Character.AI Founders Return to Google

New Thread

Ctrl

Home

Discover

Introduction
Shazeer and De Freitas Return
Attention Is All You Need
Google DeepMind and Licensing Deal
Character.AI User Engagement

Character.AI Founders Return to Google

Curated by

dailies

2 min read

38,380

2,484

According to reports from TechCrunch and Reuters, Noam Shazeer, co-founder and CEO of Character.AI, is returning to Google along with co-founder Daniel De Freitas and select team members, as part of a deal that includes a non-exclusive licensing agreement for Character.AI's technology.

Character.AI CEO Noam Shazeer returns to Google - TechCrunch

techcrunch

Google rehires founders of consumer chatbot startup Character.AI

siliconangle

Noam Shazeer - Character.AI - LinkedIn

View 11 more

Shazeer and De Freitas Return

tribume.com

Returning to their roots, Noam Shazeer and Daniel De Freitas are set to rejoin Google after a three-year hiatus, during which they founded and led Character.AI

. The move comes as part of a broader agreement that will see Google licensing Character.AI's technology on a non-exclusive basis

. This strategic shift reflects the changing landscape of AI development, with Character.AI acknowledging the increased availability of pre-trained models and seeing an advantage in utilizing third-party LLMs alongside their own

. The reunion is particularly significant given Shazeer's prominent role in AI research, having co-authored the seminal paper "Attention is All You Need" in 2017, which introduced the concept of Transformers—a cornerstone of modern language models

2 sources

Attention Is All You Need

Research paper by Google that introduced the transformer architecture and revolutionized artificial intelligence

Authors

Eight scientists

Published

2017

Significance

Founding document for modern artificial intelligence

towardsdatascience.c...

The "Attention is All You Need" paper, published in 2017 by a team of eight Google researchers, introduced the groundbreaking Transformer architecture that revolutionized natural language processing

. This landmark work, which has been cited over 100,000 times as of 2024, proposed a novel approach to sequence transduction tasks that relied solely on attention mechanisms, eschewing recurrence and convolutions entirely

. The Transformer model demonstrated superior performance in machine translation tasks, achieving state-of-the-art results on English-to-German and English-to-French translation benchmarks

. The paper's impact extends far beyond its original focus, as the Transformer architecture became the foundation for modern large language models and has been applied successfully to various AI tasks, including question answering and multimodal generative AI

. Notably, the diverse team of authors, most of whom were born outside the United States, have since left Google to found their own AI startups, further influencing the field's development

5 sources

Google DeepMind and Licensing Deal

As part of the agreement, Shazeer will be joining Google DeepMind's research team, though his specific role remains undisclosed

. The deal includes a non-exclusive licensing arrangement allowing Google to use Character.AI's large language model (LLM) technology

. This strategic move is expected to provide increased funding for Character.AI, enabling the company to continue its growth and focus on developing personalized AI products for users worldwide

. The licensing agreement reflects the evolving AI landscape, with Character.AI planning to leverage both its own and third-party LLMs to enhance its product offerings

3 sources

Character.AI User Engagement

Character.AI has demonstrated remarkable user engagement metrics, particularly among younger demographics. The platform's unique approach to AI interactions has resulted in significantly longer session durations compared to other AI chatbots.

Metric	Value
Average daily time spent per user	~120 minutes
Percentage of users aged 18-24	57.07%
Average visit duration	29 minutes 30 seconds
Pages per visit	13.91
Bounce rate	20.87%

Character.AI's user engagement far surpasses that of other prominent AI chatbots. Character.AI users engage for approximately 120 minutes daily

. This extended interaction time is a testament to the platform's ability to maintain user interest and provide a compelling experience. The platform has a strong appeal among younger users, with 57.07% of its audience falling in the 18-24 age group

. This is significantly higher than other AI platforms; for comparison, only 27% of ChatGPT's traffic comes from this age bracket

. Character.AI's website metrics further underscore its engaging nature. Users spend an average of 29 minutes and 30 seconds per visit, browsing an average of 13.91 pages

. This level of engagement is particularly impressive when compared to ChatGPT's average visit duration of 7 minutes and 12 seconds with 4 pages per visit

. The platform's low bounce rate of 20.87% indicates that users find the content relevant and engaging, encouraging them to explore further

. This is notably lower than ChatGPT's bounce rate of 39.39%

. Character.AI's success in user engagement can be attributed to its unique offering. The platform allows users to interact with AI renditions of celebrities and fictional characters, as well as create their own AI entities

. This creative and interactive approach has resonated strongly with users, particularly the younger demographic, leading to high engagement levels and rapid growth in popularity.