S2. Ep36 - The AI Labs Are Hitting the Brakes — And It Won't Affect Your Business At All

Katie (00:27)

Hello and welcome back to another episode. Hi, hello, I'm Katie and as always I have Noel here with me on today's podcast episode. Hi Noel, how are you doing?

Noel (00:41)

I'm doing absolutely fabulous yet again this week. How are you doing?

Katie (00:44)

Good, had a good week so far.

Noel (00:46)

Very good week. Yeah, it's been a busy week, lots of things going on. I've got a new release coming out for Clyde soon.

Katie (00:55)

Exciting. What does that involve?

Noel (01:00)

Well, we all love a group chat, so that is the new feature.

Katie (01:06)

And has this group chat got a really unhinged name, or is it literally just group chat?

Noel (01:12)

I haven't thought about an unhinged name. I could probably give it one. But essentially, with Clyde, the AI agent that I built, you talk to Clyde and then he goes off and talks to the team. But I was kind of thinking, well, sometimes you don't need the manager in every meeting. You could just have a chat with the team member, or a few of them. So yeah, you can create little group chats.

Katie (01:31)

Yeah, sometimes it can even be an email.

Noel (01:37)

It could be exactly that. So yeah, that's going to come out this week, hopefully.

Katie (01:44)

Okay, so people will have their own group chat. It's not a whole group chat for the whole community, that's still available on the Skool platform.

Noel (01:53)

No, it is, yes. So there's a community to come and chat with me and other people that are building with Clyde. The group chats are just you and your agents within the Clyde session.

Katie (02:06)

Okay, good to know.

Noel (02:10)

We don't need unhinged chats between multiple people's agents.

Katie (02:15)

No. Okay, that's fair enough. So any updates that we should be aware of this week?

Noel (02:25)

Yeah, so OpenAI released their latest image models. So if you're into creating image assets and things like that for your business, then great, great little model. Although, as with every model release, comes a new trend.

Katie (02:40)

Yep. And this time it's the 80s. What did you look like in the 80s? And some of them are really good that I've seen.

Katie (02:53)

But I love the people who aren't really into using AI and they're just posting actual photos of what they looked like in the 80s. And I was like, yeah, I am here for that.

Katie (03:11)

If I had a photo to hand of me as a child in the 80s, I would 100% be posting that. But sadly I don't have one to hand. Sad times.

Noel (03:26)

No, I know. But that trend, I remember we used to get in the almost matching tracksuits and shell suits and things like that.

Katie (03:39)

Yeah.

Noel (03:44)

We were a walking fire hazard in the 80s.

Katie (03:47)

Yeah, and created a lot of static. Maybe that's why the hair was so big.

Noel (03:56)

Well, that in itself, if it has a bit of the hairspray.

Katie (03:59)

True.

Noel (04:03)

Yeah, it'd be a lot safer to create an image of you in a shell suit.

Katie (04:08)

Yes. Okay, any other updates, other than people creating 80s images?

Noel (04:18)

So DeepSeek came back, came out with a new model last week, which is really interesting actually, because it's a lot cheaper than most of the models. Compared to GPT-6 Astra, or Claude Opus 5 from Anthropic, it's kind of at that sort of level in terms of intelligence and usability, but it's hardly any price at all. It's super cheap on the API.

Noel (04:50)

So definitely worth checking out if costs in your automations are really important. You can go off and check that one out. I checked it out last week and I quite enjoyed using it, actually. It was really, really good.

Katie (05:06)

I kind of thought we had ruled out DeepSeek. I literally didn't think anyone was using it this year. I thought it had kind of died a death, as it were, because I don't think we've spoken much about DeepSeek on this podcast since last year really, and it's only had a few brief mentions.

Noel (05:35)

Yeah. I mean, it changed the way in which we do reasoning and thought. That was the big thing when it first came out, because it was the first model to do it. So that was really exciting. But I think with this one, what's getting people excited is the knowledge and the intelligence behind the model for a fraction of the price.

Noel (06:00)

On the API, if you're using it in your automations, the output token usage cost is like 60 cents per million tokens. Compare that to something like Fable, that's $50. So it's making people maybe think twice about which model they look at. But I would say worth a test. Whether or not you adopt it, that's up to you.

Katie (06:25)

Okay, good points.

Noel (06:28)

Yeah. And obviously the main part of this podcast is also another big bit of news, but we'll get into that.

Katie (06:36)

So Noel, do you want to tell us what we are talking about in this week's podcast episode?

Noel (06:43)

So in this week's episode, we're going to go through some of the main news that's come out of the AI labs this week, where OpenAI, Anthropic and Gemini are going to put the brakes on AI development. So we're just going to go through that and how that could potentially impact your business.

Katie (07:00)

Okay, so we're saying these AI labs are talking about slowing down AI, but what does that actually mean for us, for business owners?

Noel (07:16)

So what that actually means is they're slowing down the development at the very frontier end. Where they're looking at artificial general intelligence and creating models which are even more intelligent than what we have today, they're starting to slow that down, or they would like to slow it down. Whether or not they will is another thing. But they have a few concerns, mainly from a security perspective.

Noel (07:45)

Yes, they can create them, but they want to slow it down just so they can make sure that these models are safe when they get released into the wild. So that's kind of where they're coming from.

Katie (07:57)

Okay, so nothing has changed at the moment, it's just a conversation that they're having.

Noel (08:10)

Yes, but it's an interesting conversation, because it's actually unified across all of the main AI labs in America. So even Elon Musk's xAI has agreed with everybody else, which goes the opposite to everyone else sometimes.

