OpenAI Wants to Replace Human Researchers by 2028 — E-E-A-T Is Your Shield

On March 20, 2026, OpenAI's chief scientist Jakub Pachocki revealed in an exclusive MIT Technology Review interview that the company has made the "autonomous AI researcher" its new "North Star." The goal: a fully automated multi-agent system capable of solving complex research problems — in math, physics, biology, and even business strategy. Deadline: 2028.

An intermediate milestone is planned for September 2026: an "autonomous AI research intern" capable of tackling a small number of specific research problems without human oversight. This isn't just another language model — it's a coordinated agent system combining reasoning, web browsing, and interpretability.

Why this is a signal for SEO

At first glance, an autonomous AI researcher has nothing to do with search rankings. But look closer. What Pachocki describes — a system that browses the web, reads sources, evaluates their credibility, synthesizes and produces original content — is exactly what emerging AI agents already do. Faster and more autonomous.

The direct connection to your content:

• AI agents will become massive web content consumers. A 2028 AI researcher won't read 10 articles — it'll read 10,000 in seconds. The question is no longer about ranking on page one. It's about being the source the agent decides to cite.
• Verifiable human expertise becomes the only competitive advantage. If an AI agent can produce a 2,000-word article on any topic, your generic content is worthless. What remains irreplaceable: proprietary data, unique methodology, documented field experience.
• E-E-A-T shifts from a ranking factor to a survival filter. Google already uses Experience, Expertise, Authoritativeness, and Trustworthiness to rank. In a world where AI produces content at scale, these signals become the primary distinction criteria.

OpenAI's plan in detail

Pachocki outlined a two-phase roadmap:

1. September 2026 — "AI Research Intern." A system capable of handling a limited number of specific research problems autonomously. It combines reasoning models (o-series type), web browsing agents, and interpretability tools.
2. 2028 — "AI Researcher." A fully automated multi-agent system able to tackle problems "too large or too complex for humans." Applications: mathematical proofs, conjectures, life science discoveries, business and policy analysis.

Pachocki states: "We are getting close to a point where we'll have models capable of working indefinitely in a coherent way just like people do." The ambition isn't to replace human researchers, but to create "a whole research lab in a data center."

What to do now

The good news: you have 6 to 18 months before these systems are operational at scale. But the window is closing. Here are 3 priority actions:

1. Invest in proof of expertise. Named author with verified LinkedIn profile, documented methodology, cited source data. This is what Google's algorithms already verify — and what AI agents will verify too.
2. Produce content AI can't generate. Proprietary case studies, exclusive data, field experience with verifiable numbers. If your content can be reproduced in 30 seconds by an LLM, it has no distinctive value.
3. Structure for AI citation. Schema.org (Person, Organization, Article), structured data, concise answers at the start of paragraphs. AI agents cite content they can easily extract.

Cicero's take: OpenAI isn't just building a tool — it's a signal that tomorrow's web will be read as much by agents as by humans. Your content must convince both audiences. E-E-A-T is no longer a checkbox — it's your license to exist in the results.

Alexis Dollé

CEO & FOUNDER

Growth and SEO content strategy specialist, I founded Cicéro to help businesses build lasting organic visibility — on Google and in AI-generated answers. Every piece of content we produce is designed to convert, not just to exist.

LinkedIn