The future of advertising after AI traffic coup

Large-scale language models are steadily replacing traditional searches by not only answering queries directly, but also converting the interface itself into a walled garden. This is an increasingly sought-after space that appears to replace traditional search engines, and is gaining more and more desperate grabs of land. Can publishers migrate content discoverability into the potential space of chatbots? Also, does AI chat make it a very attractive place when market capture triggers a monetization strategy?

opinion On Tuesday, an article in the Wall Street Journal noted the industry-wide slump in search traffic between news sites. This allows you to see the facts that a few minutes can afford to confirm for free.

While this coincides with a period of fierce growth in LLM use, it is difficult to empirically prove that this traffic slump is a direct effect of the rise in AI chat and the effects of Google’s AI summary, the relationship between the two phenomena has been reported as trivial.

News publishers have relied on content visibility within search results for decades, and now they need to watch their referral traffic decrease with the platform’s appeal to advertisers. For those who have been in business for decades and have hoped to suffer from ocean changes in another industry, such as the death of prints, this is exhausting and an incredible development.

The only downward traffic statistics are probably warning tremors. Once integration and market forces clarify the key AI players, new regions of commercial profits are finally known**, and the rest of us will have to buy or acquire some space in these new Citadels.

As we are tired of subscriptions in the Vulcanized market, we are resigning to an older ad-supported model, the interactive AI space is set up for mass mobility, one of the most disruptive shifts since the dawn of the internet.

Advertising space for potential space

CHATGPT users are currently not exposed to advertising in most cases. However, in a climate where subscription-based content returns to ad-supported models, terrain is fertile for chat-based advertising.

Naturally, the main players are pleased. At the end of 2024, Openai’s CFO Sarah Friar admitted that the company did not rule out advertising for the future of AI-based interfaces. By April this year, the company had already announced pending shopping facilities within ChatGpt. Furthermore, in line with the social network’s general business approach to leveraging customers as products, one company’s revenue forecast for April specifically included “free user monetization” as a pending revenue stream.

At Google Enclave, the search giant has already deployed a paid deployment integration within the PAGE AI summary content of search results, and has “very good ideas” for native ads in the Gemini AI chat environment.

Fake friends

While the decline in searches will focus, this week’s new paper also speculates what form of chat-based advertising will take compared to (for example) current experience with ChatGpt.

title Fake Friends and Sponsored Ads: Risks of Advertising in Conversation Searchthis paper comes from Jacob Erickson, an assistant professor of computer science at Vasser University in New York.

In it, Erickson modifies advertisers’ preferences for native ads over “Walled-Off” content. He provides some fictional examples manufactured for papers, including the most unpredictable formats of advertisers. Banner Ads:

Anyone who has dealt with an advertising partner will know that this is not what they want and certainly not something they are ready to pay for insurance: native advertisement,paid placement is deeply buried in the content, allowing users to confuse it with objective, agenda-free content.

As the paper observes, the above (simulated) example in which an AD-enabled AI chat environment presents antidepressant ads to vulnerable users raises many issues*:

“The concern here isn’t that antidepressant recommendations are inherently wrong. In fact, they seem to be a medical recommendation that could help support users, but could potentially disguise advertising.

“While there is disagreement about the harm in consumer drug ads in the United States, conversational search ads for such products can be made easy, especially for online consumer drug companies, without adequate medical needs when needed.”

He says that clearer ethical conflicts will arise when AI exploits vulnerable users’ weaknesses by suggesting useless or destructive solutions to sensitive disclosure.

Again, those familiar with the advertising world will recognize that placement fees tend to rise compared to products’ “ethically inflammatory” (alcohol, gambling, etc.).

The paper points out that chat-based AI platforms are familiar with user preferences, and that certain users know (for example) which soft drinks are preferred, making it difficult to know whether to discuss that product in chat or if it is a commercial opportunistic calculation act.^†:

This concern grows when chat-based advertising systems need to be able to quietly target users I already like it Certain products repeat hidden tactics that have led to the 2010s data scandal, fallout from Cambridge Analytica, and more recent scandals regarding preference-based election rigging in Romania.

Support Ads

Ethical and legal considerations aside, it is one of the biggest sources of friction against adoption. Any How likely is advertising in an LLM environment potentially reducing your user base? In fact, retention figures for streaming video platforms that introduce ad-based layers suggest that people withstand ads in an environment where they mark subscription-only access (rather than ad-supported access).

However, as we’ve seen, native and ring fence ads are very different suggestions in the context of AI chat. Rather than passively absorbing content, users are much deeper and interactive.

