As part of the AI takeover we can now no longer set timers for 12 minutes using Google Assistant on google devices.
A very strange bug has turned up for people using a Google's Assistant, all of a sudden it won't set a twelve-minute timer. Happening to people throughout the English-speaking world. You can see what happens if your device has a display - you will see it transcribing the word "twelve" as "12" but then it "corrects" itself to be "1-2" so it sets a 2-minute timer going not a 12 minute one.
Folk are wondering is this because Google have started rolling out their incorporation of their "Bard" AI into Assistant as they said they would. It does look like the prompt is being overwritten as you sometimes see when Bard composes a search reply.
Of course, this is only a minor inconvenience - you can set a timer for "11 minutes and 59 seconds" if you need to but if it is the roll out of Bard it does raise some serious concerns, what happens if the "AI" starts to "correct" itself for something more critical - perhaps you use it for reminders for medication.
How does Google (and the other companies with similar approaches to "AI") test their AI's output? If it can screw up a simple timer request it doesn't bode well for anything complex.
A very strange bug has turned up for people using a Google's Assistant, all of a sudden it won't set a twelve-minute timer. Happening to people throughout the English-speaking world. You can see what happens if your device has a display - you will see it transcribing the word "twelve" as "12" but then it "corrects" itself to be "1-2" so it sets a 2-minute timer going not a 12 minute one.
Folk are wondering is this because Google have started rolling out their incorporation of their "Bard" AI into Assistant as they said they would. It does look like the prompt is being overwritten as you sometimes see when Bard composes a search reply.
Of course, this is only a minor inconvenience - you can set a timer for "11 minutes and 59 seconds" if you need to but if it is the roll out of Bard it does raise some serious concerns, what happens if the "AI" starts to "correct" itself for something more critical - perhaps you use it for reminders for medication.
How does Google (and the other companies with similar approaches to "AI") test their AI's output? If it can screw up a simple timer request it doesn't bode well for anything complex.
via International Skeptics Forum https://ift.tt/gGeYq20
Aucun commentaire:
Enregistrer un commentaire