The real problem with the current AP systems -- the ones that ~30 groups are racing each other to get approved and bring to market -- is that they are all reliant on inadequate technologies. First of all, an AP is reliant on an insulin pump that can only deliver insulin subcutaneously (not via the portal circulation like an actual pancreas). Even though most pumps are filled with rapid-acting analogs, the absorption of insulin is still somewhat time delayed and cannot be shut off quickly enough for spontaneous activity or when glucose levels start to plummet during or following exercise. The only real benefit of pumps over injections is that pumps make it easier to lower levels of basal insulin prior to, during, and following the activity, but the pump user currently has to make those decisions and plan ahead.
Second, everyone is trying to be the first to create a completely closed-loop system driven by algorithms that will manage insulin delivery without user input, but I fail to see how that can ever work effectively to reduce circulating insulin levels in time if exercise is "unannounced" to the AP system and, therefore, lacking the time to lower basal insulin delivery in advance. Some exercisers lower their basal insulin for 1-2 hours before they start an activity, along with during it. Current AP systems are not likely to ever effectively manage spontaneous activities without requiring the user to consume extra glucose and calories, which would make weight gain likely.
Some groups are trying to overcome this issue by using dual delivery of insulin and glucagon, but currently that would require wearing two separate pumps unless a dual-chamber pump can be developed. Then there are issues with the shelf-life and stability of glucagon. Personally, I would much rather treat lows with a mini-dose of glucagon that have to eat more -- especially when I'm not hungry -- so hopefully the companies working on improving the stability of glucagon can soon make glucagon pens to treat hypoglycemia, regardless of what happens with AP systems.
Moreover, the continuous glucose monitors (CGM) that are being paired with a number of different insulin pumps for various AP systems are all seriously flawed in their abilities to work well for exercise. Even when resting, CGM technology is limited by measuring interstitial glucose levels, which lag behind blood glucose values by as much as 20 minutes. If you have to run through the airport to catch your next flight (an intense, but short, activity that will likely raise your blood glucose), it's likely that you'll be seated and ready to take off before your CGM even begins to register that your blood glucose was affected by your airport sprint. My own personal experience with using CGMs has been less than inspiring as well, given that I seldom have had sensors work the full seven days, give accurate readings in the critical lower ranges (60-120 mg/dl), or even trend correctly -- even without exercise as an added variable. They ought to come with a warning label that says, "User experiences may vary by individual, and this device may or may not work well for you."
Finally, some groups are trying to use exercise monitoring devices to measure heart rate and movement so that "smarter" algorithms can be developed for closed-loop AP systems that will be able to interpret and adjust for exercise, but these are a long way from being effective yet. Having all of these extra measurements to factor in also means an AP user would have to continuously wear some type of biometric data collection device or devices (in my case, both a heart rate monitor and an accelerometer when I was trying to get all the information a system would need to make those determinations), along with an insulin pump and a CGM, and carry around the device(s) controlling the integration of all of these various technologies.
All in all, I personally think it will be a long time before a truly effective closed-loop AP system is developed and ready for consumers to use. From a marketing standpoint, using all these devices may be a hard sell to the average person with diabetes (especially since troubleshooting of various devices will be inevitable) and the expense may be prohibitive. Also, who really wants to wear upwards of four or five different devices at all times for the rest of their lives? A real, functioning pancreas would be a much better "cure" in my opinion, especially for anyone who wants to exercise regularly!