So, several things to unpack here, I'll address in order of your response.
1. AIOs are not just gimmick, they do work, but there is a perceived assumption by people that they are some form of holy grail of cooling at a significant cost savings. Consider the saying 'sounds too good to be true'.... AIOs aren't magic and the claims the average user makes based on their purchase is likely shrouded a bit in their pride and knowing they spent $100-$200 on one. I've tested literally dozens of AIOs for Tom's Hardware and while they're decent and mostly positive reviews, there isn't anything really that special about them....nearly ALL Of them are based on the same 2 or 3 designs....nearly identical. But to also close this out, custom watercooling can cost several hundreds of dollars (or more) and may not offer a lot of improvement over a large AIO or large air cooler, or, they can offer great impact - it is all subjective to build, airflow, coolant flow and loads being dissipated. For example, my 9700k and GTX 2080 run at 45-50C at load in the same cooling loop. Depending on ambient room temp, this fluctuates up or down a bit, which all cooling does- air or liquid. (You cannot have sub-ambient liquid or air....this requires other cooling methods, regardless what your temp readings seem to be telling you.)
2a. Heat output is a direct relation of a components TDP (thermal design package/thermal design power/thermal design profile, depending on who you ask). This is the rated wattage used by the component at maximum utilization based on a given configuration, usually factory or base clock speeds. Overclocking requires more voltage to stabilize, meaning that TDP rises as a result. Setting your overclocking BIOS to AUTO often overcompensates and provides more voltage than needed, resulting in higher temperatures, but at the benefit of being 'easier'. Manually defining these settings can lead to lower temps due to lower voltages needed for a specific clock speed and stability.
2b. Killing your CPU isn't that likely to happen. Motherboards and CPUs have failsafes in place to prevent damage. They'll reset or refuse to operate if conditions aren't good enough to do so. In all honesty, the biggest threat to hardware comes in terms of shorting something out (screwdriver or other metal on motherboard or components while running) or by static or other electrical discharge. Liquid cooling leaks can be an issue and cause shorting, which AIOs are usually not known for as they are factory sealed, but leaks can happen...just rare. Custom watercooling can leak and usually is due to installer or operator error, usually where tubing or fittings connect. Just remember, when you Google search for 'Corsair cooling leaks' you are going to find them because that's exactly what you searched for, not an indicator of how common they are. The actual percentage of failures is remarkably low given the volumes of millions and millions of AIOs around the world in-use each day.
3. An AIO or custom watercooling may not actually be what you need at all - this is where understanding what YOU want to do for your build comes into play. Your CPU is a 4790k, based on your info above from a previous response. This is a CPU which was originally sold 2014-2015 or so with a TDP of 88w according to the Intel ARK site.
https://ark.intel.com/content/www/u...-4790k-processor-8m-cache-up-to-4-40-ghz.html If you do not plan to overclock at all and let it run with standard factory configuration, something like a 4-6 heatpipe cooler will likely suffice for you, provided you have good case airflow. I've tested the Cooler Master Hyper 212 on the Tom's Hardware 10-core, 20-thread i9-10850k at 4.6ghz and while it wasn't happy, it did OK on this monster CPU while allowing 100% fans to be utilized. We're easily talking about 225-250w of CPU in that configuration.
https://www.tomshardware.com/reviews/cooler-master-hyper-212-evo-v2
4. AIOs are not silent. Well, they can be if you run an aggressive PWM curve on the fans to spin down at lower temps reported by CPU die readings. If you want an AIO to perform as well as it can, you either need a larger AIO (280 or 360mm) or have an AIO big enough for your CPU TDP to mitigate thermal loads with lower fan speeds. Otherwise, you have roaring fans for those temps...the price to pay.
5. Making a decision. It sounds like you are at a crossroads, but I'll give you advice if I were in the same shoes. Assuming the AIO you currently have is working: pump moving coolant, fans spinning, temps are decent....don't change any hardware. If anything, learn how to setup your AIO to be more effective at high CPU utilization and then have the fans slow down when temps and load are at idle. I personally prefer to set any liquid cooling pump to run at 100% at all times and allow radiator fans to spin up or down as needed. Liquid cooling pumps are usually very quiet in operation, so if yours makes a lot of noise and very audible up to a meter away, something else could be wrong. Otherwise, AIO noise typically comes from radiator fans. If you can setup your fan curves to adequately go up and down as needed for cooling, this can help with noise and temps as needed.
The reason I say leave the pump at 100% and only alter fan speeds via PWM curve is this:
Thermal conductivity of cooling component metals (copper or aluminum blocks/radiators) and coolants themselves as well as specific heat of water and water-based coolants allow for a large volume of thermal energy to be stored in coolant before temps raise by a single degree Celsius. (from my sticky)
https://forums.tomshardware.com/thr...-watercooling-sticky-v2-0.331454/post-3502926
This means that CPU temp spikes which fluctuate up and down over the course of seconds, even to minutes would try to trigger the AIO pump to ramp RPM up and down, even though the amount of wattage being output would not immediately change the temperature of the coolant itself unless it was sustained. CPU temps (reported by die temp or software) DOES NOT indicate coolant temperature, only reported core temp. Coolant temp is not quickly spiking up and down...coolant takes time to accumulate thermal loads as a result.
Radiator fans are the single, best way to dissipate additional thermal loads, provided the flow rate is consistent. Even with pump speeds ramping up and down, cooling still depends on fans and radiator dissipation as a limiting factor of overall cooling capacity. Liquid cooling systems (AIOs or custom watercooling) still abide by the same sets of rules which is often considered a product of multiple different things: coolant flow rate (based on pump speed and head pressure), radiator surface area, radiator airflow restriction, fan airflow and the thermal properties of both coolant and the metal components (blocks, radiators). Lower any one of these and you MUST increase another in order to achieve the same level of performance. Ex: lowering fan speed to make a cooling system quieter likely means keeping the pump and flow rate consistent but adding an extra radiator or extra set of fans for push+pull airflow.