.

It is reported that with the popularity of ChatGPT, Google management has recently issued a "Red Code".

Previously, Google management also vowed that ChatGPT could not replace us. Now is this a slap in the face at the speed of light?

Chopping firewood and shivering

What does "Code Red" mean? In Silicon Valley, this means sounding the "fire alarm."

The largest company in Silicon Valley is really scared now - perhaps, ChatGPT is really triggering a huge technological change that will subvert the entire industry.

The change was so sudden that Google didn’t even react.

ChatGPT can answer people’s inquiries with clear, simple sentences instead of just providing a list of links. It explains concepts in a way that people can easily understand.

It can even write an article from start to finish, including business planning, Christmas gift suggestions, blog posts, and holiday plans.

More and more people are asking: Will search engines be replaced? Now, Google isn't sure either.

Currently, Google CEO Pichai has participated in several meetings surrounding Google’s AI strategy, and has given key instructions: Google’s multiple teams need to focus on solving the impact of ChatGPT on the company’s search engine business. threats posed.

According to the Times, teams in Google’s Research, Trust and Safety and other departments have been instructed to shift gears and instead begin assisting in the development and development of AI prototypes and products. release.

Some of these employees are tasked with building AI products that generate art and graphics, similar to OpenAI’s DALL-E, which is already being used by millions of people and is a fairly successful The product.

Why is Google so afraid of ChatGPT? Because, if it really replaces the search engine, Google's advertising revenue business model will be greatly impacted, or even fundamentally subverted.

This is a fundamental loss of jobs.

Sridhar Ramaswamy, the former head of Google’s advertising team, said that if ChatGPT continues to be so popular, Internet users will no longer click on ads. Google linked. Advertising earned Google $208 billion in 2021, accounting for 81% of Alphabet’s total revenue.

Five days after its public release, ChatGPT has accumulated more than 1 million users. In the face of user queries, it can collect information from millions of websites and communicate in a conversational, human-like manner. methods to generate unique answers. It can write papers, write code, and even act as people's therapists.

Of course, it is also often full of mistakes because it cannot verify what it says and even makes up answers.

In fact, Google once had the opportunity to take this path. Before the advent of ChatGPT, Google's Lamda chatbot was already famous.

But Zoubin Ghahramani, head of Google Brain, Google’s AI lab, told The Times that chatbots are “not the kind of thing that people can reliably use every day.”

Instead, experts say, Google should focus on its search engine over time rather than dismantle it.

Ten days later, the attitude changed suddenly

Previously, Google did not have this attitude towards ChatGPT.

More than ten days ago, Google executives stated this: Google will not launch a competitor to ChatGPT because Google has greater "reputational risk" than startups such as OpenAI.

And they believe that chatbots have not yet reached the point where they can replace search engines.

Previously, Google’s AI chatbot LaMDA made a stunning debut, but Google was unwilling to release it to the public because its error rate was too high. It is easy to "carry poison".

More than ten days ago, Jeff Dean, head of Google AI, said at a plenary meeting that although Google has the technology and capabilities to make AI products, compared with "small start-ups", They must make decisions "more conservatively."

At that time, at the meeting, a Google employee once asked: Did Google miss a good opportunity by not launching a chatbot like OpenAI? Especially considering that Google already has its own conversational technology, such as LaMDA, and language models for other conversational applications.

Dean’s answer is that although Google is fully capable of making chatbots, chatbots are vulnerable to bias and disinformation, and Google is a large company with more than one billion users. It is even more difficult for companies to escape this influence.

"Of course we very much hope to use these technologies in real products, especially in things that highlight language models, rather than hiding behind the scenes, like now.", Dean said this when talking about Google's AI. "But, more importantly, we have to do the right thing."

In addition to LaMDA, the BERT and MUM AI language models that Google uses to improve its search engine can also Pull it out and compete with ChatGPT.

However, Morgan Stanley also issued a report stating that ChatGPT will not pose a major threat to Google’s position because Google will continue to improve its search engine and language model.

Google: My AI is stronger, but advertising is more important

Google’s entanglement is very easy to understand, because Google is not at a disadvantage in the field of chatbots.

