Is what I type being watched somewhere? A talk about AI and data
Since I started using AI, there is something I have been somewhat careful about.
I have been avoiding entering things like my family's names, financial information, or detailed work matters.
However, I haven't been able to explain well why I do that.
I have been using it while feeling a vague sense of unease, like 'it feels creepy' or 'I'm just a bit worried'.
The AI of today is not inside your smartphone
This is obvious once it's pointed out, but at first, I didn't really understand it.
ChatGPT, Claude, and Gemini are not inside your smartphone.
They are inside massive computers on the other side of the internet.
Every time you talk to them, your words fly over there.
When I realized it was like 'calling a distant telephone counseling service,' it made a bit more sense.
The feeling that the person on the other end is kind, but you wonder 'isn't this conversation being recorded?'—that feeling was correct.
'Being saved' and 'being used for training' are different things
This is a point I want to clarify.
It is the same for almost every AI that the content you enter is saved on a server.
That is why you can look back at your conversations later.
But being used for training is a different matter.
Training means that my conversations become data used to create the next version of the AI.
'Remaining' does not equal 'being used for training.'
Now, this is sorted out.
I checked the settings for three AIs
I was curious, so I actually checked three AIs.
Claude
According to my research, it seems the default setting is not to use data for training. It says that even without changing anything, the content you speak will not be used as training data.
ChatGPT
When I checked, it was turned on.
Settings → Data Controls → 'Improve the model for everyone'—this was left on.
I had been using it with that setting without knowing. I changed it to off. (By the way, if you don't want to keep your history, there is also a 'Temporary Chat' mode. This way, the conversation disappears as a one-time thing.)
Gemini
My learning setting was also turned on. Since it is a Google service, I changed it from within account management. It was a bit hard to find. Google Account Management → Data & Privacy → "Gemini Apps Activity" → Change to Off
It seems that when learning is turned on, content may be reviewed by "human reviewers" after being anonymized. So this is another reason why it is better not to enter personal information.
It seems that future AI will move into our smartphones.
While looking at what lies ahead, I found something interesting.
It appears that "small AI that works without an internet connection" is emerging.
Since it runs directly on the smartphone, the content you enter does not go outside.
It can be used even in places without a signal.
The day may come when family worries and diaries can be kept entirely within your smartphone.
The feeling that "it is scary to enter personal things" might change with the evolution of technology.
What I am doing now
I changed both ChatGPT and Gemini to off.
Also, whenever I use any AI, I make sure not to enter names, addresses, or passwords. Just doing that makes a big difference in how secure I feel.
Before checking the settings, I was only careful about what I talked about. Since learning that I can choose in the settings, I feel a bit more at ease.
Well then.
Rather than using it while feeling uneasy, looking into it a little bit made me feel much calmer.
Just knowing about it makes you feel a bit more secure, doesn't it?
If you enjoyed reading to the end, I would be happy if you could click the ♡ button.
Thank you for reading to the end.
I am challenging myself to post every day.
If you follow me, you will receive notifications for my next articles.
Written by: Sate. | Digital Utilization Supporter
I have spent 30 years at an IT company creating "easy-to-understand manuals."
Together, we will turn the "I don't understand IT" of those in their 50s and beyond into "Oh, I can do it after all."