Sollte es einen legalen „Mit einem Menschen sprechen“-Button geben, bevor KI-Kundenservice zum Standard wird?
https://economictimes.indiatimes.com/tech/artificial-intelligence/over-78-of-firms-expect-ai-agents-to-handle-customer-support-within-18-months-adobe-finds/articleshow/131999118.cms?utm_source=chatgpt.com&from=mdr
28 Kommentare
Submission statement: Adobe’s 2026 AI and Digital Trends research suggests that organizations expect agentic AI to move quickly into customer support, with 78% expecting AI agents to handle customer support interactions within the next 18 months, while only 16% report organization-wide deployment today.
If AI becomes the default layer between people and companies, the future issue may not be whether AI can answer routine questions. It may be whether people can escape the automated layer when the issue involves money, health, travel, identity, legal access, billing, fraud, or an unusual case that does not fit the menu.
Should future customer service systems be required to offer a clear “talk to a human” option? Should that right apply only in high-stakes sectors such as banking, healthcare, insurance, government services and travel, or should it apply broadly? And would such a requirement protect trust, or make AI-first support too expensive to scale?
I would rather requirements that anything your customer service AI agent agrees to, you’re beholden to honor. No clawbacks.
Yes.
AI: Could you describe the problem you’re having?
Me: I can’t log in.
AI: Try changing your password.
Me. I don’t receive the email for the password change.
AI: What’s you username?
Me: redclawx
AI: I can’t seem to pull up your account.
Me: Let me talk to a real person.
AI: Could you describe the problem you’re having?
No but companies that offer it will destroy the completion that doesn’t
„Of course! We’ll be happy to connect you to a human.
Your estimated hold time is 9 hours.“
Facebook has no way to contact them which is absolutely ridiculous . AI should never replace human customer service
I called my cell company the other day and it was ridiculous how I kept saying customer service and AI kept asking me questions. In the end it said “ im having trouble understanding you,let me get you to someone who can help“
I think it is fine for the initial interaction to be will AI, but if not satisfied with the AI interaction you should be able to talk to a human then. But not just talk to a human, the human needs to have decision making authority. For example one of the big banks closes your account. Yes, you can call up customer service and talk to a human but often they can’t do anything about it „the algorithm made the decision to close your account“.
AI in a customer support position should only be for finding support pages and FAQ.
:: sarcastic tone:: „I know you’d like to speak with a human, but…“ 😒
Only if the user is able to cogently explain their issue without deception.
Most of the work of CSRs is trying to figure out what people who lack the intelligence to correctly explain their problems want. The other side is the people that may be more intelligent, but they just lie their asses off for some reason.
If you have ever worked with the public for any extended period of time, it becomes rapidly apparent that the vast majority of people cannot even articulate their ‚problem‘ or become frustrated and abusive extremely easily because they are unable to properly communicate.
A similarly sized population are pathological liars.
The issue usually isn’t the automated system, it is the worthless user base that has no idea what it is doing.
Am I the only that experiences rage when being forced to talk to a machine?
I’ve never had Alexa or any crap like that. I despise voice interfaces. Crap AI customer support would just make me rage even more.
Any chance of this becoming required is going to be a big ole no thanks from lobbyist.
In 2026, what “should” be doesn’t exist and everything is lame.
I was literally using Gemini last night to get an API from Google cloud setup and for the life of me could NOT get it to tell me my account ID for a my business account.
It took an hour of massaging the correct prompts and STACKOVERFLOW incidents for me to finally get it.
This is Google…can’t provide a simple ID for a property associated with an account im logged in to.
But these toys should be let loose everywhere as CSRs lol.
I get some of the reasons they use it (outside of cost cutting), like people calling for basic info (like operation hours), or to mitigate volume, but if I’m calling a business it’s because I have a super specific question or a quick simple question that AI can’t answer. Having to wait through layers of dialogue trees when I’m trying to find out if they’re available for walk-ins today or something is *infuriating*.
Please yes.
I tried BestBuys online CS. It freaking ran me un circles for 10mins before I could find anything about a human.
how not a legal requirement anything an AI tells a customer, companies have to honor? that’ll slow their roll
I don’t know, I guess? Will it change anything? Probably not…
As you may or may not notice, especially with larger corporations, they are posting messages that are more or less saying „What you can see and do on the site is the extent of what our CS agents can do/have info on“. Since it was a race to the bottom to find the cheapest possible staff, they also don’t want to give them the power to do other things. That meatbag CS agent won’t have any more ability to do things than the bot, so I suspect you may be escalating to a human in either case.
Has any one here had experience with customer service in the last 15 years? It’s been dog shit for as long as I’ve been using it. When you get escalated to the human it’s going to be a 3 hour wait on hold for them to say „sorry we can’t do anything“ and hang up.
The goal should be AI for all initial contact and then escalate to a human as a fall back. The IT department/vendor should be judged on how many calls get escalated to a human.
No there shouldn’t, talking to a human is the default. End of story.
In the UK it is a legal requirement to list an email address on your website. Unfortunately the absolute vast majority violate that law.
It’s also a huge accessibility issue.
AI customer service should be illegal. Big tech companies should be forced to have minimum human response time to the tickets. For example right now you almost can’t talk to a human in Anthropic, this is unacceptable.
I forgot whether or not I left a cast iron pan on an active burner, two hours from my apartment. Immediate emergency, due to fire risk, right? I call my building to ask for someone to go into my apartment and check, and I get my buildings -new- AI customer bullshit instead of a real person. I started asking to be connected to a person and all I got was „I understand. Can you explain the issue to me?“ EVERY FUCKING TIME I ASKED TO TALK TO A PERSON. The fucking thing only shut the fuck up and listened when I said „fire risk“. Eventually I give up and explain the maintenance issue and the thing says „Okay your maintenance request has been logged. Goodbye!“
A few minutes later I got this text from the AI bot.
„I apologize but I was unable to reach our team at this time. We will continue trying to contact them to assist you.“
Eventually I got a call from a maintenance guy who said the stove was off when he got inside.
Talking to a fucking person would have made my experience significantly less stressful and frustrating. AI can in no way replace human contact. I dont need a robot telling me „I understand you are frustrated.“ and „Can you please repeat that?“ There should absolutely be a „speak to a human“ command.
Sorry for my rant.
Sure. The benefit would mostly accrue to foreign support staff (think poorer people) while the costs would mostly be borne by western (think richer people) customers.
Not ideal from market efficiency perspective but it would a relatively agreeable form of wealth transfer.
AI phone agents are worthless when you actually need help
Yeah, I think that’s a pretty critical point. AI is great for simple stuff, but when you have a complex problem or something emotionally charged, you just need a human. It’s not just about getting an answer, it’s about feeling heard.
Plus, if something goes really wrong, who’s accountable? The AI? Companies need to keep an easy path to a real person for tricky situations. Making it a legal requirement makes sense.
Yes. Any service call I make starts with „operator, human, person, representative“. Quit wasting my time and I won’t waste yours.