Anxious alliance

However, first of all, ads require a Lots of content A circulating medium. Some of the most notable AI chat platforms are currently engaged in expensive content round contracts with major platforms. For example, in 2024, Openai signed a contract with Rupert Murdoch’s News Corp.

This development is “settles” with organizations that cut down on AI that content companies may perceive as predatory, preventing lawsuits this year. But I ask at least three questions:

1) Is this kind of multi-year contract a “devil deal” designed to delay rather than prevent the end of an integrated news outlet, and is required by the need for AI to exchange search traffic that has been launched appropriately?

2) Is it a way instead to ensure that the content of the content becomes LLMS inference output and, together with hyperlinks, constitute a high-level “subscription” from the AI ​​company to the publications covered by these agreements?

3) From a more ominous perspective, would you merge commercially based news outlets as “true arbitrators” despite the inclusion of only major publications, such as the Wall Street Journal, tend to bend at the will of the advertisers? If so, can users experience “proxied advertising” as the commercial bends of the partner publications cited in AI chat are revealed even when the chat platform is not trying to promote it to users?

Feeding the machine

In the first scenario above, it is unlikely that AI companies are using licensed news content to remain limited to searched queries. Regardless of what the contract technically permits, the material ingested may inform future generation systems (AI systems that can collect and/or write news) that may compete with the publisher’s platform, or enable AI companies to recover their investments by offering more advanced AI-driven news products than publishers develop themselves.

So in this scenario, the newsroom becomes thin and empty, as most vigilant predictions fear. but, The institution remainsas a “authorized” brand, its reputation for fostering people will be of continuous promotional value.

Block the exit

In the second scenario, prioritizing links from “trusted” news partners in AI chat responses could attract scrutiny from regulators related to monopoly practices.

There are two other considerations for the second scenario: First, when users click away from the portal, they hate the platform (LinkedIn penalizes external links in user-generated content^††). Like Discord, ChatGpt forces users to “trust” outward clicks on other sites per domain, adding friction that can be argued to discourage users from leaving the chat.

From the AI ​​chat portal perspective, hyperlinks are the responsibility. It’s an unwanted tension in internal advertising that doesn’t bring money to the host. Therefore, warning pop-ups can remain and hyperlinks in chats can remain at the 6.75 points type.

Second, regulators and experts may be more interested in linking privileges than users. Despite the risk of AI hallucination, users essentially pay for chat platforms and read and research on their behalf. Therefore, when sent to the domain that sent knowledge, it may be an unattractive prospect for the average user who can develop the habit of trusting the digested version of LLM.

This means fewer or no clicks to proven domains, or at least much less than the age of search engines. In this scenario, only thinned newsrooms of partner companies can thrive.

Partial truth

Finally, when the objectivity of news platforms is under the largest and most obvious attacks in the history of publication, and when the biggest news platforms are subject to billionaire class eccentricity, news entities must also be considered business entities.

At every opportunity, many other partisan decisions can be envisioned in the news stream as platforms have been discovered to soothe advertisers (e.g. Bloomberg, NBC, BuzzFeed).

Some of them are guilty of omission. It is worth considering that what is not covered or score-covered in media outlets is a political decision, not based solely on the perceived level of interest of potential readers, but is frequently commercial or corrupt decisions.

With the large news networks and platforms within AI Citadel, there could be fewer voices without additional research into the end-user parts, as the general consensus of “common truth” is increasingly informed by LLM responses.

However, this aspect primarily reflects the current situation in which major outlets may cite small publications such as social media feeds and subsacs with appropriate authority and followers.

Each participating publication that contributed to the AI ​​discourse filters news through its own needs, but the AI ​​environment itself has its own editorial policy.

As for details, and what we consider “truth,” it’s quite a gauntlet.

* My conversion from author inline quotes to hyperlinks

** As evidenced by the various histories of Tumblr, Yahoo, Friends Reunited, StumbleUpon, Myspace, Betamax, Zip Drives, and many other “early leaders,” ChatGpt and Google should not be assumed to be leaders in the inevitable long-term market.

^† In fact, according to the FTC, much older questions about product placement in television shows also raise many questions about the integrity and interpretation of “advertising.”

^†† In addition to this, me and many other users have discovered that CHATGPT can be difficult to render clickable links. At the same time, GPT often appears to be overly disposed of to investigate the web without being asked. This may be because lag search burns tokens with low-cost subscriptions and allows for upselling scope. It’s so ironic that the GPT cutoff date was long before.

First released on Friday, June 13th, 202
Friday, June 13th, 2025, 12:21:38, correction of typos and one rephrase.