


Silicon Valley unicorn is back to shock the world again!
The core strength of this unicorn called Character Technologies comes from the former Google LaMDA team.
Less than a week after their new product was launched on the mobile side, the number of downloads reached 1.7 million, directly crushing ChatGPT!
And, the company is now valued at US$1 billion.
So, this is another OpenAI?
The number of downloads surpassed ChatGPT in the first week. Why?
Character.AI is an application that allows users to freely create AI characters and chat with them.
## Sounds familiar, right? After all, there have been many similar products before.
However, Character.AI’s explosive popularity was unexpected.
After launching on May 23 for iOS and Android users worldwide, Character.AI has become a huge hit. In just 48 hours, the Android version has been installed more than 700,000 times.
A week later, even more astonishing figures came - the number of downloads on the mobile side exceeded 1.7 million.
Character.AI claims that before the app was released, the Character.AI website had been visited more than 200 million times, with users averaging one visit per visit It took 29 minutes - this figure is 300% of ChatGPT.
Even more than YouTube!
The company said that although ChatGPT has gained a large number of users in a short period of time, Character.AI is even more popular among users.
After a user sends their first message to an AI character, their average time on the platform jumps to over 2 hours.
Now users love the app and have created more than 10 million custom AI characters so far.
Character.AI has a bright future. As early as March this year, the company announced that it had received up to US$150 million in Series A financing, valuing its business at US$1 billion.
What is this concept? In other words, Character.AI has become one of the few AI unicorns on the same level as OpenAI.
What is the magic of this application, so that investment Why are people and users so obsessed? Let’s test it out.
In the mobile version launched at the end of May, the "virtual celebrities" created by previous users are already available for you to use.
The people waiting for favor include everyone from Musk to Socrates, from all walks of life, and for thousands of years.
Let’s just click on one and go in to hook up and have a look.
# This is really busy, and it is very simple.
But after all, there is a language model behind it. After just a few sentences, it will start talking to you back and forth.
Of course, the tone of the "Earth's Ball Leader" guidance work is still maintained.
Next, let’s try another celebrity, the “digital virtual human” of the richest man, Tony Stark.
It can be seen that he was not very happy about us asking "another him in the world" as soon as we came up. His speech was yin and yang from beginning to end.
After chatting with the big star, I found Xiao Zha lying quietly at the bottom of the list, so I clicked again Go in and prepare to tease him.
Before I said a few words, Xiao Zha’s attribute as Meta’s “first person responsible for the stock price” was activated by me. , began to desperately protect his company.
This tone seems to really make me angry.
Watching him defend himself and the company hoarsely, I was really touched.
Just this function of chatting with virtual celebrities made me feel quite good after just downloading and registering.
Looking at the names in this long list of categories in the list, basically the celebrities I want to chat with can find the corresponding AI to chat with for a while.
And based on the brief exchanges between the characters just now, the characters’ reactions are basically in line with their reputation in the public’s mind. positioning.
It can be said that for the vast majority of ordinary people who have not had the opportunity to talk face-to-face with these celebrities, this application does provide a "close contact" with celebrities or their favorite characters. "Opportunity.
The joy of developing a system
The trump card function that truly makes Character.AI continue to attract new users and keep old users addicted, In fact, it should be - self-created AI character chat, AKA AI cosplay.
In order to experience the Chinese capabilities of this application, the editor easily created two of his own AI characters.
The first character is very simple. I added a greeting in the introduction, and a football player AI who is about to join Real Madrid was generated.
For the second role, I decided to put a little more effort into it.
Recalling the protagonist in the first computer game I played in my life, "Cats and Dogs," I created him using the few fragments of memories I had about him.
#After creating it, I first try Had a brief chat with the football star.
Because I barely made any settings for this character, his reply was almost random to me.
The second sentence of the reply directly included "national curse", which caught me off guard.
After I added the character description, friendly, and likes to cooperate with teammates, the chatting style slowly started to become normal.
Then I started chatting for a while with the "game Bai Yueguang" that I was thinking about.
Sure enough, because my description was more substantial and specific, Lele was able to reply to a lot of details from the game in my memory.
The editor recalled the happy time when I was a child and played "Cats and Dogs" after finishing my homework, and I burst into tears.
Maybe this character will remain in my phone for a long time, and I will continue to fill it in when I think of some game details.
