I don't remember there being typos when captions were generated by dedicated graphics operators using systems like aston etc. Just because journalists enter this information into the system themselves doesn't excuse not proof reading what they have entered.
Yep - though even then the astons were only as good as the operators in the chain. Two people typing on two keyboards also increases the chances of typos.
In many traditional galleries there would have been :
Producer/Assistant Producer who typed the aston request (or hand wrote it)
Production Assistant/Script Supervisor who will have put the aston request in a script (or seen it there and also looked at it), and will also have seen the caption on a screen in the gallery during transmission before it was aired, as well as counting to the aston time if it was tied to a specific point in a
VT/film.
Graphics Operator who will have actually typed the aston into the Aston machine
Director who will have seen the aston request in the script and looked at it, and also previewed it before directing it to air
Vision Mixer who will have transitioned the keyer to put the aston on air.
Now - in many systems the journalist enters the aston and an aston time and the graphic will be automatically generated, and keyed on-screen automatically at the time the journalist has entered, with no preview by anyone.