Most of this is noise; the idea of a fundamental opposition between gating networks and basic methods for venture capital prediction is an oversimplification.
A gating network, which combines textual rationales and company attributes, does not cancel out existing approaches but can complement them.
Focus on what is controllable and relevant for action, not on conceptual struggles.
For example, in South Korea, a gating network could refine predictions of LLM models by integrating local market signals or structured financial data, rather than replacing them.
The question is how to use them effectively, not whether one opposes the other.