A few weeks ago, I explicitly disabled it everywhere I could find within account.google.com and I’ve not used anything gemini since. Now I get this email and find it enabled on my devices.
By default Gemini has permissions to access everything on screen, view your contacts/messages, and can be used from the lockscreen…
I’m not all that surprised; but I’m still annoyed. Especially with the opt-out of data collection/access after it’s been given access to everything.
I tried Gemini before. The problem was reliability.
E. g. I say “add an event to my calendar: tomorrow 2 p.m., doctor visit”, and it parses the voice perfectly. The regular assistant would then just create the entry.
But Gemini sometimes goes like: “Adding an entry to your calendar is easy! Do you have an iPhone? Then these are the steps: …”
This is why LLMs are not appropriate for applications where regular syntax is needed, like calling an API. I don’t understand why they made the LLM the first step, handing it off to the old hard coded Google assistant second, rather than the other way around. Having everything go through the LLM first is wasteful, slow and unpredictable. I am very confused about Google’s decision here
All of the voice assistants have been complete dogshit since day 1 (sometimes even getting worse) and yet somehow we still have them.
The funny part is when you say something and it corrects you
Or that weird thing ChatGPT was doing were it screamed “no” in your own voice before replying like nothing happened.
Edit: it was “no” not “help”
Wait what the fuck
https://arstechnica.com/information-technology/2024/08/chatgpt-unexpectedly-began-speaking-in-a-users-cloned-voice-during-testing/