# Prediction 1 — Phill Agnew: AI must 'show its working' & the rise of text speak

- **Episode:** [262](https://agents.truthliesandwork.com/episodes/262.md) (2025-12-30)
- **Type:** other
- **Tags:** ai, transparency, text-speak, communication
- **Time:** 4:36–11:12 (start 276s)
- **Listen:** [https://truthliesandwork.com/262?t=276](https://truthliesandwork.com/262?t=276)

---

[4:36] Al Elliott: Nice. Welcome back. Right. Our first prediction comes from a friend of the show, fellow HubSpot Network podcaster and host of the UK's number one marketing podcast, The always cheerful, always amazing Phil Agnew from the Nudge Podcast. My prediction, like probably most people's prediction on today's show, is to do with AI, because what else is there to predict other than AI? I have 2 predictions actually. Both of them are based around a bit of psychology because that's where my head's usually at. I'm often reading behavioral science journals and figuring out how people make decisions. The first one is this idea that I think I think AI, come next year, will start to be very clear in terms of showing it's working. So right now it sort of sometimes says what it might be doing and then gives you an answer, but it doesn't really tell you what it's been doing. And I think AI companies will actually start to build tools to show the working of the AI. So when they write you an email, it'll give you, you'll be able to click a button. It'll say, here's exactly why I wrote the email in that way. Here's why I used this line rather than this line. Here's why I said this rather than this. I think the reason they will do this is because Showing the working will make customers actually value the output even more because the magic has worn off now. So two years ago, when we were using ChatGPT to write an email, it was magical. It really blew our mind. Now it just seems normal, and actually, I think we undervalue how impressive it is because it just happens so fast and we don't really think about it. And I think AI companies will realize something that psychologists have realized for a long time, which is that if you show you're working, for example, people will value the thing you create more. I'm sure I've heard an episode around open kitchens or something, which is where you can see the chef actually make it. Is this a similar kind of idea? So that's a study by Ryan Buell from Harvard University, and he's studying something called operational transparency. And it's exactly that. In Harvard canteens, he sets up these studies where sometimes the people who are dining in the canteen can see the chef cooking Their meal. And other times he places a partition between the chefs and the diners, and so no one can see each other. And what you find consistently is that diners enjoy their meal more when they can see the chef, whether they could see the chef working. Interestingly, the chefs also enjoy creating the meal more when they can see diners enjoying the meal. So this is operational transparency. If you can see behind the scenes, you value things more. And there's other elements of this as well, which, and I think the one which is pertinent to the AI example, Is that this is again a Michael Norton and Ryan Burrell study. They did a study with Kayak where people were shown the Kayak homepage. They entered their search. So I want to do whitewater rafting in New Zealand. And Kayak, because this was back in the sort of 2010s, early 2010s, it took a bit of time to load. So there was a loading screen. Sometimes the loading screen just showed a wheel loading. And other times the loading screen showed it's working. When it does that loading screen, they find that people value the results more than when the loading screen is blank. Interestingly, if they show they're working, the loading screen can take like 3 times as long and people still value the results more because you can see all the effort going into it. And I think because AI has got so fast, it's become so omnipresent, it's lost a bit of that. magic, a bit of that shine. And I think companies will realize that if they actually reveal a bit of what's going on behind the, under the hood, people will value it a bit more for that reason. So you have a second prediction? Yeah, this one's silly. And it's for the same reason. There's this weird thing that happens when you realize something is created by AI, which is you start to devalue it, undervalue it. So there's this amazing study we've shared on my show by Professor Matt Johnson, where he shows individuals 2 pieces of art. Now, it's a podcast, so I can't describe, I can't show you the art, but one of the pieces of art is actually quite beautiful. It's really symmetrical. The colors sort of mesh really nicely. It's pleasing to look at. The other piece of art is a bit janky and it's not quite as nice to look at. So when people look at these 2 pieces of art, me included, and you're asked, which do you prefer? Pretty much everyone prefers that one, which is nice to look at. I did too. However, the study then reveals to the viewers that that piece of art, the nice one to look at, is AI-generated, and the janky one is human-created. And then they show these labels to people, and they find that when the piece of art is labeled as AI-generated, we actually don't like it anywhere near as much. And we suddenly say, oh no, I much prefer the human-created piece of art. I don't like the AI piece of art. I think this is maybe a bit of a problem with AI. I know you see it a lot. I spend a lot of time on LinkedIn because I'm a loser, and you see it a lot on LinkedIn. You see people calling out other people for using em hyphens and clearly using AI comments and clearly using AI. And so my prediction next year as a sort of fight back against AI is that text speak is going to become a thing again. So text speak, you know how we always used to speak on texts when you paid by the letter and you couldn't be asked to write out a See you late with an 8. I think that's going to start coming back or dripping into our vocabulary. I think there's going to be this weird value that people place on messages that have so clearly been not created by an AI because they're chaotic and weird and grammatically incorrect in some ways or cut short in others or meshing words together in a way that AI wouldn't. Previously, before AI, that would have just been nonsensical and would have looked unprofessional. But now, in a world where potentially so many of the messages, especially on somewhere like LinkedIn, are written by a robot, not a human, that'll become really weirdly valuable
[10:49] Leanne Elliott: If you enjoyed that, you absolutely need to check out Nudge. And here's the thing, right? You're already 90% of the way there because you've got your podcast app open right now. So press pause, type in Nudge, press follow, press play. We're all happy
[11:02] Al Elliott: Well, don't press play just yet because I've got a few more predictions for you
[11:06] Leanne Elliott: So I meant come back and then press play on host. Wasn't very clear there
[11:09] Al Elliott: Well, okay, on to prediction number 2 then
