Doesn't how easy it is to monitor gas usage manually depend on the predictability of one's dives? Sure, if you dive a square-ish profile, it's easy to use your known SAC rate to estimate how much your SPG needle should drop in 5 minutes at your current depth. For cave and wreck diving, etc., it works fine. In that kind of diving, the diver also takes an abundant supply of gas, carefully planning the volume needed for the dive and leaving plenty of reserve for contingencies. If he brings enough gas and sticks to the plan, what his SPG reads should not affect whether he can get out safely. He just checks it now and then to confirm what he predicts it will read at any given point in time. Time, not gas, is likely to be the more significant parameter to monitor. He knows at about what time he will have to turn the dive.
In contrast, I can understand why AI is attractive to people whose typical dive plans involve meandering up and down over something like a reef, stretching their gas supply as far as they believe it can safely be stretched, leaving only a relatively small reserve--the proverbial "be back on the boat with 500 psi" kind of dive. AI's calculation of gas-time remaining helps this kind of diver maximize gas time the same way a computer's calculation of NDLs helps this kind of diver maximize bottom time. If I understood correctly, I believe Stuart pointed out the inconsistency in an argument that it's okay to rely on a computer's calculation of NDL but not okay to rely on a computer's calculation of GTR. I would think those who are okay with stretching no-deco time would be just as okay with stretching their gas supply. I still use an SPG because I enjoy the simplicity of it and was trained to do it that way. Also, I like to leave large safety buffers and am often back on the boat with a large amount of gas left, not to mention no-deco time left. However, if I were less risk-averse and willing to leave smaller safety buffers, I could envision how riding my computer's GTR display might allow me to extend my bottom time the same way riding my computer's NDL display might allow me to extend my bottom time.
I can't say I appreciate the advantage of being able to see my gas pressure on my wrist rather than a separate SPG, but others have said that it can be helpful in certain situations, such as when you're hanging on for dear life in a current. I don't foresee myself diving in conditions that are so difficult that I can't easily check my SPG.
As far as the data logging capability of AI, I guess some people find analyzing their data to be a fun exercise, but I sure don't. I calculate my SAC infrequently, and if I am planning something that I know is likely to stress me, I simply figure I'll need more gas than my SAC would indicate, maybe using a figure of 1 cfm if really stressed. I see no need for, and would derive no fun from, minute dissection of logged data. But some of you may. I think I referred to you (in good fun) as data nerds earlier in this thread or a similar one. If logging and analyzing dive data floats your boat, have at it. Some people just enjoy the idea of control panels with lots of data and measurements and that sort of thing. A surprising number of divers, especially tec divers, seem to be engineers and scientists. I'm a former engineer, and I don't enjoy numbers as much as I used to. I'm looking forward to the day when I can measure time in units of a season. Until then, I'll reluctantly deal with minutes and PSI to the minimum extent necessary to safely dive.
As others have repeatedly said, we're all different. For some kinds of diving, AI may provide benefits, and for other kinds it may just be an expensive toy. Again, whatever floats your boat.
Let's face it, AI is a gimmick. A usefull gimmick, but not something you need or even something you should need. If AI solves a problem for you you better solve the problem first.
