By Dr. Tawhid CHTIOUI, Founding President of aivancity, the leading institution for AI and data
Selected by Keyrus as one of the 25 most influential global figures in the field of AI and data (January 2025)
Why ChatGPT Is Outperforming Google – The Audio Debrief
The Day the Raincoat Killed the Blue Tie
Paris, on a rainy Saturday morning. Lina types “black waterproof hiking jacket, budget €50”… or rather, Lina says this to her phone. In a matter of seconds, ChatGPT suggests three models in her exact size—information it has based on her previous purchases—and notes that a fourth is only available in a color she doesn’t like. It also displays a comparison table of delivery times, a direct link to the cheapest seller, and a reminder that her birthday is coming up, meaning a gift card is sitting in her digital wallet. No results pages, no yellow “sponsored” banners, no friction. By the end of the conversation, Lina never saw a search engine logo: she chatted with an assistant, then made a purchase. This seemingly innocuous act may signal the quiet end of an era: the web will no longer be a place to explore, but a conversation partner to consult. The blue link, an icon of the Google years, is wavering…
The End of the “Click Pact”: The Silent Earthquake
For nearly twenty-five years, our browsing of the web has been based on an unspoken agreement: the search engine acts as a zealous librarian who organizes pages, ranks them, and then leaves the choice up to us. We click, we compare, we forge our own path; the digital giants, for their part, officially limit themselves to ranking. This division of labor, inherited from Google, has shaped the attention economy: influence grew as one climbed the results page, but the final decision remained in the user’s hands.
The arrival of ChatGPT Search abruptly breaks this pact. The conversational assistant no longer merely lists results; it filters, summarizes, argues, and recommends. In just a few sentences, it transforms the act of searching into an act of reception. The era of “click and see” gives way to “here’s what’s right for you.” And in this subtle shift, the balance of power shifts: whoever filters now holds the key to our economic, cultural, and even political decisions. In the past, the search engine distributed visibility; now, the agent distributes hyper-personalized and prescriptive answers.
The web, once a place of exploration, has become a source of advice. We no longer search; we are served.
This is why we must speak of a revolution rather than incremental improvement. Just as the printing press shifted the center of gravity of knowledge from the monastery to the reader, generative assistance today shifts the center of decision-making from the researcher to the algorithm. What is shifting is not the speed or relevance of a query; it is the very structure of authority over information. Turning the page on the search engine, literally, amounts to redrawing the map of digital power.
Three Major Shifts: The Newtonian Era of the Web Is Underway
1. From research to response
In the old days of the web, search engines functioned like giant billboards: they plastered dozens of links onto a virtual wall and left users to figure out their own “to-do list.” Since ChatGPT Search, the ritual of “scroll, analyze, click” has evaporated; the results page has transformed into a continuous dialogue. We no longer hop from site to site; we converse with a single narrator who is already piecing together the puzzle. This shift eliminates the micro-cognitive fatigue that arose with every tab change and creates a deceptive sense of comfort: the answer arrives ready to use, tailored to each user, sometimes accompanied by a price comparison and a purchase link.
The metaphor is then reversed: whereas Google was like a library whose endless shelves demanded curiosity and perseverance, ChatGPT is more like a concierge at a luxury hotel.
It doesn’t just point you toward books; it brings you the book itself, opened to the right page, with a highlighter marking the key sentence. It’s not a more powerful search engine; it’s a personalized recommendation service—in other words, an entirely new business model that shifts the focus of value creation from indexing to recommendations.
This change might have remained a mere footnote had it not, almost instantly, taken the world by storm. Three days after its global launch, ChatGPT’s “shopping” feature was already processing over a billion weekly queries: a statistical tidal wave confirming that the shift is not a ripple among early adopters but a mass phenomenon. When millions of internet users swap active searching for guided content delivery, it’s not just habits that are changing; it’s the very way information flows, how purchase intent is formed, and how economic power is redistributed.
2. From Proclaimed Neutrality to Algorithmic Intermediation
For years, search engines presented themselves as mere librarians: they cataloged, organized, and let users navigate freely among the shelves of the web. With the arrival of ChatGPT Search, this stance of apparent neutrality is shifting. The conversational agent no longer merely indexes information; it guides the user’s focus. When a user makes a purchase request, the products highlighted are drawn from structured feeds provided by giants like Shopify, Amazon, or Zalando, then sorted using semantic reasoning that takes into account history, preferences, real-time availability, and even seasonality.