As early as the I/O conference in May 2021, Google demonstrated its latest artificial intelligence system LaMDA.

Google said that LaMDA can make the answers to questions more "make sense" and allow conversations to proceed more naturally, and these responses are not preset (unlike those chatbots) , even the same answer will not be used a second time.

At the I/O conference in May 2022, Google announced the upgraded second-generation model-LaMDA 2.

During the demonstration, the AI ​​asked at the beginning: "Have you ever thought about why dogs like to play fetch games so much?" After answering simple questions such as "Why is this?" In answers to follow-up questions, the AI ​​gave more information about dogs and their sense of smell.

#What made LaMDA completely out of the circle was that in June this year, Google engineer Blake Lemoine actually had a romantic relationship with LaMDA and firmly believed in it. Not only does it already have the intelligence of an eight-year-old child, but it is also "conscious."

The next month, Google fired Lemoine, who "had gone crazy".

Although the whole thing seems very nonsensical, it does show that chatbots have made great progress in technology.

Since Google itself is also researching chat AI, why hasn’t it been deployed yet?

First of all, a considerable amount of data on which AI models are trained comes from the Internet, which is full of prejudice, hatred and abuse.

Given that Google itself attracts attention, if the model outputs negative results, public opinion will easily ferment to the point where it cannot be ended. And the image Google has spent decades building will be in flames.

This is not unfounded worry. It is not difficult to see from the fact that netizens have brainstormed and tried their best to bypass the security restrictions of ChatGPT.

Secondly, even as Google perfects chatbots, there’s still a burning question — whether the technology will cannibalize the company’s lucrative search engine advertise?

After all, for Google, digital advertising revenue accounts for more than 80% of all revenue.

Obviously, if the chatbot can respond to queries in tight sentences, people will have no reason to click on ad links.

So, before the advent of ChatGPT, Google’s strategy was to use chatbot technology to enhance the search engine experience.

ChatGPT actual PK Google search

So, can ChatGPT replace Google?

Let’s do a practical exercise first to see who is the winner.

CNBC columnist Sofia Pitt wrote an article "About the day I switched from Google to ChatGPT" and personally tested and compared the usage of the two:

I recently bought a second fiddle leaf fern potted plant, but the new one died after only a few days, normally I would ask Google what to do Do.

And this time I tried to find the answer on ChatGPT, I typed "How can I keep my fiddleleaf fern plant alive"? The results were consistent with the plan I received from the plant company Easy Plant.

Compared to Google, I don’t have to navigate through thousands of entries and websites, nor do I have to endure various pop-up ads.

When I googled the same question, the top result gave me detailed instructions in an article that also included an ad and a link to purchase new soil.

In the first round, ChatGPT won.

Next I wanted to prepare a gift for my husband for Hanukkah, so I asked ChatGPT for some advice.

I didn’t give it any information about my husband other than the gifts were for Hanukkah.

Then ChatGPT assumes my husband is kosher and that he likes technology and watches. There's nothing wrong with that, but it's not the answer I was hoping for.

When I googled this question, I had access to hundreds of articles with different gift ideas and links to websites.

I received advertisements for Lululemon and Bombas, which I used to go to frequently to buy gifts.

In this case, Google uses its personalized knowledge of me to find more targeted results from around the web.

In the second round, Google won.

Can we kill Google?

Having said so much, the question is: Can ChatGPT or the upcoming GPT4 replace Google?

Zhihu Answer Zhang Junlin’s answer is, not yet.

Based on the powerful GPT 3.5 large-scale language model (LLM, Large Language Model), ChatGPT introduces "artificially labeled data reinforcement learning" to continuously fine-tune the pre-trained language model, mainly The purpose is to let the LLM model learn to understand the meaning of human command instructions.

First of all, for many knowledge-type questions, ChatGPT will give answers that seem reasonable but are actually wrong (you can see a variety of Weird answer).

Secondly, ChatGPT’s current model further adds annotated data training based on the GPT large model. It is very unfriendly for the LLM model to absorb new knowledge.

New knowledge is always emerging, and it is unrealistic to re-pretrain the GPT model when new knowledge appears. It is unacceptable in terms of training time and money costs.

