AI model evolution

Many people have doubt on AI as: 1) Open AI delays launching GPT 5, 2) the Reinforced Learning RL for Strawberry, the new model launched by OpenAI, doesn't necessarily need more computing power for training.

However we highlight Strawberry utilizes more computing power for inferencing given  longer multi-step reasoning, and it can only specializes on one type of input/output more precisely , Eg, text, or image, different from GPT which is a general model for multi-modal variables.

As such, we will see computing power shifting more to inferencing/ASIC from training/GPU.

That's it for this week!