One useful way to think about AGI is as a progression of operating modes rather than a single threshold. Each stage changes the role AI plays in human work, decision-making, and autonomy.
Level 1: No AI
At this stage, work is handled entirely by humans or by traditional software systems without AI reasoning or assistance.
Level 2: AI as a Tool
AI helps automate narrow tasks under direct human control. This includes systems such as search, ranking, and grammar support.
Level 3: AI as a Consultant
AI contributes more substantive insight, summarization, and reasoning, but still depends on human initiation and oversight.
Level 4: AI as a Collaborator
AI starts behaving more like a working partner. It can participate in multi-step tasks, support decisions, and interact in more persistent workflows.
Level 5: AI as an Expert
AI begins to lead within specific domains, especially where deep pattern recognition or large-scale reasoning creates an advantage over manual work.
Level 6: AI as an Agent
At this level, AI acts autonomously across tasks, workflows, or digital systems. This is where identity, permissions, security, verification, and alignment become especially important.
Why ABV cares about this framework
As AI systems become more agentic, the risks also grow. That is why ABV places so much emphasis on trusted infrastructure, secure agent execution, and verification layers such as A2SPA.