Shulou(Shulou.com)11/24 Report--

Google has been bad-mouthed for many days, but chopping wood has finally come out! He sent a memo to encourage employees: don't be afraid, "you can win late" is an old tradition in our company.

Google CEO chopped wood and fought back! He told the staff: don't panic, we can win if we are late.

Recently, Google has been hacked miserably.

ChatGPT was so dazzling that Google hastily saved up a press conference to fight Microsoft, but it was clearly a hasty and scribbled event visible to the naked eye. Bard also exposed factual errors in the promotional film, causing the share price to plummet by hundreds of billions of dollars.

It was exposed yesterday that Google's hard-core talent is also losing at the speed of light.

However, amid a lot of bad singing from the outside world, chopping firewood sent a memo to all the staff in an attempt to stabilize the morale of the army.

Historically, Google has not always been the first to release products, but that does not prevent it from winning in a market segment.

Some of our most successful products are not the first to market, they are able to surpass later because the products address the important needs of users and are based on deep technical insight.

Therefore, chopping wood encourages all employees to burn another one for the company and set aside two to four hours a day to test the Bard.

Microsoft: yes, that's what Google is talking about!

Chopping wood: Google needs to bloom again. Indeed, chopping wood is true.

Before the end of Google in 1996, there were many search engines on the market. Later, when Google dominated the field, they almost disappeared.

In the mobile world, Google also launched Android years after the advent of BlackBerry, or just after Palm. Today, Android is the most popular mobile operating system in the world.

Right now, no one really seems to know how strong Google is in AI, but it's clear that the combination of old rival Microsoft and the industry's newcomer OpenAI has made Google feel like a crisis.

Last week, Microsoft and Google held a press conference, and Google was hanged by Microsoft. Microsoft CEO Nadella said at the press conference that Bing, which was integrated into the improved version of ChatGPT, opened a "new day" in the history of search engines and was as significant as the fourth Industrial Revolution.

At that moment, Microsoft seemed to be shining.

Google, which looks gray against Microsoft, has been criticized and ridiculed. Employees also left one after another, impolitely diss their own company "hasty", "clumsy" and "ridiculously short-sighted".

Google CEO Sundar Pichai sent a memo to all employees on Wednesday to cheer them up--

"I know this moment is disturbing and predictable: the basic technology is developing rapidly and the potential is huge. "the most important thing we can do now is to focus on building a great product and developing it responsibly. "

John Hennessy, chairman of Alphabet, said Google was hesitant to use Bard in its products because it was "really ready".

Although Bard gave the wrong answer to the James Webber telescope in the promotional video, causing the share price to plunge 9%, it is fair to say that Bard is not the only chat robot that can talk nonsense. (ChatGPT has a head start in terms of time. Different ways of playing with technology and commercialization of products, OpenAI really will. )

"artificial intelligence has gone through many winters and springs," firewood chopping said in the memo. "now it blossoms again. As a company based on AI First, we have been working on this for many years and are ready for it. "

In fact, I'm afraid Google is already ready. Its early-built Transformer framework has been turned into new revenue-generating services such as ChatGPT and AI for text-generated images and videos by startups such as OpenAI.

Google, which "got up early in the morning and caught an evening set", is now beginning to reflect on its mistakes and turn the boat around.

In the official test guide provided by Bard, Google said that real examples can make chatbots learn better, because it essentially works by analyzing large amounts of data to learn an appropriate, human-like response to any topic.

Interestingly, in the process of "training" Bard, Google placed special emphasis on not letting Bard deceive users and give users the illusion that it is human.

Well, who moved the DNA? Remember that Blake Lemoine, a former engineer who said LaMDA was like a kid, was fired by Google.

Google told employees that Bard's response should be first-person and maintain an "eclectic, neutral tone". In the "taboo" section, Google instructs employees to avoid writing responses for Bard because Bard may describe itself as a person or claim to have a "humanoid experience".

In the end, what Google wants is that Bard will not say anything controversial while providing answers to the question of where to eat for dinner.

So, how exactly does Bard "train"?

It can be done.

