I am an astronomer, and I can tell you that, in most of our opinions, Adaptive Optics is a complete load of crap. A lot of money has gone into that technology for very little science return. The price per scientific paper for Adaptive Optics is far higher than for other technologies.
Yes, the technology can make images a bit sharper, but due to its intrinsic properties, results in those images being completely impossible to calibrate. Making a measurement is only 1% of science; 99% of the challenge is understanding the context of the measurement, its reliability, and uncertainties. Without the ability to calibrate the image, it is useless. With adaptive optics, there are image artifacts that vary in time and can mimic details for which you are searching. If you see a new point of light around the star you are imaging, is it really a distinct object, or just an artifact of the image?
Also, adaptive optics requires bright stars as guides, because the system must operate faster than the atmosphere varies (generally 1 kHz). Very little of the sky has a bright enough star nearby for this to be useful on many objects.
You can use lasers to make artificial guide stars. You still need an actual star for one stage of the correction that the laser isn't sensitive to (because it goes round-trip), but it can be fainter, and this opens much more of the sky. However, it's still far from complete sky coverage.
Finally, background light is higher for ground-based observatories. Hubble can still see fainter than ground telescopes because of this.
For versatility, large sky coverage, and faintness in high resolution imaging, you just can't beat going to space.