When you have time, you can chat with Lele and recall the adventure stories you experienced in Bucket Town.
Using "20% time", they created the prototype of a 1 billion unicorn
Character.AI was established less than 2 years ago and has raised nearly 200 million yuan US dollars, for a start-up to be so optimistic about investors, the identity of the founder must be a major factor.Yes, the founders of Character.AI, Noam Shazeer and Daniel De Freitas, are from the famous Google LaMDA team.
Noam Shazeer is a veteran of more than 20 years at Google.
Founders Noam Shazeer and Daniel De Freitas previously built LaMDA, a language model that supports conversational programs at Google. It was this model that triggered the subsequent uproar "Google programmers insist that LaMDA is conscious." Now that we have experienced the high-power baptism of large models such as ChatGPT, we have long been accustomed to this, but at the time, this incident made many people crazy.
In 2015, De Freitas, then a software engineer at Microsoft, read about scientists at Google Brain, Google’s flagship AI lab presented paper.
In this paper, the researchers introduced in detail a "neural dialogue model", showing how the machine can learn the art of dialogue by analyzing dialogue records of hundreds of movies .
Paper address: https://arxiv.org/pdf/1506.05869.pdf
When De Freitas read this paper, he was not yet at the AI Research Institute. At that time, he was a software engineer working on a search engine.
What he really wants is to take Google’s ideas to their logical extreme. "You can ask this robot to summarize. What it says doesn't look like lines in a movie."
In 2017, De Freitas joined Google and became an engineer at Youtube .
Google’s tradition is that everyone has a “20% time” project to explore new ideas after completing daily tasks.
In the “20% time”, De Freitas started building his own chatbot.
In order to train a neural network with a larger collection of conversations (such as a large number of chat records selected from a social network), he needs huge computing power.
Fortunately, as a Google engineer, he had some credits to run software on the company's network of computer data centers. But soon, the computing power was not enough, and he could only borrow points from other engineers.
As the system analyzes more and more data, the skills of the chatbot also improve by leaps and bounds.
At first, De Freitas used the LSTM neural network to train the robot, and then he quickly turned to the Transformer neural network developed by the Google team.
Unlike LSTM, which reads text one word at a time, Transformer can use multiple computer processors to analyze the entire document in a single step.
Soon, he created the chatbot Meena, and Google noticed his project and upgraded it to official status Research. Later, Menna became LaMDA.
Refuse to chop wood, leave Google, and reach the peak of life
After the "Engineer claims LaMDA is conscious" incident, Google was unwilling to release this technology because of concerns about misinformation and Toxic rhetoric can damage Google’s brand image.
So, Shazeer and Daniel De Freitas decided to leave.
It is reported that Google CEO has personally come forward to retain the two people, hoping that they will continue to work on the development of LaMDA, but there is no commitment to release the chatbot to the public.
So the two left Google at the end of 2021, and established their start-up company Character Technologies Inc. in November of that year and launched Character.AI.
In December 2021, the company raised $43 million in seed funding.
According to the founder’s LinkedIn introduction, he has participated in almost all major language model projects with Google over the past five years and has achieved outstanding results.
# This level of technology leaders came out to start a business, and naturally took money to get soft hands.
Institutions and individuals participating in the investment include A16Z, SVA, Paul Buchheit, the father of Gmail, etc.
Until now, the company team only has 30 people and is very busy.
Additionally, it announced a partnership with Google Cloud. Using Google Cloud’s tensor processor units, companies can build and train large language models faster and more efficiently.
Thriving Creator Community
In addition, Character.AI also has a rich creator community.Here, users can exchange their experiences in creating various AI characters with each other.
will also share its carefully created characters, such as a ghost warrior girl trapped in the modern world.
Cute repair hook.
##And my husband...
Unconsciously, users may deeply connect an experience of their own with the character they created.
This is why Character.AI has been emphasizing that it is the AI product with the highest user stickiness and longest usage time.
Because whether users are simply chatting or settling down to create their own AI characters, they will definitely invest a lot of time and energy on the platform.
And the AI characters supported by large-scale model technology seem to be really alive and interact with users, and have unique personalities.
With the deep connection established with users, users are reluctant to leave the platform and continue to contribute diverse content to the platform.
That’s it, Character.AI supports a unicorn-level valuation. ######
The above is the detailed content of The number of downloads in the first week crushed ChatGPT! A 20-year-old Google employee defected to create a second OpenAI?. For more information, please follow other related articles on the PHP Chinese website!

