Open models provide much better proof and explain exactly this. They enable policymakers, who are key to grid planning, to make verifiable arguments rather than 'black box' ones. Such verifiable models are much harder to challenge.
The EU is demanding that the European Network of Transmission System Operators for Electricity (ENTSO-E) implement this ten-year power infrastructure plan transparently. Unfortunately, this modelling has so far been carried out using an expensive, proprietary tool (PLEXOS Energy Exemplar). The results were not easily reproducible. However, with the new model based on PyPSA-Eur, a transparent Ten-Year Network Development Plan can be created.
It is actually an model that has been developed in collaboration with entso-e by Open Energy Transition. You can read more about it here: https://open-tyndp.openenergytransition.org/
What most AI companies won't tell you is that creating a massive amount of new code will result in an overwhelming amount of technical debt for companies worldwide, sooner or later. Many organisations now believe that they can develop their own apps form scratch using AI, but if they don't pay for tokens capable of handling their ever-growing infrastructure code, their infrastructure will rot. The tendency to reinvent the wheel has always been a problem in the software industry, and AI will only accelerate this trend.
Nevertheless, I'm optimistic that an LLM will be developed sooner or later that is fully aware of the open-source ecosystem and can create software in the correct way: This would involve using pre-existing code and reviewed modules that are plugged together and optimised to create as little new code as possible while reusing as much open-source code as possible.