Ask HN: What does OpenAI's O3 Arc result mean for the field?

From criticism of how inadequate the transformer architecture is, depletion of training data, improvements to the attention mechanism, AGI supposedly being miles away; where does the new results put the field and what avenues need more focus and what others need less focus to the point of complete cutoff?

[flagged]