[henearkr]

I understand the push for stratospheric aerosols, but the alleged need for deeplearning is poorly justified.

All the points listed in favor of deeplearning in this article only point towards the need for heavy calculations, but I would say it's not deeplearning that's required, but instead very precise physical simulations.

For example to help the deployment of drones to spread the aerosols, I don't think that you can model winds better with deeplearning than with current models used by meteorological agencies. Or, at least, it does not seem to me the bottleneck here. Or, OK maybe there will be breakthroughs that would replace physical simulation-based meteorology by deeplearning models, but I don't think it is helpful right now, or you would need to do a lot more explaining to convince me.

[acc_297]

yeah that's the weirdest part of it aerosol injection is heavily debated and reasonable people can disagree about it but standard engineering principles can be applied to build new planes and standard global circulation models can be applied to model long range particle transport in the atmosphere -- these are very understood problems by human beings and have been for years -- sure AI can assist with tweaks / optimizations i.e. with airfoils etc. the emphasis that these are problems where AI / ML are ripe to be applied is bizarre and speak to a greater bias in favour of AI / ML development and adoption than a bias in favour of geo-eng solutions to climate change