Source: New Smart Element
Image Source: Generated by Wujie AI
Perplexity AI, an AI search company co-founded by a Chinese-American genius programmer, announced today that it has received a Series B financing of $73.6 million, and many big shots have come forward to support this star company.
Perplexity AI announced today that the company has received a Series B financing of $73.6 million, with a valuation of $520 million, and the total financing amount has exceeded $100 million.
This round of financing was led by IVP, with participation from institutions such as NVIDIA, Databricks, NEA, and others. Amazon boss Jeff Bezos and former GitHub CEO Nat Friedman also participated in the investment.
The founding team of three includes IOI gold medalists, genius programmers who have worked or interned at Google, OpenAI, and AI scientists.
Perplexity AI was established less than 2 years ago, and currently has a staff of no more than 40 people. Their headquarters is still located in a co-working space in San Francisco.
However, they have created a product that can challenge traditional search giant Google, with a current monthly active user base of over 10 million.
In 2023, they have already processed over 500 million queries, and the download volume on mobile devices has also exceeded 1 million.
Although compared to Google, the user base of Perplexity AI is still very small. However, the user value and experience they can provide have made many big shots optimistic about its future development.
Compared to the 10 links provided by traditional search engines such as Google, Perplexity AI can directly provide AI-processed answers to users, saving users the process of sifting through the vast sea of information.
By combining the search engine results with ChatGPT's summarization and question-answering capabilities, they have maximized the avoidance of potential issues that ordinary chatbots may encounter.
Moreover, in response to the AI-summarized results and answers, they can further provide follow-up questions through ChatGPT to help users actively inquire and dig deeper into the answers.
The founder of Perplexity stated that their advantage in response to user search queries is to directly provide answers using the most advanced AI tools, without the limitations that users may encounter when answers are provided by large companies, ensuring that users obtain the most direct and effective answers.
Big Shots Come Forward to Support It
In addition to the strong product capabilities and rapidly growing user base, the attention drawn to Perplexity AI's financing this time is also due to the participation of many big shots.
Bezos invested in Google in 1998, when Google had not yet figured out any business model and search traffic was not growing rapidly.
Therefore, his investment in Perplexity now seems to have a more important symbolic significance.
—Search is changing, and we are entering a new era.
NVIDIA also participated in this round of financing for Perplexity AI. Jonathan Cohen, Vice President of Applied Research at NVIDIA, commented on Perplexity AI as follows:
While enterprise-to-enterprise and developer-related artificial intelligence capabilities are exciting, artificial intelligence will also change the way we obtain information. Perplexity is at the forefront of this transformation and is one of the few consumer artificial intelligence products that can achieve this goal.
In addition to Bezos, many former Google employees have also invested in Perplexity AI in previous financing rounds:
Paul Buchheit: Inventor of Gmail
Susan Wojcicki: Former CEO of YouTube
Ashish Vaswani: Author of the Transformer paper
Andrej Karpathy tweeted his congratulations: "For a consumer product, being able to be used by users every day and being willing to pay monthly for it is already a high standard."
Jim Fan, Head of OpenAI developers and scientist at NVIDIA, and other industry big shots have also sent their congratulations.
Yann LeCun congratulated CEO Denis Yarats and Aravind Sriniva, who had studied for a Ph.D. at NYU and worked at FAIR.
Aravind Srinivas also expressed his gratitude:
We thank you and New York University for providing us with lab space for our first whiteboard and hackathon meetings. We have fond memories of wandering around the campus and doing our Twitter search demo in your office! Also, thanks to Soumith Chintala and LLama 2 from Meta!
Soumith Chintala, the founder of PyTorch, also has high hopes for the future of Perplexity AI and tweeted his congratulations.
By the end of 2023, Perplexity AI has become my most frequently used AI application. I use it to search for factual questions—including recent news/facts, summaries of product opinions and suggestions, and more.
ChatGPT + Browsing can do similar things, but it is 100 times slower and is usually random.
Founding Team
Similar to the founding teams of many previous star companies, Perplexity AI was founded by the following three individuals: Aravind Srinivas, Denis Yarats, and Johnny Ho.
Although the three of them look very young, they are all outstanding figures in the field of AI, including genius programmers and academic experts.