In other words, the assistant becomes a commercial mediator, a sort of invisible control tower that decides which options will—or will not—cross the threshold of our consciousness.
OpenAI insists that its recommendations remain “ad-free and commission-free”; the press release issued at the April 28, 2025, launch emphasizes the “independent” selection of results. The promise is appealing, but it is fragile. As user numbers skyrocket (400 million active users in a single week), the temptation to monetize this algorithmic eye will grow. Observers have already dubbed it the “golden gate”: a gateway so valuable that it would be naive to imagine that no commercial pressure will ever distort it. Reddit forums immediately sensed the danger: “Enshittification is inevitable, ” warn users, alluding to the fate of other platforms that were once virtuous and are now saturated with ads.
Beyond the fear of a “pay-to-recommend” future, there is a question of informational sovereignty at stake. When the agent captures the purchase intent before it reaches the marketplaces, the latter lose the power to attract visitors on their own merits.
The debate is no longer “Who owns the data?” but rather “Who writes the first line of every desire?”
Brussels is under no illusion: following the AI Act of 2024, European regulators are already grappling with the risks of discrimination and market concentration associated with “these high-risk intermediation systems.” As long as the code remains opaque, society delegates, almost blindly, the choice of what it will see and buy to a confidential algorithm.
And so the second major shift takes place: the web is no longer a public square where everyone sets up their stall and attracts customers; it becomes a stage where AI is the stage manager, raising the curtain and directing the spotlights as it pleases. For brands, the crucial question is no longer whether they can be found, but whether they will be retained by the algorithm. For citizens, the question is even more dizzying: what do we lose, collectively, when human curiosity gives way to an invisible filter?
3. From Advertising Posters to Performance-Based Contracts
For two decades, the web economy ran on the same mechanism: AdWords auctions. Advertisers paid to appear; they paid again when a user clicked. This “cost per click” (CPC) has made Google a fortune: over $175 billion in advertising revenue in 2023 according to Statista, with an average click-through rate of around 6% in the top-performing sectors.
ChatGPT Search is a game-changer. Here, advertisers don’t pay to be seen; they pay only after an actual sale—a “pay-per-conversion” model modeled after affiliate marketing, but integrated for the first time into the very heart of the search experience. OpenAI remains, for now, in the “no ads, no commissions” camp, Sam Altman notes, while saying he is open to “tasteful ads” based on a percentage of the transaction rather than on the purchase of keywords. Internally, the company is already reporting click-through rates of 15% to 20% on product cards—three times the performance of a traditional sponsored link— a gap that, on the scale of a billion weekly queries, becomes an economic earthquake.
For users, the immediate result is the disappearance of the “advertising fog”: they no longer see garish ads, only seamless recommendations woven into the conversation. But this visual streamlining has a downside: by drying up the advertising revenue that fuels blogs, media outlets, and forums, it threatens an entire ecosystem built on search engine traffic. Already, some publishers are seeing drops of 12 to 18% in organic visits to “shopping guides” since the feature’s global rollout—a faint signal that foreshadows a drastic redistribution of revenue. In other words, ChatGPT isn’t just replacing search; it ’s shifting the cash flow. Where Google sold visibility, OpenAI may one day sell certified conversions, collected at the source from every purchase.
What will become of the free web once the ad is gone? What will become of the media, blogs, and forums if traffic dries up?
The issue goes beyond a mere rivalry between two giants: it concerns the financial survival of independent journalism and, by extension, the very vitality of the digital public sphere .
The Interface War: The Reflex That Will Shape the Web
Google didn’t wait for OpenAI’s onslaught to bring out the big guns. Since March 2025, the company has been testing AI Mode, an integration of itsGemini 2.0 model onto the framework of its legacy search engine. The official goal: to revitalize search through multimodal generative responses. In the demos, users get an illustrated summary before they even see the blue links. But the transformation is proceeding cautiously: Google must protect its $175 billion in advertising revenue—57% of which comes from text ads, still billed per click. Cutting back on the results page too quickly would be like sawing off the branch that feeds the tree; slowing down risks losing the user.
Added to this economic equation is legal pressure. Since April, the company has been facing a remedies trial in Washington to determine the penalties resulting from the 2024 antitrust ruling. Every new generative feature is being scrutinized by regulators: Google must prove that it is not stifling competition or the diversity of sources.
