Altman predicts AI research interns by 2026, full researchers by 2028

OpenAI CEO Sam Altman has outlined an ambitious roadmap for artificial intelligence development, predicting that AI models will function as “research interns” by 2026 and as fully independent researchers by March 2028. This timeline represents one of the most specific AGI development schedules from a major AI company, potentially signaling a shift from AI as a responsive tool to AI as an autonomous collaborator in research and development.

What you should know: Altman’s vision involves AI evolving from today’s chatbot capabilities to systems that can independently conduct research and make decisions.

• By 2026, OpenAI expects to deploy AI “research interns” that can read papers, compare research, highlight novel findings, and suggest next steps without requiring detailed human instructions.
• The 2028 target involves “true automated AI researchers” that can design experiments, test hypotheses, and present results independently—functioning more like colleagues than tools.
• Altman acknowledged that the term AGI has become “hugely overloaded” and prefers defining specific capabilities rather than satisfying broad AGI definitions.

The technical challenge: Moving from current chatbot capabilities to reliable research agents faces significant hurdles that could derail Altman’s timeline.

• AI hallucinations (when AI confidently provides incorrect information) remain a critical barrier, as research tools must be completely accurate to avoid requiring constant human verification.
• Current AI systems can perform research-like tasks only when given specific structures to follow, lacking the autonomous decision-making Altman envisions.
• The leap from responsive AI to proactive research agents represents what experts consider a steep technological climb.

Why this matters: Altman’s predictions suggest AI development is moving toward systems that work alongside humans rather than simply responding to them.

• An AI research intern could revolutionize knowledge work by autonomously analyzing information, identifying patterns, and suggesting improvements without human guidance.
• Independent AI researchers could fundamentally reshape the workforce, potentially replacing human researchers in certain domains.
• The timeline positions OpenAI as targeting practical AGI capabilities rather than pursuing abstract artificial general intelligence definitions.

The broader implications: This roadmap raises questions about safety, control, and workforce disruption that extend beyond technical capabilities.

• Safety concerns emerge around who controls autonomous research systems and how they make decisions.
• The quality of AI outputs matters more than meeting calendar deadlines, as confidently wrong AI systems could prove “worse than useless.”
• Success would require AI that can show its reasoning, cite sources, and acknowledge limitations—capabilities that remain challenging for current systems.