All AI agents essentially do three things. First is pre-training, which is like us hitting the books and attending classes. Second is post-training, like chatting with seasoned pros to gather experience.
Third is reinforcement training, akin to us in the real world, grinding away and getting real feedback.
Open AI has defined five stages of standards regarding AGI.
The first phase is the chat bot—like an entry-level trader.
The second phase is the Reasoner—like a trader analyzing the market.
The third phase is agents—these are the smart players in the market.
The fourth phase is the Innovator—think of it as the new player in the game.