Will search engines be the next big revolution in generative artificial intelligence? While Google and Bing are developing their own AI-powered models, Gemini and Copilot respectively, young start-ups are also developing in the sector. But one of them is particularly talked about at the start of 2024: Perplexity.
Perplexity was founded in July 2022 by Aravind Srinivas, an Indian-American engineer, himself formerly of OpenAI and Google. At the start of 2024, the start-up announced that it had completed a new fundraising round of nearly $73.6 million, with renowned investors, such as technology giant Nvidia, the former vice-president of Twitter Elad Gil, or even Jeff Bezos and former YouTube boss Susan Wojcicki. Enough to bring the company’s valuation to more than $500 million. Still far from the estimated 80 billion of OpenAI, but, for a start-up which has not even celebrated its two candles yet, the performance remains notable.
“Response Engine”
It is above all the concept of Perplexity which could well mark a new turning point in the place taken by AI. Aravind Srinivas says it himself: his company, which boasts nearly 10 million active users each month, wants to be a “response engine” and no longer just a search engine. So, instead of writing a search and being offered a series of links answering it, as we all have the reflex to do, Perplexity directly provides a concise and clear answer. All while citing several sources, whether media or even scientific studies.
“The traditional approach to search was to return 10 blue links, which the user then had to navigate to find the information they were looking for. Over the past few years, this model has evolved to provide direct answers to users’ questions. This is what I mean by ‘answer engine’: users can ask any question directly and receive an actual answer, not just a list of web pages that may or may not contain the answer.” explained the CEO of Perplexity in an interview with Forbes, in September 2023. This “response engine” makes it possible to limit the generation of results to these scientific studies, providing security against the risk of misinformation. Finally, the user is suggested other questions related to their query, in order to explore the subject in more depth.
Despite these options, Perplexity is not free from criticism for the moment. If we take the Franco-French example of the government reshuffle, Perplexity always tells us that Gabriel Attal is Minister of National Education and that Aurélien Rousseau, that of Health… All justified by press articles dating back of several months. The platform itself warns: “It should be noted that political information is subject to change, and it is recommended to check official sources for the most recent information.”
An economic model to be confirmed
When using Perplexity, the resemblance of the interface to that of ChatGPT is obvious. Far from being a coincidence: the search engine uses artificial intelligence from OpenAI to feed its responses, but also those from Google and Anthropic. Perplexity is developing its own generative artificial intelligence model, but readily acknowledges that ChatGPT is currently the industry benchmark. “I expect OpenAI to have the best models for at least another two or three years. Nobody knows the future after that… You just have to be pragmatic. […] We are very happy to work with their model,” explained Aravind Srinivas in an interview last July.
Perplexity is still far from being profitable, firstly because the platform can currently be used (almost) completely free of charge. There is still a premium offer, at 22 euros per month, allowing you to choose which generative AI to use between ChatGPT, Gemini (Google), Claude (Anthropic), Perplexity, or even to have access to other services, like image generation. The start-up also faces very strong competition in the sector, whether from other start-ups such as You.com, which itself recently raised more than $45 million, but also from tech behemoths, which are developing their own AI models at high speed.
The field is also particularly expensive, and the costs of running and enriching artificial intelligence remain extremely high. Enough to put into perspective the fundraising and the current valuation of Perplexity, obviously impressive but far from being gigantic when we take into account the operating and research costs of a company in this sector. But, for Aravind Srinivasthis is not an insurmountable obstacle to conquering the market, on the contrary: “Having less funds can actually be better, because you have even more motivation, you need to win.”