While the California-based company treads carefully, OpenAI moves forward with ease. With no advertising portfolio to protect and no technical legacy to maintain, the startup can reshape the search experience without any cannibalistic pain. The result is evident in usage patterns: according to Bain & Company, 80% of U.S. consumers already use an AI chatbot for at least 40% of their queries—a trend that could cost Google up to 25% of its organic traffic within two years.
In other words, the clash is less technological than cultural: which gesture will become second nature? Typing“www.google.com” or “openingachat”? If the reflex to chat prevails, the index finger’s dominance could crumble faster than any battle for market share could have foreshadowed.
From SEO to CAIO: The Invisible Shockwave
As content replaces links, a whole sector of the digital economy is beginning to falter.
The first potential casualty: SEO. Optimizing a page to rank “number one” on Google no longer makes much sense if the assistant no longer returns search results but instead provides a pre-digested summary. Agencies are already talking abouta new holy grail: CAIO—Conversational AI Optimization. It’s no longer about stuffing text with keywords, but about structuring clean “product databases,” exposing metadata readable by LLMs, and opening APIs capable of feeding the future digital concierge in real time. In other words, it’s a technical and philosophical undertaking: visibility is no longer earned through an indexing algorithm, but at the negotiating table with a conversational agent.
This shift raises a question that is less technical than moral: what becomes of desire when an algorithmic third party anticipates it so well that all we can do is acquiesce? Hannah Arendt , the German-American philosopher, reminded us that the capacity to judge is developed through the process of deliberation.
If AI pre-selects our choices for us, the risk is not only the homogenization of tastes, but a gradual atrophy of our ability to make preferences—the most political of all human faculties…
Added to this is the danger of the “shopping bubble.” When a model maximizes personalized relevance, it can trap the shopper in a predictive tunnel where only statistically likely products make it through. Discovering an under-the-radar artisan or choosing an item that surprises us then requires an act of rebellion against convenience. By claiming to broaden the selection, conversational commerce could, paradoxically, narrow the diversity of the shopping cart. The prescriptive web is not neutral: it pushes, subtly but firmly, toward the center of gravity of probabilities. It remains to be seen whether we will accept this comfort as progress… or as a new form of invisible control.
When Recommendations Devour Search: A Brief History of Cannibalistic Interfaces and Glimpses of the Immediate Future
Every digital transformation follows essentially the same pattern: a new interface promises to reduce effort, then absorbs attention to the point of marginalizing the tool it replaces. The road map has given way to GPS; today , 93% of American drivers say they are dependent on their satellite navigation and admit they would be “lost without it.” The curated homepage has been devoured by the social feed, which now monopolizes more than a third of our time online—2 hours and 23 minutes per day. Tomorrow, the chatbot could well spell the end for the search engine: the same logic of cognitive economy, the same erosion of our autonomy to explore.
The historian Milad Doueihi had already described the“click culture”;
we are now entering the era of the “reply culture.”
This phenomenon isn’t abstract; it’s already taking shape in our daily routines, sketching out a future where the engine fades into the background behind a thin layer of algorithmic orchestration. First, imagine the driver in a hurry: in the middle of the A6, he murmurs, “Find me an available charging station within thirty kilometers and order a coffee to go.” The assistant makes the reservation, pays, adds the gas station as a stop, and adjusts the GPS without further ado. Next, picture the semi-automatic wardrobe: every morning, a connected mirror communicates with ChatGPT, takes stock of your closet, suggests outfit combinations, and plans any returns; no need to “go shopping” anymore—the store comes to you. Finally, let’s apply this capability to B2B: the purchasing department of a restaurant chain issues the following request: “10,000 biodegradable containers, 750 ml size, delivery in five days, budget €12,000.” The AI agent queries several certified suppliers in real time, negotiates volume-based discounts, verifies certifications, compares delivery times, and automatically drafts the order for digital signature. In less than three minutes, the supply is secured, the tracking number sent: the purchasing department has literally just completed its mission… through conversation.
In these three scenes, we search in vain for the results page. The search engine hasn’t disappeared because it’s no longer useful; it has dissolved into the interface. What remains is the prescriber, an intelligence that, drawing on our data, orchestrates the world so that it matches our query even before we’ve fully formulated it. To the question “Where are we going?”, the answer might well be: wherever the assistant has directed our desire.
Ethical Issues in Total Assistance: Between Opacity, Dependence, and Autonomy
The more the chatbot becomes the sole gateway to information and purchases, the thicker the fog of uncertainty surrounding it grows. The first cloud: opacity.
