Currently the HTML generator protect each character having a special meaning in HTML, typically the character < i produced as <
There are several ways to avoid that :
[*] to add an option to produce the content of the descriptions as they are, without protection
[*] to add an option to search for the HTML tags and to produce them unchanged, in this case a single < will produce < but <i> will produce <i>. Of course this option is more complicated to manage, because this supposes to know all tags ...
What do you prefer ( may be an other way ) ?
However, let's note in my thinking the normal use of the HTML generator is not to use it as it is, but to adapt it to your own needs
