Artificial intelligence is taking center stage on websites. Its promise: faster navigation, personalized content, enhanced accessibility. But behind these advancements lie crucial questions: what does the future hold for the websites themselves, for the diversity of information, and for our digital footprint? How can we position ourselves as website publishers and as a Drupal web agency?
It’s 11 p.m. An anxious parent logs onto the local hospital’s website: their child has a fever and they’re looking for a quick, reliable answer. Rather than dig through complex menus, they simply type: “My son has a fever of 39°C, should I see a doctor tonight?” Within seconds, an assistant built into the site responds: basic medical advice, emergency numbers, directions to the nearest on-call clinic.
This scenario isn’t the norm yet… but it illustrates what tomorrow’s user experience could be, where every site is interactive, responsive, and human-centered. With tools like Drupal AI—connecting the open source platform to top AI services (OpenAI, DeepL, or other specialized modules)—organizations now have an unprecedented toolkit to transform the online experience.
The benefits are clear: reducing repetitive tasks, improving accessibility, personalizing the user journey. In short, AI allows websites to become full-fledged conversation partners, no longer just static showcases.
But as AI becomes the gateway to the web, the fragile balance of the digital ecosystem is being disrupted. What will remain of the websites themselves if users get all their answers directly from a conversational interface? What will be the impact of this reliance on algorithms for our economic models, plurality of voices, reliability of content… and the ecological footprint of our digital behavior?
Promises: saving time, accessibility, and personalization
The advantages of AI applied to the web are undeniable.
Where a writer would spend hours crafting meta descriptions for SEO, a generative model can now do it in just a few minutes. Where a communications department once painstakingly translated its content, tools like DeepL, connected to Drupal, automate this work with unmatched quality.
Beyond saving time, AI also makes the web more inclusive.
A visually impaired user browsing an e-commerce site can, thanks to integrated AI, benefit from automatic image descriptions.
Since the Drupal AI initiative, the Drupal CMS has positioned itself at the forefront of AI. The ecosystem already offers modules that we have integrated into our Drupal projects, enabling:
- the generation of SEO-optimized content,
- automatic page translation,
- custom conversational assistants,
- easier comment moderation.
These tools pave the way for more dynamic websites, able to interact with their users and tailor their responses to the context.
AI, the new gateway to the web: what impact on the business models of our Drupal sites?
In the 2000s, Google established itself as the gateway to the Internet. Tomorrow, the shift might happen again. But this time, instead of browsing from link to link, users will ask their questions directly to an AI.
This is already happening with ChatGPT, Perplexity, or Microsoft Copilot. Ask a question, get a concise answer, without ever visiting the source sites: an impressive efficiency, but a risk for the web economy.
Because if AI becomes our one-stop shop, who will still visit the original sites? Business models based on advertising, subscriptions, lead generation, or sign-ups risk collapsing. Content publishers might lose their audience, and therefore their reason to produce.
This shift doesn't just concern media outlets or bloggers. It directly affects our clients: B2B companies, major corporations, local governments, universities, or public institutions. For years, they have invested in strong SEO strategies: producing quality content, optimized search visibility, sites designed to capture new prospects or users. But if Internet users now get their answers directly from ChatGPT or Perplexity, this entire strategy will have to adapt!
Efforts made to generate traffic will benefit organizations less…and feed the AIs instead, sparing users from consulting the source sites. In other words: your content, designed to attract and persuade, becomes the building blocks that feed artificial intelligences, with no direct return for your organization. While major news publishers like the New York Times or Le Monde have reached financial agreements with AI companies, smaller outlets see their content scraped and their traffic plummet. According to Similarweb, traffic to American news sites from search engines has reportedly dropped by 26 %.
It should also be noted that beyond simply AI using free content, the arrival of conversational agents has considerably accelerated the rate of online text production. As Sundar Pichai, CEO of Google, pointed out in a podcast for The Verge, “the number of web pages increased by 45% over the past two years.” We are witnessing a vicious cycle: a web increasingly produced and consumed by AIs rather than by humans. Conversational agents generate content, publish it online, which is then absorbed by those same agents to answer users’ queries… who then no longer need, and sometimes no longer want, to visit the source sites directly.
A democratic risk: when the diversity of voices fades away
Even for the end user, this filter is not without risk: an AI “pre-chewed” web, where sources are no longer consulted, is also a web where the diversity of viewpoints fades away.
Why does AI tend to erase this plurality? Because a generative model does not reason like a journalist: it does not rank arguments by their relevance nor sources by their reliability. It operates through statistical average, producing the most probable sequence of words. Minority opinions, original analyses, or dissenting voices are thus diluted in a standardized discourse.
