When creating Jenkins slave nodes via /jenkins/computer one has the option to override environment variables and tool configurations (i.e. the path to the executable) for the slave. If one does not do this the configuration of the master will be used. 

The Amazon EC2 Plugin does create slave nodes automatically if none are present at the time a job requests one. But it does so by creating them without overrides, so the master's settings are used. As our AWS slaves are commonly configured differently than our master it is quite annyoing to have to edit the created nodes by hand every time a new one is started. Otherwise Jenkins won't be able to execute the tools as the path is different. 

Example: tool git
Master config: /somepath/git/bin/git
Actual slave location: /usr/bin/git

The same options which are present in the slave node's config (/jenkins/computer/slave-name/config) "Node properties" should be available in the EC2 plugin's AMI section as the paths should usually not differ when using the same AMI.