The reason you don't want to use a computer that incorporates workload is actually quite simple. When you use a computer with a published, i.e. disclosed model, you can plan and execute dives more logically. Granted I'm not going to advocate live human experimentation, but in effect we're all doing it anyway.
Here are my observations having owned multiple AI computers in the past, and owning exactly none now. A few things, if you are genuinely planning your dives, the concept of air integration is unnecessary, but if you like it - go for it. Before I jump in the water, I how much gas I have, the gas fractions of all gas, and I even know to the closest 100psi how much gas I'm going to have every ten minutes in the dive, and upon my return to the surface. I actually, ahem, plan my dives which includes more than just pressing plan mode on a computer and seeing how much bottom time it lets me have within the no decompression limit.
The reason I don't like whiz-bang adaptations of the model is that I don't know where I started on the model and I certainly don't know where I ended up. Say we plan a dive using ZHL16-B 30/85 and we setup our Shearwater to ZHL16-C 30/85. We can expect our computer will produce a result during the dive that is shockingly scary close upon execution. When we surface we feel tired, groggy, but perhaps not symptomatic, but we’re tired and take a nap on the couch for hours eat some soup and go to bed. I think I need more shallow time. Next dive we try 30/70 and the difference, how we feel is night and day different! In fact, after a 170ft dive we go home and cut the grass, take the kids out for ice cream, and fire up the grill and make some tasty treats. What we've done is admittedly live human experimentation, but we discovered in a one off trial that 30/85 is slightly too aggressive for that particular dive plan.
With whiz-bang gizmo computers, what we find are all sorts of model interpretations by programmers under the direction of product liability lawyers. You can search for a fascinating conversation I had with RonR here on ScubaBoard about the Atomic Cobalt. The Cobalt runs "Proprietary RGBM" (whatever that means) and I observed it did weird and highly unpredictable things beyond 150ft. Turns out, the manufacturer indeed included a "Folded Model" (whatever that means) and it does modify the model (how we'll never know) in real time at greater depths. Okay. No Problem. How do we plan a technical dive with your "multi-gas" computer accurately? Anyone who's ever spent time around a Suunto computer knows about the "Suunto Minute", also a recreational interpretation of "Proprietary RGBM".
Now I know by now people are going to jump on me for picking on bubble model, but don't waste your time. I'm not picking on bubble model, what I'm doing is pointing out that when the manufacturers or their lawyers manipulate the model we can't understand where we started or what we need to adapt our processes to best utilize it.
It is my belief these features exist to convolute the dive planning process and limit legal recovery. I can imagine that a diver using one of these devices with a half dozen settings has a reduced opportunity to sue/recover damages upon being injured. I can visualize the manufacturer's attorney saying, "You failed to set the computer to morbidly obese as shown on page 134 paragraph 2, therefore you didn't do it correctly, sue your instructor who failed to teach you how to use this properly and oh btw – learn to read the manual."