Yet, the plurality of media is an essential condition of democracy. When a reader visits a news site, they don’t just access raw information: they benefit from editorial transparency, an acknowledged context (editorial line, values, angles of coverage), and, above all, different points of view. Confronting these perspectives makes it possible to weigh pros and cons, refine one’s judgment and, ultimately, become a better informed and more enlightened citizen.
If AIs become our only intermediaries, we risk shifting towards a web where nuance vanishes in favor of a “single truth” generated by algorithm. A truth without plurality, and thus without true democratic debate.
The environmental cost: an invisible but massive expense
A less visible but equally crucial question concerns the environmental impact of artificial intelligence.
Today, it’s difficult to obtain a reliable measurement: major digital players remain opaque about the energy consumption and resources required for their services, as well as the locations of their data centers. And even if this information were available, the real footprint of AIs would still be underestimated, since indirect effects, known as “rebound effects,” are still impossible to model.
However, studies are beginning to shed light on the issue. An analysis conducted by Mistral AI, Carbon4 and ADEME highlights two major findings:
-
Data center location plays a decisive role. More than 85% of a model’s emissions come from electricity consumption related to its training and use, and therefore from the energy mix of the country where it’s hosted. In France, the heavy reliance on nuclear limits CO₂ emissions—an advantage that doesn’t exist in most other regions of the world.
-
Carbon footprint increases with model size. The bigger the model, the higher its environmental cost soars. That’s why it’s important to tailor AI to real use: there’s no reason to use a giant model just to generate a simple email or calculate an Excel formula.
Given its environmental impact, it's therefore necessary to question the use of artificial intelligence. Without discernment, we risk a considerable increase in our digital footprint. As Julia Meyer from ADEME reminds us, “very few digital tools today have more benefits than impacts”:
-
Positive example: some AIs already help optimize water or energy consumption in industry and cities, or track land artificialization (IPCC, ADEME).
-
Negative example: others, on the contrary, aggravate the problems. Shein’s fast fashion illustrates this danger: thanks to AI, the company produces faster, sells on a larger scale, and causes its carbon footprint to skyrocket.
Thus, a question arises: is it relevant for our institutions to invest millions in ever more powerful AIs without questioning their purposes? AI must be put at the service of the ecological transition, not its aggravation.
Is a “frugal” AI possible? Yes, experts say, provided we redefine needs and rely on existing frameworks (REEN law, RGESN, AFNOR initiatives, European AI Act).
Towards a reasoned AI: a strategic choice, not a trend
In light of these challenges, one conviction stands out: AI must be a strategic and measured choice. For us as web agencies and for our clients who manage Drupal sites, every intended use should come with some simple questions:
- Does this use of AI address a real need?
- What added value does it bring (time saved, improved user experience, increased accessibility)?
- Is its ecological and societal cost justified?
This thoughtful approach aligns with an emerging trend: that of sustainable AI. It’s not about rejecting innovation, but about adopting it selectively and responsibly.
And in this context, Drupal offers a particularly well-suited framework. As an open source CMS, it allows for transparency regarding the modules used, modularity in AI integration choices, and the possibility to prioritize local or sovereign solutions.
Conclusion: for a reasoned use of AI
Does this speech seem alarmist to you? That is not our intention.
The purpose of this article is to highlight the endless complexity of the new questions raised by artificial intelligence and its potential impacts on our economic, social, democratic, and environmental models.
We are not anti-AI. In fact, as we have seen, artificial intelligence combined with Drupal offers many advantages! We have already integrated AI into our Drupal Starter! But we are convinced that it is essential, as digital professionals, to discuss these questions, to sound the alarm when necessary, and to examine our relationship with AI.
The primary aim of human innovation is ultimately to increase well-being. Their use therefore cannot come at the expense of our economy, our democracies, or our planet.
The web was born out of a promise: that of an open, thriving space where everyone can publish, explore, and compare different points of view. AI can be an incredible lever to enrich this experience, as long as its use is controlled and regulated.
Tomorrow, websites must not vanish behind uniform conversational interfaces. On the contrary, they must become spaces where AI amplifies value: making information more accessible, more relevant, more human.
The question isn’t: “Should we use AI?” But rather: “Where and how should we use it so that it’s truly worthwhile?” Because behind every use case lies a collective choice: to preserve a web that is vibrant, diverse, and sustainable.
Do you have a Drupal project? Are you wondering about potential AI integration? Now is the time to think it through together. Contact us!
Sources:
- How AI-generated answers threaten the foundations of the Web,
Le Monde, Morgane Tual - Artificial Intelligence: A partnership agreement between “Le Monde” and OpenAI,
Le Monde, Louis Dreyfus - Our contribution to the creation of a global environmental standard for AI,
Mistral.AI - Perspectives on the impact of generative AI,
Ademe