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

Does Google still have a chance to turn around in this AI war?

GPT-4 came out of nowhere, which made Microsoft win again!

On the same day, Google went to war with Microsoft again and came up with a full set in one breath:

-PaLM API & MakerSuite, an open big language model

-the generative AI is packed into the Google Workspace family bucket.

-generative AI supports Vertex AI platform

-New platform Generative AI App Builder

The release of so many shows that Google is in a moment of panic.

Challenge GPT-4! PaLM API announced that Google officially opened its PaLM large language model API for the first time.

This means that people can use it to complete tasks such as summarizing text, writing code, and even train PaLM to be a chat robot like ChatGPT.

In April 2022, Google announced a 540 billion-parameter PaLM for the first time. Like other LLM, PaLM can perform a variety of text generation and editing tasks.

PaLM is the first large-scale use of Google's Pathways system to expand training to 6144 chips, which is by far the largest TPU-based system configuration for training.

Its ability to understand is outstanding, not only can read jokes, but also can explain to you where the joke is.

In addition, to make it easier for developers to train PaLM to perform specific tasks, Google is launching a new tool along with PaLM API called MakerSuite.

With MakerSuite, you can iterate over prompts, enhance datasets with composite data, and easily adjust custom models.

This process can be done in the browser, while the computing-intensive work of training and deployment is handled by Google Cloud.

PaLM API and MakerSuite can be accessed by selected developers in Private Preview and a waiting list will be launched soon, Google officials said.

In addition, Google also introduced generative AI to Google Cloud.

For example, support for generative AI has been expanded in its Vertex AI platform. The platform is designed to help enterprises train and deploy machine learning models.

Vertex will have access to more models built by Google Research and DeepMind, and will be able to take advantage of open source and third-party systems in the future, Google said.

Finally, Google will launch a new platform called Generative AI App Builder.

The platform allows developers to ship new experiences quickly, including robots, chat interfaces, custom search engines, digital assistants, etc.

Developers can access Google's basic model through API and use out-of-the-box templates to start the creation of generative applications in minutes or hours.

In addition to exposing PaLM API and other tools that are good for developers, Google has also introduced generative AI into its own office components Google Workspace, such as documentation, Gmail, Sheets, Slides and so on.

Currently, Google has more than 3 billion Google Workspace users worldwide. This time Google will integrate the generative AI into the "family bucket", mainly targeting Microsoft Office office software.

Microsoft is expected to reveal more about the ChatGPT version of Office at a press conference on Thursday.

In the documentation demo given by Google, an outline of the job description is created based on a very brief tip.

In the future, with the help of AI, you can brainstorm, proofread, write, and rewrite documents.

You can simply list some key points about the work meeting, and Google Docs can expand it into a "more elaborate summary."

You can even specify the tone manually, such as whether it should be "more fantasy" or "formal".

When trying out the Gmail email service for AI, you can type "draft an email to the team" in the text box of the email, and Gmail will create a "three-paragraph" thank-you note.

You can not only edit and add content, but also ask for a rewrite.

In the demonstration, you can also use AI to write personalized marketing messages for customers, convert the main points into a complete email, and summarize a long list of email chains in Gmail.

Of course, Google is currently testing artificial intelligence products that can only be accessed by a limited number of Workspace users.

Google also said it plans to introduce more artificial intelligence features to Workspace later this year, including generating formulas in worksheets, automatically generating images in slides, and taking notes in Google Meet.

Here is a list of AI capabilities that Google has announced to add to Workspace in the future:

-drafting, replying, summarizing and prioritizing Gmail

-brainstorming, proofreading, writing and rewriting in documents

-use automatically generated pictures, audio and videos in slides to turn your creative ideas into reality

-gain insights and analysis from raw data through automatic filling, formula generation and context classification in Sheet

-generate a new background and capture notes in Meet

-enable workflow in chat to complete tasks.

Upgraded version of Big Bard, under testing. Google also stepped up the launch of the artificial intelligence robot Bard after Microsoft released the new Bing.

However, Bard's debut turned upside down, wiping out hundreds of billions of dollars from Google's market capitalization.

In a demonstration, "how to tell a 9-year-old about the discovery of the James Webb Space Telescope".

Bard replied: "the Weber Telescope took the first pictures of planets outside the solar system." In fact, as early as 2004, the European maximum Telescope captured an image of a galaxy outside the solar system.

In the only demonstration, Google made such a big mistake that it lost face and had to quickly remove the YouTube video of the presentation.

As a result, Pichai asked employees to test and improve its internal chat robot Bard before it is released to the public.

But it has been revealed that many employees are already playing with a more advanced version of the artificial intelligence chat robot, internally known as "Big Bard".

Of course, Big Bard is the same as Bard, which is based on the LaMDA model, but the former has larger parameters and is smarter.

Insider looked at users asking the same question to both versions, and Big Bard output richer, more human-like answers. It is also generally more talkative, more casual, and more likely to swear.

One employee said that when he tested Big Bard, he would see a prompt instructing him to test with a regular version of Bard.

Some leaders require employees to use only the standard version, and the early version of the internal test is called "Apprentice Bard".

Pichai once introduced that Bard is a lightweight model based on LaMDA and requires less computing power.

Because of the cost, Google may want to launch a limited edition of Bard. John Hennessy, chairman of Google's parent company Alphabet, recently said that using artificial intelligence to search, Google may cost more than 10 times as much as ordinary keyword searches.

Next, Microsoft will hold a "Reinventing productivity with AI" event on March 16, when officials may announce the integration of ChatGPT into Office 365 office software.

Google's integration of generative AI into Workspace is still being tested, and Microsoft will certainly be one step ahead.

Does Google still have a chance to turn around in this AI war?

Reference:

Https://www.theverge.com/2023/3/14/23639273/google-ai-features-docs-gmail-slides-sheets-workspace

Https://developers.googleblog.com/2023/03/announcing-palm-api-and-makersuite.html

Https://blog.google/technology/ai/ai-developers-google-cloud-workspace/

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.