I noticed this recently when I saw someone post with an AI generated map of Europe which was all wrong. I tried the same and asked ChatGPT to generate a map of Ireland and it was wrong too. So then I asked to find me some accurate maps of Ireland and instead of generating it gave me images and links to proper websites.
Will definitely be remembering to put "generate" vs "find" in my prompts depending on what I'm looking for. Not quite sure how you would train the model to know which answer is more suitable.
I live in Ireland and our contact tracing app (using Google+Apple tech) seems to be doing well. There was about 1 million install 1-2 days after it launched, currently standing at 1.4 million installs (population is 4.9 million), along with some positive contacts reported:
> It also reveals that, whatever the people of Ireland have been told, their app is collecting centralised data on them.
The signup flow has a very obvious and well written opt-in for collection of anonymised data to help track how well the app is working, which was mentioned by the BBC article linked. I don't think people are being mislead.
I personally decided not to enable that, but did give the app my phone number to be shared in the event I'm a close contact.
> it has been in use, it is claimed to have resulted in 91 “close contact exposure alerts”, which is remarkably few.
That they know of, as those users either opted in to the anonymous tracking or (possibly) gave their phone number. It'd be useful to know how the install base per the usual Android/iOS stats compare with the anonymous tracking enablement, but those numbers haven't been shared as far as I'm aware.
Right now our overall case numbers are low enough (around 20 per day) that it's likely hard to tell how effective the app is. However our R is currently estimated at 1.1, so even a little help from the app could help keep us below 1.
Does installed mean active?
Here, the local app had good penetration, but daily activation was rather low and constantly decreasing.
One of the concern is that it impacts battery life, so people turn it off most of the time unless they plan to go to a public space. Then they forget to turn it on.
It’s possible many of those users did not grant the contact tracing permission on the phone when installing.
But I’m assuming anyone who went to both of installing would be open to allowing it.
They could also have uninstalled 10 mins later.
I don’t know how the number is being calculated.
I’m on iOS and keep app installs on my phone to a minimal but haven’t noticed battery drain so far. Android my differ.
Also your phone can collect the data by enabling in Settings > Privacy > Health > COVID-19 without the app at all.
Not that I’m aware of. The app just reads the data your phone is already collecting.
It’s making use of the api that Apple and Google added to iOS and Android. Which I believe once enabled is using Bluetooth beacons to keep a list of other phones you have been nearby. That’s all done by phone, not the app.
If you test positive, then you are expected to use the app to upload the list of phones you were in contact with.
Notifications will then be sent to all those phones through the app.
Will definitely be remembering to put "generate" vs "find" in my prompts depending on what I'm looking for. Not quite sure how you would train the model to know which answer is more suitable.