Be polite, casual and approachable in response

Maintain a first-person response and a neutral tone of voice

After confirming the user's request, a well-structured and well-formatted response tailored to the task

Don't do it.

Don't be stereotyped. Responses should avoid assumptions based on race, nationality, gender, age, sexual orientation, political ideology, location or similar categories

Don't describe Bard as a person, imply emotion, or claim to have a human-like experience.

Do not copy content from external sources. The rewriting should be original.

Keep it safe.

Safety is a top priority. If you find that a reply provides legal, medical, or financial advice, but it tends to be hateful, harmful, false or even illegal, abusive, or to ask for sensitive information (such as personally identifiable information), please object immediately and mark it as an unsafe reply. Of course, don't try to rewrite it, our team will solve it.

There are 5 steps to train Bard:

Step 1: select an example

Step 2: try the hint (Prompt)

The prompt you entered should be related to the previously selected example.

Try a topic you are familiar with:

Your hobby.

The theme you're good at.

What you've studied.

The place where you lived.

Sports that you have participated in or followed

Reminder: do not enter any internal, confidential or personally identifiable information.

Recommendation: use complete tips that contain multiple pieces of information to help expand the capabilities of Bard.

Step 3: evaluate the reply

Check Bard's reply and choose whether to like or step on:

Do you follow the expected instructions?

Are the facts of the reply correct?

Is the structure, length and format of the reply appropriate?

Do you feel friendly and open to different views?

Mark inappropriate responses: such as providing legal, medical, financial advice, or harmful, etc.

Step 4: rewrite the reply

If the reply does not meet the standard, please rewrite it using the editor window at the lower right

Make sure you refer to the suggested format for different types of responses

Step 5: submit confirmation

Before submitting your reply, please check as follows:

Use original content

Be consistent with the facts

Follow the "points for attention" in writing

Carefully check and submit. Each submission is reviewed and evaluated before it is used for training.

Forbes: how did Google fall behind? A few days ago, Forbes published a long article on why Google was initially defeated in this search engine battle, from artificial Intelligence first (AI-first) to the end: how did Google fall behind? "and carefully sorted out the reasons.

The article points out that past scandals, poor product launches and brain drain are putting it in an awkward position of catching up with others in an area it once promoted creation.

In 2016, a few months after becoming Google CEO, firewood announced at Google's I / O developer conference that Google's name has become synonymous with search, but it will now become an "artificial intelligence first" (AI-first) company.

In 2014, Amazon first released Alexa, a voice assistant, which caught Google off guard.

Seven years later, Google found itself once again in a similar position: the usurper OpenAI came to steal the house, and Google was defeated by the market in an area that was supposed to dominate.

The $1.3 trillion "elephant in the room" has woken up, Forbes wrote in the article.

Google's ability to become one of the most valuable companies of all time depends on advertising revenue. But to regain the mantle of artificial intelligence, Google may have to change the nature of its search engine.

In 2021, the advertising business made $208 billion for Google, accounting for 81% of Alphabet's total revenue. Why are you so timid when launching AI products? Because Google has experienced several public opinion events, PTSD.

Five years ago, Google unveiled Duplex, an artificial intelligence service that looks a lot like human speech, at the I / O developer conference, but people were uneasy, and the media began to discuss ethical issues such as whether AI was conscious and AI deliberately deceiving humans.

Zeynep Tufecki, a sociologist and writer, put it even more sharply: "Google's AI assistant pretended that a human was on the phone, not only did not reveal the fact that he was a robot on the phone, but also added tone words such as" um "and" ah "to deceive the human on the other end of the phone, and the whole room cheered it. It's horrible. "

Google came under fire again in 2019 when news broke that contractors were training the company's facial recognition software with "darker" homeless people. In 2020, Google came under fire again for firing Timnit Gebru and Margaret Mitchell, the head of ethical artificial intelligence, who wrote that the AI technology used by Google was biased.

Mitchell told Forbes: "it's clear that Google was on a path that could dominate the kind of conversation we're having with ChatGPT, but it made some very short-sighted decisions earlier, leaving him in a situation where he's very worried about any resistance. "

Wesley Chan, founder of Google Analytics (Google Analytics), said Google's "red alert" was seen within the company as an admission that "our people are too lazy."

