People are right to doubt the claims of notOpenAI and others about the capabilities of their models. The nonlinear output gains do not mean that the quest for i

People are right to doubt the claims of notOpenAI and others about the capabilities of their models. The nonlinear output gains do not mean that the quest for intelligence is over. It's already hard to steer them with RL to make proper math. It's more likely that the transformer will only be a part of the larger architecture.