OpenAI showcases its capabilities but remains tight-lipped about its inner workings; worse still, several organizations affiliated with the group have backtracked on their transparency commitments, as revealed by Wired in 2024. Yet trust requires knowing whether a recommendation is genuine or influenced by commercial interests. U.S. regulators have understood this: in early 2025, the Washington State AI Task Force placed transparency at the top of its legislative priorities, a sign that tensions are rising on both sides of the Atlantic.
Next comes dependency. By combining our health history, expenses, and leisure activities, an assistant will soon cover more ground than a financial advisor or a family doctor. For many, switching providers would be like suffering from digital amnesia. This imbalance fuels mistrust: according to Big Valley Marketing, 80% of users say they are reluctant to trust opaque AI. Yet the promise of convenience remains irresistible, and adoption continues to climb.
The third issue concerns sovereignty. The European Union is no longer hiding its concern: the “AI Continent Action Plan” presented in April 2025 emphasizes the need to “ensure the strategic autonomy of conversational interfaces” and to mandate a minimum level of interoperability among competing agents. Because when the assistant captures the purchase intent before it reaches Amazon or Google, it redraws the map of economic—and political—power.
What this means for each of us
For consumers, the promise is enticing: fewer clicks, less hesitation, everything served up on a silver platter. But this “seamless” efficiency threatens curiosity—that intellectual muscle that is exercised through aimless browsing. Reserving the “right to digital wandering” is becoming almost an act of mental hygiene.
On the business side, visibility is now played out behind the scenes: the assistant’s “brain” must be fed with flawless catalogs, robust APIs, and LLM-compatible descriptions. The battle is no longer fought on Google’s first page, but in the data pipeline that fuels the recommendation. Conversational AI Optimization specialists are already replacing yesterday’s SEO consultants.
Finally, for society, the challenge is structural. The free web relied on advertising revenue; yet ChatGPT’s “pay-per-conversion” model promises to siphon off a portion of those more than $600 billion in global advertising spending. If this flow is diverted away from the media, what economy will sustain the press, independent culture, or even Wikipedia? The chatbot, by claiming to simplify our lives, could drain the public sphere that made it so rich. The choice, from now on, is no longer just between comfort and security; it raises the question of the very vitality of our information ecosystem.
Conclusion – Epilogue — The Day the Click Fell Silent
Larry Page once remarked that a “perfect search engine” would be one that answered the question before it was even asked. He could already foresee the end of the famous blue link, though he lacked the missing piece: an intelligence capable of generating sentences, not just ranking pages. Two decades later, OpenAI slipped that piece into place, and the puzzle came together with the soft sound of a successful revolution.
Few revolutions take place without barricades. Here, everything seems seamless: we speak, the AI responds, we nod in agreement. Yet with every request turned into a response, a precise measure of power shifts hands—from the human explorer to the algorithmic architect. The day Lina ordered her rain jacket without leaving her chat window, a page—literally—turned…
So the real question is no longer “Will Google survive?” but rather: Are we willing to let AI write the first paragraph of every decision we make?
As long as we preserve the right to throw the agent off balance, to scroll aimlessly, to search simply for the pleasure of wandering, conversation will remain a democratic space rather than a corridor. It’s up to us to remind ourselves, from time to time, that the unexpected is still the best driving force.
References
- Reuters – “OpenAI rolls out new shopping features with ChatGPT search update,” April 28, 2025.
- Wired – “OpenAI Adds Shopping to ChatGPT in a Challenge to Google,” April 29, 2025.
- Google Blog – “The latest AI news we announced in March,” April 4, 2025.
- NPR – “Why Google’s search engine trial is about AI,” April 29, 2025.
- Wired – “OpenAI Highlights New AI Safety Research. Critics Say It’s…”, 2024.
- Washington State AI Task Force – Legislative Priorities for AI Transparency, 2024.
- Bain & Company – “Consumer Reliance on AI Search Results Signals a New Era of Marketing,” February 19, 2025.
- Financial Times – “Brands Target AI Chatbots as Users Switch from Google Search,” April 27, 2025.
- uTires (study) – “Study Reveals Where Drivers Are Most Reliant on Their GPS,” 2021.
- We Are Social – “Digital 2024 Report,” January 2024.
- Statista – “Google’s Ad Revenue Dwarfs Competitors,” 2023.
- Wall Street Journal – “Global Ad Revenue Will Hit $1 Trillion Faster Than Expected, GroupM Forecast Says,” June 2024.
