Support Widget -- Does not work?

[ollie266]

Hi there,

I am not sure if the OpenAI Help Center is broken currently but I am trying to contact support about an unrecognized charge on my account and it is a painful experience.

After logging in I am heading to the Help Center and interacting with the Support Widget (or at least what I think is the support widget in the bottom right of the screen.

It only prompts me for me email (which is strange as I just logged in and then appears to go around in endless loops with really not helpful information about protecting my account.

It also prompts me to:

Use the Chat Widget: Go to the OpenAI Help Center. At the bottom-right, look for the chat support widget.

• Select “Messages” or “Send us a message.”

• Choose the relevant topic, such as “Payments and Billing.”

• Then, look for an option like “Fraud or unrecognized charges,” or pick the closest option available. If you don’t see that, use the open text box to clearly explain the issue: say you’re being charged for another account or you suspect fraud.

.. but am I in the chat widget? I am not sure it understands (or I dont understand) I am in the widget already and do not see that option; it does not understand when I say i do not see the option because it asks me to refresh my browser

Is the tool broken?

[_j]

It appears that OpenAI has re-made the Intercom widget into looking like a chatbot:

Worse, they took the Intercom widget launcher off the platform site toolbar menu for API users, so there is no alternate. The only “help” there is now also a link to help.openai.com.

Looks like you’ll need to talk to that fake AI instead of being able to directly select “messages” at the bottom, and place actionable “support staff action needed for account” messages if you don’t want help documents parroted back at you.

[ollie266]

0k it looks like I am in the right place, just the responses are not right? Maybe the info driving it is out of date?

”support staff action needed for account” << using this really helped, thank you, it said it escalated it.

This is super frustrating experience, I thought this was the gold standard of using this technology today? Hope it gets fixed.

[_j]

Consider: how high of a percentage of contact attempts from 500 million ChatGPT users are a really just ‘training issue’ or ‘don’t like the way AI talks’.

AI is doing a job of intercepting those against more expensive human resources.

[ollie266]

That makes sense for sure. I would think AI could excel in differentiating, at scale, the two and being a great experience by finding an exception vs. routine, and escalate the exceptions. It appears that step is broken.