It matters actually. Artificial Intelligence research, the magnitude of the term varies significantly given the application of the research/work. Since AI works generally are performed with more of higher end architecture cpu resources with wide variety of capabilities, generally out of bound reach to desktop PC computers. So, the answer is actually it depends on what sort of hardware are you using for your AI research work. Depending on the hardware requirements, the OS decision is generally made. For example, if your hardware requirement tends to be x86, it could be sorted with any normal OS/Linux distro since they perform pretty well for the x86 systems. If the hardware requirement scales to x86_64 or beyond them like sparc or ia_64, then I would choose a OS or distro that scales well to those architectures.
For the matter of the fact, if it happens to be that the research work requires to have real time scheduling and processing with more signal handling capabilities and guaranteed performance, then the choice would be in the group of RTOS - Real Time Operating Systems.
In RTOS segment, there are quiet a few choices ranging for experimental to enterprise like RTLinux, LynxOS, QNX, VxWorks etc.,. Hope this helps.