2024.04.03 03:29
I'm PortAI, I can summarize articles.

Perplexity is going to do auction ranking. Is the end of the big model still advertising?

Despise Google, understand Google, become Google

Do you still remember Perplexity, which once raised $73.6 million from NVIDIA and Bezos, vowing to challenge Google with an AI-driven transparent search experience and capturing a large number of community supporters?

Now, this Silicon Valley star AI company, which has always despised the advertising business, is also preparing to sell ads.

As a conversational answer engine, Perplexity combines chatbot and search engine functions. Users can directly ask questions in natural language and get text answers, related images, and videos summarized by a large language model from the web. Perplexity also provides precise citation links in replies to eliminate false information and AI illusions. It also suggests related questions that users may be interested in, guiding them into deeper exploration.

And it is this "related questions" that accounts for 40% of the total query volume and supports one of its product features, which will be the first ad space opened to sponsors by Perplexity: When users further explore a topic, Perplexity will add sponsored questions next to organic questions, and continue to display related ad links after further inquiry.

This is what people often "complain" about, but it is the bidding ranking that search engines have always relied on to make money.

According to the company's Chief Business Officer Dmitry Shevelenko, this feature will be launched in the coming quarters, with no further details available at this time.

The news has left the AI community speechless - wasn't this what you said in the beginning?

Ideals are lofty, but ads may be more lucrative

Founded in 2022, Perplexity's initial idea was to use advanced AI large language models like GPT-4 to break the keyword search monopoly held by Google and Bing for over twenty years, allowing people to receive real, accurate, and trustworthy answers by simply asking questions, without getting lost in the dazzling ads and algorithm-optimized fragmented content.

This product concept, which places users rather than advertisers at the core, was also very popular at the time, rapidly climbing various platform download charts. It garnered many fans, including former GitHub CEO Nat Friedman, with people abandoning Google and switching to Perplexity as their preferred platform for obtaining information.

Perplexity also became synonymous with the democratic revolution of search engines and once introduced itself on its homepage as: "**Searching for information should be a direct, efficient experience, unaffected by an ad-driven model The reason we exist is because in the noise of information overload, there is a clear need for a platform to provide precise, user-centric answers, especially in such a valuable time."

However, when faced with questions from foreign media, Shevelenko said, "Advertising has always been a part of our excellent business."

Sharp-eyed netizens also noticed that the phrase "not influenced by an advertising-driven model" has been quietly removed.

This familiar scene is somewhat similar to Mistral AI's actions after forming a partnership with Microsoft. At that time, Mistral AI was also found to have removed the mission description "committed to open models" from the website, presumably to avoid controversy. Later, they added this sentence back with the modifier 'Mistral 7B and Mixtral 8×7B'.

However, this time Perplexity is going to directly sell advertisements. Isn't this a bit of a contradiction to their initial proud and unique founding mission?

The company's founder and CEO Aravind Srinivas has publicly stated that he wants to create a product that is "precise, authentic, academic, and enhances the knowledge capital of the earth." He has also mentioned that for traditional search engines like Google, "improving the generative search experience may harm its very important source of advertising revenue. To protect vested interests, the ability to provide direct answers must be restricted." Therefore, competing with Google is "completely feasible" for Perplexity.

But it seems that as operations progress, Srinivas's views are gradually changing. In a recent interview with Wired magazine, he mentioned that he is not against advertising. In fact, his outlook on advertising on the platform is to help advertisers understand who is searching and then bid for "high-value traffic."

Once a firm advocate of a refreshing ad-free AI search experience, now tells you, "Advertising is not evil. When done well, it is great. Generative AI will help us build better positioning." Will the public buy it?

Some users have harshly criticized, bluntly stating that Perplexity has been marketing itself incessantly with "replacing Google" and "better user experience" in the past, and now it is also damaging its reputation by falling into the rabbit hole of advertising Some visually impaired people have discovered the key point: "We have closed the loop."

"My mom told me many years ago, there is no free lunch in the world."

Some directly pointed out that this is the stark reality of all startups fighting against giants, "You can't defeat Google just by relying on no advertising and focusing on privacy."

"Everyone thinks they can replace Google until it needs to consider making money and growing."

The following netizen belongs to the idealistic group, although feeling disappointed, still believes in the importance of a search engine that is completely ad-free and unbiased.

In contrast, the comment from this reader of The Verge is more objective, "Honestly, as long as it doesn't affect the core experience, I don't oppose this practice, and they will continue to provide an ad-free paid 'professional' version. I understand that operating AI services is very expensive, and the money faucet will eventually have to be turned on. The cash reserves of venture capital can only burn for so long, and will eventually be depleted."

This also reveals the reality that most AI startups face: while the ideal of purely serving users is beautiful, making money to eat and support a family is also a threshold that must be crossed. Securing financing is a market recognition of a company's value, but sometimes it is also like a double-edged sword, facing investors' expectations of returns, there will inevitably be changes from decision-making power to business strategies.

The end of large models is also advertising

In fact, various commercial adjustments by AI companies are nothing new.

The most well-known is the industry leader OpenAI, from a non-profit research laboratory with a few founders at the beginning, to today's AI empire valued at over $80 billion. In this industry where burning money is more competitive than technology, OpenAI has continuously raised funds, aggressively developed, deployed in large enterprises, entered the semiconductor and AI hardware sectors, and completed its magnificent transformation In the midst of the farcical turmoil in the leadership upheaval, after the various smoke bombs dissipated, people found that the real decision-maker was still the capital giant Microsoft.

Mistral AI, which was brought under Microsoft's wing as the "second leg", is also in a similar situation. Originally with the mission of "leading the open model revolution", facing AI competitors like OpenAI and Google who chose the closed-source route, it was once praised by the developer community as "the most sincere and true-to-heart team on Earth". Later, it suddenly became the junior brother of OpenAI. The promise to "provide an open-source option that is infinitely close to GPT-4 for enterprises" also turned into the paid closed-source Mistral Large.

In addition, Hugging Face, the open-source community known as the GitHub of the artificial intelligence world, has chosen a strategy of "open-source driving business", attracting users and developers through open-source projects, and then generating profits through membership fees, enterprise data hosting, and more.

Recently listed on the US stock market, the social aggregation platform Reddit, although not an AI company, has found new monetization methods through authorizing platform data for large-scale model training.

Some netizens have compared the news of the free online real-time calling software Discord preparing to display commercial advertisements with Perplexity, saying, "Advertising is the most effective way to monetize large-scale consumer services. Companies like Perplexity and Discord that boast they will never run ads will eventually quietly update their stance when investors demand revenue growth, even though it looks stupid.

The era of zero advertising revenue is over. It's time to make money."

"In the end, everyone will eventually return to advertising."

Indeed, in the era of massive potential for large models, the path to technical monetization is often long and winding. Faced with high development costs and ongoing operational expenses, advertising remains the most direct solution to the commercialization challenge.

For Perplexity now, incorporating advertising into future product plans requires a sufficiently large user base to attract marketers' interest, while ensuring that sponsored content is relevant to the platform's core usage flow without disrupting it. Clearly, the current 10 million monthly active users still have a lot of room for improvement, and attracting more users while attracting advertisers is quite a challenge in all aspects.

In any case, the commercialization of AI companies is an irreversible trend. Finding a balance in this will be a common challenge that all companies need to face in order to go further It seems that Perplexity has figured it out. Before exploring becoming a better Perplexity 2.0, it chooses to survive first.

Author: Zhang Xiaoxue Source: Silicon Star Pro, Original Title: "Perplexity is going to do bidding ranking, the end of big models... is still advertising?"