Google DeepMind researchers and others proposed creating a kind of “ID card” for AI that displays its autonomy, influence, goal complexity, and generality levels. Researchers from Carnegie Mellon University and Google DeepMind proposed an “agentic profile” in the journal “Nature” on August 13, evaluating AI across four axes: autonomy, influence, goal complexity, and generality. The first criterion is autonomy (A), which measures how independently an AI can act without step-by-step human instructions. The second is influence (E), indicating how directly an AI’s actions can alter the real environment. The third, goal complexity (GC), evaluates how well an AI can break down complex tasks into steps and organize them. The fourth, generality (G), reflects whether the AI is specialized for one task or can handle multiple domains. The researchers compared AlphaGo, ChatGPT-3.5, Claude 3.5 Sonnet, and Waymo’s autonomous vehicle using these four criteria, assigning each a 0–5 scale. AlphaGo received an autonomy rating of A3, influence E1, and generality G1. ChatGPT-3.5 scored G3 for generality, but its autonomy was A2. Claude 3.5 Sonnet scored A3 for autonomy, with influence (E3) and goal complexity (GC4) higher than ChatGPT-3.5’s. Waymo scored low in generality (G2), but received the highest autonomy (A4) and influence (