Johnny Ho is the co-founder and Chief Strategy Officer of Perplexity AI.
He is responsible for the development of ranking and backend systems, building key components for news sources and summary emails, and launching and expanding new parallelization systems.
Before founding Perplexity AI, he worked at a quantitative trading company.
Prior to that, he was an engineer at Quora, the largest online Q&A platform in the United States.
Johnny started programming at the age of 12. In 2012, while still in high school, he won the championship at the International Olympiad for Informatics (IOI), defeating Gennady Korotketvich, a six-time gold medalist from Belarus.
Johnny achieved a perfect score at the 2012 IOI, a rare occurrence in the 25-year history of the competition.
In 2013, Johnny began working at Quora, and was later admitted to Harvard University to continue his studies and research in information technology and mathematics.
Johnny stated that the key to winning the competition was "familiarity with all available algorithms," and he also won a gold medal at the ACM-ICPC in 2016.
Aravind Srinivas is the co-founder and CEO of Perplexity AI.
Before founding Perplexity AI, he worked at OpenAI and Google.
Aravind Srinivas graduated from UC Berkeley and interned at OpenAI, DeepMind, and Google.
Later, Aravind Srinivas worked at OpenAI, researching language and diffusion generation models along the lines of Stable Diffusion and DALL·E 3.
Denis Yarats is the co-founder and Chief Technology Officer of Perplexity AI.
Denis Yarats holds a Ph.D. in artificial intelligence from New York University, and has worked at Microsoft and Quora, and served as an advisor to Alessandro Lazaric at Facebook AI Research.
He has published multiple papers in the field of machine learning and has open-sourced as many as 37 GitHub repositories.
Making Strides
According to statistics, the website and mobile network of Perplexity received 45 million visits in December.
CEO Aravind Srinivas stated that the advantage of Perplexity lies in focusing on fine-tuning various performance-optimized AI models, rather than locking in on a single model.
Unlike traditional search engines, Perplexity provides a chatbot-like interface that allows users to ask questions in natural language—such as, "Do we burn calories while sleeping?" and "Which is the least visited country?" and more.
The artificial intelligence on the Perplexity platform responds with a summary containing source citations (mainly websites and articles), at which point users can ask follow-up questions to further delve into specific topics.
Srinivas stated that with Perplexity, users can immediately obtain answers to any question, including complete sources and citations. Perplexity is suitable for anyone using technology to search for information.
The foundation of the Perplexity platform is a series of internally and third-party developed GenAI models.
Subscribers to Perplexity's Pro plan (at $20 per month) can switch between models—Google's Gemini, Mistra 7Bl, Anthopic's Claude 2.1, and OpenAI's GPT-4.
Pro users can also use features such as image generation, and have unrestricted access to Perplexity's Copilot (which considers personal preferences during the search process).
Additionally, Perplexity allows Pro users to upload documents, including images, and lets the model analyze the documents to formulate relevant answers (e.g., "Summarize pages 2 and 4").
Perplexity also promptly follows up on model updates from various providers.
For example, Perplexity previously used the text-davinci-002 for prototyping, which resulted in a certain degree of hallucination (1/2~1/3).
After OpenAI released text-davinci-003, Perplexity will launch a new version that significantly reduces hallucination (1/10).
Srinivas believes that the search filtering and discovery options provided by Perplexity are more powerful than those of most companies. For example, users can restrict searches to academic papers or browse popular search topics submitted by other users on the platform.
"With Perplexity, there is no need to click on different links, compare answers, or endlessly dig for information. The era of filtering SEO spam, sponsored links, and multiple sources will be replaced by a more effective mode of knowledge acquisition and sharing, driving society into a new era of accelerated learning and research," Srinivas said.
Furthermore, Perplexity's ambitions go far beyond search. It has already begun training its own GenAI models and using Perplexity's search index and public network to improve model performance.
Srinivas said, "Google will be seen as traditional and outdated, while Perplexity will become the next generation and the future."
Perhaps in the future, this startup company can truly reshape the new competitive landscape in the search field using AI.
Reference:
https://www.wsj.com/tech/ai/jeff-bezos-bets-on-a-google-challenger-using-ai-to-try-to-upend-internet-search-0859bda6?mod=followamazon
免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。
