After Google’s “smart compose” and “smart reply”, here comes this time Google’s “Help Me Write”.
Google’s “Help Me Write” automatically generates a reply to an email after telling the toolkit to write a response. Hence can be called the Email Maker.
Yes, you may need to write what you want “Help me Write” to do, that is to compose a mail but with what inputs.
Smart compose and automatic fills were fine. But help me write goes one step further. Does this mean intrusion into privacy? Or does this mean reading emails and sending them to the server to generate the reply?
For this, one needs to see what is the algorithm behind Auto Composing emails. The steps can be as follows, though exact steps only Google people can explain.
- Give an order to the mail server to generate the reply to an email versus giving the command on the client side.
- This is needed as probably LLM is used to generate the output.
- Use AI on the server versus using AI on the client side.
- Get the AI-generated message from the server or generate AI generated message on the client side.
These things only the service provider can emphasize.
If, however, the command sends data to the server, this would be concerning as the personal email data can be confidential and it should not have been sent on the server for NLP.
And if you are lucky and all new composed mail is generated on client side, then email maker can be used, as it poses lesser danger of data privacy breach.
There are ways in which data can be secured when using a server to generate new emails. If these security concerns are handled by the new toolkit, it’s well and good, else it’s never late and the data security concerns can be added in now.