One of the big unknowns in the model is "where will the clouds show up?" That unknown stems from our understanding of the water capacity of air by temperature, the increase in air temperature leads to the air holding more water, and water is the basis for cloud formation. If the clouds form "low" they increase albedo and create colder temperatures, if they form "high" they act as a semi-mirrored surface and reflect light that has been reflected from the surface back down for another shot at generating heat.
Much of the IPCC's work has been done in MATLAB[1,2] so if you have a reasonably powerful workstation you can play around with various initial conditions and settings yourself to see what might happen in the future.
No matter what the far future holds, the near future holds more violent storms as storms are powered by the temperature differentials of the air, land, and sea.
It is of note (for me, probably not for many others) that we don't have good models for how an ice age starts. There are a few papers that talk about ice ages being a response to warming (hit a tipping point, generate clouds, and get a "nuclear winter" scenario without the nuclear part). But much of the nuclear winter work has been refined and that scenario is generally considered unlikely AFAICT from what people seem to be publishing these days. Turco's work[3] and things that cite it are a good jumping off point if you want to read up on that. It isn't perfect because smoke/soot are not clouds (different albedo numbers, different cooling attributes) but the accumulation and dispersion of atmospheric obstructions is solid stuff.
[1] Some code and information used to generate plots in the IPCC reports -- https://github.com/IPCC-WG1/Chapter-9
[2] Mathworks trying to get you to buy their climate data toolbox -- https://www.mathworks.com/discovery/climate-stress-testing.h...
[3] Climate and Smoke: an Appraisal of Nuclear Winter -- https://www.science.org/doi/abs/10.1126/science.11538069