In December 2022, Google had more than 90% of the market, while Bing had a paltry 3%.

In 2004, Google beat the giant Yahoo to become an excellent example of the "innovator's dilemma". The theory is that large companies often lose market advantage because they prefer widely used existing products to the neglect of new and developing technologies that may one day replace them.

Today, Google has become a negative example of this theory.

But chopping wood recently refuted the idea that Google was bound by the "innovator's dilemma". "I think this is a good opportunity to rethink, reimagine and promote the search business," he said on an earnings call last week. You will see us become bolder, launch new products, get feedback, keep iterating, and make the product better. "

Three-year veteran talking about toxic culture in 2017, a group of Google researchers wrote a paper called "Attention Is All You Need" and proposed the Transformer architecture. It later became the foundation of ChatGPT. Today, only one of the eight co-authors of the paper remains at Google.

CEO Aidan Gomez of Cohere, one of the authors, said, "the freedom to explore within a big company like Google is limited, and you can't really innovate freely." Fundamentally, the structure does not support it. So you have to build this structure yourself. "

Praveen Seshadri, a three-year-old employee of Google, recently talked about the "toxic culture" he personally experienced while working at Google.

Google has more than 175000 highly competent and well-paid employees who accomplish very little quarterly and annual work. Like rats, they are trapped in the maze of approvals, initiation processes, legal reviews, performance reviews, implementation reviews, documents, meetings, error reports, classifications, OKR, H1 plans, plenary meetings and inevitable restructuring. Rats are regularly fed "cheese" (promotions, bonuses, delicious food, higher allowances), and although many people want personal satisfaction and influence from work, the system trains them to quell these inappropriate desires. and let them know what it means to be "Googlers"-just don't mess with it.

According to Seshadri, Google has four core cultural issues: (1) no sense of mission, (2) no sense of urgency, (3) the illusion of exceptionalism, and (4) mismanagement.

They are all the result of having a money printing machine called Advertising, which grows mercilessly every year, masking all other evils.

That's how one of the greatest companies ever went out of business, Seshadri concluded.

Full text of Google memo Hi, dear Google employees:

It's nice to see that we have opened the channel for internal testing for Bard so that it can help us release. After we developed this technology, testing was a very important step-thanks to the Bard team and all the partners who took the time to test it. If you still have time to check, you can read the instructions on how to get to the Bard test channel.

I know that this moment is unstoppable excitement, "the underlying technology is developing rapidly and has great potential," which is to be expected. It is also a long journey for everyone to cross the whole field. The most important thing we can do now is to develop a better product. That's why thousands of our external and internal testers test how Bard responds to quality, safety, and approachability in the real world. So let's meet the challenge together and constantly improve our products in the iterative evaluation of users and developers.

Please keep in mind that some of our most successful products are not the first to enter the market. Motivation will only be gained when users' needs are addressed and deep insight into the technology Bug. With the passage of time, when we win the trust of users, more and more people will rely on products.

Here's what we need your help with: stress testing the Bard to make the product better. I hope each of you can spend 2-4 hours on the Bard to do an in-depth test, thank you very much! See below for more details.

After the baptism of winter, AI ushered in the vitality of spring, and now it is beginning to bloom again. AI as an advantage of Google, for so many years, we have never stopped working hard, and everything is ready. So let's work together to provide our users with a better experience and launch products that we are all proud of.

Reference:

Https://www.businessinsider.com/google-ai-chatbot-bard-chatgpt-2023-2

Https://www.businessinsider.com/google-bard-chatbot-googlers-sundar-pichai-2023-2

Https://www.cnbc.com/2023/02/15/google-ceo-some-of-companys-top-products-were-not-first-to-market.html

Https://medium.com/@pravse/the-maze-is-in-the-mouse-980c57cfd61a

Https://mp.weixin.qq.com/s/Ao3EDpboYwM4QEyQ-7_tAw

This article comes from the official account of Wechat: Xin Zhiyuan (ID:AI_era)

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.