Interesting article in the Economist this week. Hopefully, this resonates with all of you.
https://www.economist.com/business/2025/07/14/ai-is-killing-the-web-can-anything-save-it
Key Takeaways & Risks
- The web’s old economic model is under existential stress. If creators can’t earn from producing content, many will drop out or drastically scale back.
- Big platforms and AI firms hold the bargaining power. Without regulatory or technical constraints, the balance of value is shifting heavily in their favor.
- The web’s diversity, serendipity, and long-tail content risk being squeezed out — AI tends toward “what’s common, high‑signal, easy to summarize.”
- There’s a “tragedy of the commons” aspect: if every website blocks AI bots to protect itself, the web’s utility to the AI ecosystem shrinks, hurting the entire stack.
Why It Matters
- This isn’t “the web is dead” — more like the web is being rebuilt behind our backs. We’re in a transition phase.
- The hardest part will be for independent creators and small publishers. They lack scale, leverage, and resources to negotiate or build alternative monetization paths.
- There’s a window now to design the future web economy. If standards, protocols, and legal frameworks don’t evolve soon, we’ll get stuck with a de facto AI‑owned web.
- One possible bright spot: niche, high‑trust content (deep research, primary sources, original data) becomes more valuable. AI may always need “ground truth” inputs it can’t reliably generate from its own back catalog.
- That said — tech disrupts. Things will feel painful for years. Some sites will die, some will adapt, some new surprises will emerge.
What This Means for Healthcare IT, Providers, and Digital Strategy
1. Health Content Traffic Is Already Taking a Hit
“Traffic to health websites has fallen by over 30% year over year.” AI models like ChatGPT, Perplexity, and Google’s Gemini are now delivering direct answers to health-related queries that used to send patients to Mayo Clinic, Healthline, or even your own health system’s symptom-checker pages.
Implication:
- Patient education content you’ve spent time and dollars creating (FAQ pages, triage tools, wellness blogs) may never see a human eyeball if it’s summarized or paraphrased by LLMs.
- Digital front door efforts could be undermined—particularly SEO-driven patient acquisition strategies.
2. Epic/Oracle/Meditech Portals May Also Be Threatened
If AI search agents begin interfacing directly with patient records, appointment booking, or provider messaging (e.g., via ambient voice assistants or smart agents in the EHR), the patient portal as we know it becomes background infrastructure.
Implication:
- Health systems must plan for conversational and agentic interfaces as core front-end experiences, not add-ons.
- Integration with AI agents will be expected by patients who no longer want to navigate dropdown menus.
3. Clinical Publishers, Toolmakers, and Reference Apps Will Be Disrupted
Imagine UptoDate, VisualDx, Lexicomp, or even your internal clinical decision support tools getting “strip-mined” by AI models that learn from them but never send users back to them.
Implication:
- Knowledge vendors may shrink or gate content, limiting training access and increasing costs for licensing.
- Health systems may need to verify AI outputs against trusted, whitelisted sources, which creates new governance and liability complexity.
- Opportunity to build or co-develop internal LLMs trained on your organization’s protocols, pathways, and formulary—keeping the value in-house.
4. Advertising & Awareness Budgets in Jeopardy
Many health systems rely on digital ad traffic to promote service lines (e.g. ortho, oncology, maternal care). If Google AI Overviews or Bing Copilot respond with generic answers (“See your doctor”), your outreach ROI collapses.
Implication:
- Reinvest in community, niche, and owned media channels (email, SMS, local event listings).
- Consider AI-optimized content strategies that feed the LLMs rather than fight them (i.e. structured, schema-annotated content the AI models “see”).
5. Legal & Ethical Headwinds Are Coming Fast
If AI bots are crawling internal medical content or practice standards that weren’t licensed—do you detect it? Do you stop it? Or charge a crawl fee? That may soon become the new “privacy policy.”
Implication:
- CIOs and CISOs should prepare policies on AI crawler access, bot detection, and digital rights.
- Potential for consortia across health systems to demand fair AI compensation, similar to what media outlets are attempting.
Why it Matters
This isn’t just a tech disruption—it’s an economic and existential shift in how health information flows, how patients engage, and how systems differentiate.
The AI layer isn’t just summarizing the web. It’s replacing it. And if your hospital, practice, or vendor isn’t part of the new training/response loop, you’re out of sight and out of relevance.
| Risk | Response |
| AI siphoning health info traffic | Focus on trustworthy, differentiated content and explore structured syndication into AI ecosystems. |
| Portal irrelevance | Invest in voice/chat interfaces and APIs for AI assistant integration. |
| Clinical content devaluation | Partner with vendors on co-branded, protected datasets or build private AI assistants trained on your protocols. |
| Outreach ROI decline | Diversify into owned media, email/SMS, and local community awareness tactics. |
| Legal/ethical gray zones | Craft bot policy, access control, and IP strategies now—before the crawlers come knocking. |
