When an AI agent is too stupid to exist.

[MineralMan]

I called my clinic to make an appointment with my doctor. I called the number the clinic's website said to call to book appointments.

The AI agent answered the phone,

"Hi, this is Alli, your AI assistant. How can I help you?"
"I need to make an appointment, please."
"Who is your doctor?"
"Dr. [last name] (He is Pakistani. I pronounced his name as he does.)
"I don't find a doctor with that name. Can you give me his full name?"
" Dr. [first name & last name]"
"I still do not find a doctor with that name at this clinic. Which clinic are you calling/"
"The [city name] Clinic."
"":Let me check. No, I don't see that name at that clinic either. I'm sorry, but I am unable to help you."

Then it hung up on me. It didn't offer to transfer me to a human voice. It just hung hp without further comment."

I suppose it didn't see the name based on how it decided it was spelled. I don't know. I do know the doctor and the clinic, though. So, I called the nurse line. The person who answered the phone knew the name I said and quickly made an appointment for me with that doctor, offering me a range of days and times. Took less than a minute.

So, Alli, the AI agent has no more skills than the old voice menu system they used to use. There was no intelligence in that, and none in the AI system. However, in the voice menu system, the last choice on the menu was "Speak to someone in real time." That option no longer exists. I know that there is a staffed nurse line. Some people might not know that. Someone who got hung up on by "Alli" and who probably never called back.

AI assistants should have at least some intelligence. They have not been coded well if they do not. The old voice menu system was better thought out and never just hung up on the caller.

There has been zero improvement through using AI for this purpose. None. No improvment; no advantage provided.

[question everything]

1. Most of my clinics expect appointments to be done online. Actually works OK. So far..

[MineralMan]

2. Sure

I know how to do that, too. I wanted to see how the AI assistant they keep bragging about is working. It's not. Somehow, they're proud of something that doesn't work. Odd.

Further, there are many people out there who are helpless with online things. More than you can imagine.

[dickthegrouch]

3. I tried to pay a $56 bill for friend in need yesterday

Automatics failed twice.
Eventually spoke with three different representatives after fighting more automation.
Spent 50 minutes paying that $56 bill 30 minutes of which was connected to an actual person.
By my estimate they lost ~$44 on the transaction.
Insane, unbelievable and EXCEEDINGLY INCONVENIENT.
T-Mobile complaining at you!

[TheProle]

4. AI is getting smarter and real humans are getting dumber.

[dickthegrouch]

5. Keep telling yourself that

Abundant Iniquity will never be anything more than statistical suggestions of what word comes next.

[TheProle]

7. Will do. I see it in action every day, recognize its flaws and am amazed by its capacity.

Enjoy your horse and buggy.

[MineralMan]

8. BS.

[Jirel]

6. Refuse to talk to them.

It's what I do. I either repeat “give me a human” until it complies, or I start playing extension roulette until I reach a human. Pro tip: they almost always answer their BILLING line, or the doctor to doctor referral line (if they have one).

[MineralMan]

9. OK. I prefer a different approach.

I know how to reach a human. However, I also like to test new systems. I report issues like this one in writing, and suggest changes. My goal is to improve access for others. I don't yell. I explain. I believe that works better.