What if nobody wants to talk to your AI assistant?

Lately a new AI assistant comes out pretty much every week. Grok Bot, Meta's Muse, OpenAI's dots. And every time my feed fills up with the same list of things it's going to do for us. It'll call customer support for you, write your complaints, book a table at that one restaurant that still only takes reservations by phone, or message a guy on Facebook Marketplace and try to get his used desk cheaper while you do something better with your life.

And I want all of that, really. I use AI every day and I have calls I've been putting off for weeks that would take five minutes. If my assistant wants to spend its afternoon listening to hold music, that's fine with me.

But there's one thing I keep thinking about and I almost never see anyone talk about it. Someone will probably pick up the phone or open the email, that's not the problem. But what if nobody wants to talk to your AI assistant? The person on the other end might have zero interest in talking to a bot. Maybe it's their tenth one today or maybe they just think that if it's so important you can call yourself. So they hang up or ignore the email and that's it.

Think about which calls we actually want a bot to make. Nobody needs a bot to call their mom. We want to get rid of the calls where we already expect the other side to be difficult. I've had so many calls with customer support where the person clearly didn't care. I was polite, I explained everything, I stayed calm and later I got a bit more serious, but sometimes none of it mattered. The call ended and the problem was still mine. So now the same person gets a call from my bot instead of me. Are they suddenly going to care more? I really doubt it. If anything a bot is even easier to get rid of, because hanging up on a person feels a bit rude and hanging up on a bot feels like closing a pop-up.

To be fair, a lot of the time it's not even their fault. They have a script, they can't make any real decisions and someone measures how fast they end calls. The calls we hate are usually the ones where the process is already broken and the first person sends you to the second person who sends you back to the first one. So if the people you call don't really care about fixing your problem or they're just too lazy to do anything, they're not going to help your bot either. Why would they?

Email is the same story. You pay some companies directly and some offices with your taxes and still the answer you get depends on the mood of one person that morning. You know the type I mean, every country has one. The person who has been sitting behind the same office window for thirty years, has heard every possible complaint and is already tired of talking to actual humans. Now picture that person getting a call from my AI. I don't think my bot even gets to finish its first sentence.

Ok, there's an obvious answer to all of this and I've thought about it a lot. The bot doesn't need them to care. It can wait on hold for three hours, call back twenty times and ask for a manager every single time without getting angry, which is more than I can say about myself after the fourth "unfortunately". Maybe that works sometimes, I don't know. But I can't see companies just accepting an army of bots that never get tired. When something takes too much of their time they find a way to block it. Maybe they add more checks or maybe they just stop talking to bots. Then someone builds a bot that gets around that and we're back where we started, just with more steps.

And what happens when my bot calls their bot? A lot of support lines already start with a bot anyway. Companies spent years putting bots between us and their people and now we're building bots to talk to their bots. So it's two computers talking over a phone line with fake voices. We basically reinvented the API, except now both sides can interrupt each other and get the street name wrong.

I was thinking about this recently when I had to call the electricity emergency line because of a power outage. Years ago a person would pick up, I'd say the address, they'd confirm it and it was done in a few seconds. This time I got a machine that asked me for the city, repeated it and asked me to confirm it, then did the same with the street and then with the house number. I think it took like eight minutes. I get it, nobody should be copying addresses all day, so automate that, fine. But this wasn't even some fancy AI, just one of those old phone menus and it turned three form fields into a whole conversation. Now imagine my assistant calling it. My bot and their bot reading an address to each other for minutes while somewhere two databases are waiting for data they could've sent in a millisecond.

There's also a much simpler reason I don't think people will love these calls. They already don't. Way before any of this AI stuff I'd hang up the second I realized a call was automated. It took me two seconds max, because the voice was off or the pause was weird or the first sentence was something no human would ever say. Yeah, the new voices are really good and maybe a bot could call me today and I wouldn't even notice. But then it only works because I think I'm talking to a person? That's a weird thing to build a product on and the moment I find out I'm going to feel kind of tricked.

There's the cost too. When I call someone it costs me my time and it's stressful. The other person kind of knows that, so the call means something. If my bot calls it costs me nothing, but for them it costs exactly the same as before. That doesn't feel very fair when I write it down. Same with Marketplace. Imagine you're selling a desk and you get fifteen super polite messages from fifteen bots, each offering a bit less than your price. Which one are you answering? Probably none. You're answering the one guy who writes "can pick it up in an hour, cash".

I'm not saying this will never work. If a bot called me wanting to buy something at a good price, suddenly I'd be very patient. A restaurant wants to fill its tables and a shop wants to sell stuff, so when both sides win, sure, bring the bots. Most of those places already have a booking page or a form or a chat, but not all of them. There are still restaurants where booking means a phone behind the bar and someone trying to hear you over the coffee machine and there the assistant makes total sense.

It's just funny to me that the examples everyone loves are exactly the hard ones, like negotiating a bill, getting a refund or arguing with some company. We're sending the bot because we don't want to deal with that person and that's exactly why that person won't want to deal with the bot either.

And I really do want AI to do stuff for me. A bot clicking through some horrible website, filling forms, comparing options or fighting with an awful internal system? Yes please. That's boring computer work on my side and nobody else has to pay for it with their time. A call is different, because it takes someone else's attention. And we keep talking about that person like they're part of the tool, like they'll happily help because my assistant speaks clearly and never gets tired.

Maybe that changes. Maybe companies will open special lines just for agents. Honestly the best version is where assistants don't need to call humans at all, because everything has a proper way to connect. But until then I keep picturing my assistant calling that office window for the fifth time today, super calm and super polite with a great argument ready. And the person behind the window just doesn't feel like talking to a bot.