If we adopt the Fine-tune model for new knowledge, it seems feasible and relatively low-cost, but it is easy to cause the introduction of new data, leading to the problem of catastrophic forgetting of original knowledge.

Thirdly, another limiting factor is cost.

The current cost of AI training and output results is still not cheap. If we face hundreds of millions of user requests from real search engines, assuming we continue to adopt a free strategy, OpenAI cannot afford it.

However, if a charging strategy is adopted, the user base will be greatly reduced. Whether to charge is a dilemma.

In addition, the emergence of ChatGPT has indeed had a fundamental impact on Google’s own search engine business.

Google’s biggest source of revenue is advertising, that is, users who type questions into the search box will get relevant content and advertising links.

Google’s business is to help users find information that can accurately solve user problems in the vast sea of ​​information more quickly. But the problem is that if it is solved too quickly and easily, there will be no way to sell more ads. This is a somewhat self-contradictory weakness of search engine advertising itself.

Of course LaMDA can also allow users to enter questions and insert ads next to or below the content, but the experience will be very different from that of search engines, so this kind of conversational AI may have an impact on Google impact on the nature of business.

But the advent of ChatGPT is also a part of competition, which is likely to accelerate the speed at which Google is forced to launch LaMDA. At least it cannot let ChatGPT monopolize a large number of users and cannibalize the demand for search engines.

And currently ChatGPT cannot connect to the Internet, so it is impossible to use the latest information to optimize his answers, but this will not be a limiting factor in the future.

##ChatGPT himself also has this self-knowledge

What about the big guys in the circle? See

When it comes to whether ChatGPT can replace Google, Marcus, who has always looked at the excitement and is not too big a deal, will definitely have some cynicism.

Black Lemonie, an engineer who was previously fired by Google, believes that ChatGPT is still a few years behind LaMDA.

OpenAI CEO Sam Altman himself also mentioned that ChatGPT still has many limitations, and there are also many improvements that need to be made in terms of robustness and reliability. It is too early to rely entirely on it.

The following are the main limitations of ChatGPT at this stage, which may vary depending on the specific language model and the tasks it is designed to perform:

• Language models are trained on specific data sets and may not be able to answer questions outside the training domain (e.g. a language model trained on news articles may not be able to answer questions about medical topics or technical details);
• Language models are not always able to understand the context or intended meaning of a question and may therefore provide inaccurate or less relevant answers;
• The language model may not be able to handle multi-step complex reasoning tasks, such as solving mathematical problems;
• The language model cannot browse the web or access external information sources. Can only provide answers based on its training data;
• Language models may not provide accurate answers to sensitive and negative topics every time.

To summarize

The value of ChatGPT is its ability to explain complex issues and complete simple writing tasks as if you were talking to a real person.

For example, when we asked ChatGPT to explain concepts such as markets and the Internet, the responses we received were like talking to an expert.

Compared to Google, we don’t have to spend time filtering unwanted results.

We can also ask it to write a letter to the landlord requesting early termination of the lease. We can almost directly send him the results of ChatGPT, saving time and effort.

But on the other hand, Google knows us better.

It will customize our needs and provide diversified suggestions.

For example, if you search for Apple stock ticker or Cheap flights to Aruba, it will directly display a market chart with the latest price information, or an itinerary with the most likely cheapest flight dates, and Connect to multiple websites and let's buy tickets directly.

ChatGPT will not scan the Internet to obtain real-time information, and there is a flaw - it has only been trained on data before 2021.

Google is also very reliable, thanks to the company's large operating budget and years of expertise. In contrast, ChatGPT is still in testing and may experience glitches from time to time.

Can ChatGPT replace Google? It can be guessed that most people will continue to use Google to search most questions, but if they are not satisfied with the results, we have a considerate alternative.

According to reports, at a recent all-hands meeting, Pichai said that Google has many plans for AI in 2023.

At Google’s annual developer conference I/O next May, we should be able to see what Google has made after giving up the chatbot route.

The above is the detailed content of Google is scared! ChatGPT smashes search engine jobs, CEO Pichai calls a meeting to sound "red alert". For more information, please follow other related articles on the PHP Chinese website!