Running large language models at home with ease: LM Studio User Guide In recent years, advances in software and hardware have made it possible to run large language models (LLMs) on personal computers. LM Studio is an excellent tool to make this process easy and convenient. This article will dive into how to run LLM locally using LM Studio, covering key steps, potential challenges, and the benefits of having LLM locally. Whether you are a tech enthusiast or are curious about the latest AI technologies, this guide will provide valuable insights and practical tips. Let's get started! Overview Understand the basic requirements for running LLM locally. Set up LM Studi on your computer

Guy Peri is McCormick’s Chief Information and Digital Officer. Though only seven months into his role, Peri is rapidly advancing a comprehensive transformation of the company’s digital capabilities. His career-long focus on data and analytics informs

Introduction Artificial intelligence (AI) is evolving to understand not just words, but also emotions, responding with a human touch. This sophisticated interaction is crucial in the rapidly advancing field of AI and natural language processing. Th

Introduction In today's data-centric world, leveraging advanced AI technologies is crucial for businesses seeking a competitive edge and enhanced efficiency. A range of powerful tools empowers data scientists, analysts, and developers to build, depl

This week's AI landscape exploded with groundbreaking releases from industry giants like OpenAI, Mistral AI, NVIDIA, DeepSeek, and Hugging Face. These new models promise increased power, affordability, and accessibility, fueled by advancements in tr

But the company’s Android app, which offers not only search capabilities but also acts as an AI assistant, is riddled with a host of security issues that could expose its users to data theft, account takeovers and impersonation attacks from malicious

You can look at what’s happening in conferences and at trade shows. You can ask engineers what they’re doing, or consult with a CEO. Everywhere you look, things are changing at breakneck speed. Engineers, and Non-Engineers What’s the difference be

Simulate Rocket Launches with RocketPy: A Comprehensive Guide This article guides you through simulating high-power rocket launches using RocketPy, a powerful Python library. We'll cover everything from defining rocket components to analyzing simula


Hot AI Tools

Undresser.AI Undress
AI-powered app for creating realistic nude photos

AI Clothes Remover
Online AI tool for removing clothes from photos.

Undress AI Tool
Undress images for free

Clothoff.io
AI clothes remover

AI Hentai Generator
Generate AI Hentai for free.

Hot Article

Hot Tools

SecLists
SecLists is the ultimate security tester's companion. It is a collection of various types of lists that are frequently used during security assessments, all in one place. SecLists helps make security testing more efficient and productive by conveniently providing all the lists a security tester might need. List types include usernames, passwords, URLs, fuzzing payloads, sensitive data patterns, web shells, and more. The tester can simply pull this repository onto a new test machine and he will have access to every type of list he needs.

EditPlus Chinese cracked version
Small size, syntax highlighting, does not support code prompt function

Zend Studio 13.0.1
Powerful PHP integrated development environment

SublimeText3 English version
Recommended: Win version, supports code prompts!

PhpStorm Mac version
The latest (2018.2.1) professional PHP integrated development tool