Katie (08:26)

Yeah.

Noel (08:27)

So everyone's agreed with it. They're all saying, we need to maybe take a step back and slow the pace. Because like we said, we get through new models every couple of weeks from each of these labs. Do we need that? Probably not. Could we slow it down and make things safer? I think that's probably where we'll end up.

Katie (08:51)

So if AI does start slowing down in terms of development rather than usage, that's not what we're talking about, we're talking about new things coming out, new models coming out. How do you think this is going to affect us as business owners?

Noel (09:12)

So I think for most of us, it's not going to affect us at all. The people that chase the shiny objects, it will affect them, because they'll have nothing to chase, or they will have fewer things to chase. But for most people, like me and you, Katie, and people that listen to this podcast, it's not going to affect us in any way. We have agents already built that do stuff for us anyway, and those agents aren't working at the pinnacle of what AI could do.

Noel (09:50)

They're going off, they're creating written content, they might be reviewing things within your business. They could even be creating images of you in a shell suit, who knows. But it's not going to impact us that much, because as a normal business, you're kind of still at the ground floor level in terms of using it to its full capacity.

Katie (10:09)

Yeah, so the majority of business owners still haven't used all that's available to them.

Noel (10:20)

True. Even when you're creating an AI app, they always say you need to be pushing the envelope when it comes to the AI within the app, otherwise within two weeks there's no point, because the next thing's just going to overtake it.

Noel (10:42)

But most of us, we don't use AI for that. If you're just using ChatGPT or Claude to do mundane tasks, or repetitive tasks, that's just never going to change with this slowdown. It just means you'll be using the same model for longer. We just have to get used to Opus 5 and GPT-6.

Katie (11:11)

So what does this really mean, then, for the people listening to this podcast?

Noel (11:20)

I guess what it means is it gives us a bit of a point where we can have a break. We can take stock and have a look at what's running our business currently, and how we're using AI within the business. I think at the minute we can always find ourselves chasing the next big thing. We've integrated this model, something else comes out and you go, wow, now we need to switch it out.

Noel (11:55)

We always seem to be chasing ourselves with AI. So I think this gives us that chance to step back, have a look at what we do and how we do it. And I guess look at future-proofing as well, because this is a good point to look at the models we use and then go, right, well, if that disappeared tomorrow, because the AI lab can no longer afford to keep it running, or they just decide to end support for it, how is that going to affect us? What are we going to lose in our business output?

Noel (12:25)

So I think it gives us a nice little break, which is very much needed, especially for me. I'm always looking at what's next. So yeah, time to take stock is what I would say.

Katie (12:35)

Yeah, okay. And I guess as well, we talk about so many different things with AI and automations within business, but there's this thing that you should be able to do all of it. And actually, you don't need to be doing all of it, or to be good at all of it. If you've got, say, a coaching business where you coach people, you run courses and events, you don't necessarily need to be able to build an app.

Noel (13:13)

No. Nice to have, but you don't need it.

Katie (13:17)

No. And there are going to be lots of different examples where maybe you've got a brick and mortar shop, and actually you're using it to promote your shop and your products and maybe your events. So there are going to be lots of things that you don't even touch in the world of AI or automations, and literally you are just using the really basics of AI.

Noel (13:57)

Yes, absolutely.

Katie (13:59)

So some people won't even notice the AI slowdown, because it's not even relevant for their business. They don't actually need to have that skill for their business. It's kind of like a shiny object, isn't it?

Noel (14:16)

Yes, very much so.

Katie (14:19)

And it's probably only going to be real AI and automation pioneers that would be like, we've not had any updates for quite some time.

Noel (14:34)

Yeah, we're going cold turkey and getting withdrawal symptoms from my Tuesday night updates.

Katie (14:41)

Yeah.

Noel (14:43)

But like I said, a lot of people, we just don't use it to its full capacity. You might think that you are, but when you really dig down and look at what this technology can and could do, then we're probably not.

Noel (15:00)

I mean, OpenAI are currently going through and solving decades old maths problems using their new models. I wouldn't even know where to start with that. But that's me not pushing it. They're pushing that model to its limits to solve those problems. Whereas I'm just going off and, for this episode, having ideas for creating titles and the description. That's just kind of easy stuff for the AI to do.

Katie (15:30)

Yeah, it's the very basics of AI, isn't it?

Noel (15:34)

Yeah, very much so.

Katie (15:36)

Okay, so with all this in mind, do you think then the prices of AI are going to go up or down?

Noel (15:49)

It's a tricky one. What we've seen at the moment is now that the infrastructure is getting more reliable and in place, especially with OpenAI, they've got loads of data centres that have not long come online, so their prices are starting to come down. But you could also see the argument where, if they're not releasing something new and shiny, are they going to end up losing some people? And then would they have to increase the price?

Noel (16:25)

So it's a tricky one to answer, but I think for now what we're seeing is a decrease in price overall. But what we would expect to see is people getting entrenched in one camp or the other. So that could impact the revenue from Anthropic and OpenAI. They're not really going to grow their user base, it's just going to be very steady, I would imagine. So we'll have to see, I guess.

Katie (16:54)

So I think really, from this conversation, although the progress of AI and automations could be slowing down, I think we really need to look inwards and think, actually, do I need to be using all of this, or are the basics just going to remain the same, and actually I'm really happy using those basics? Because I think as well, like we've said, most business owners are only using the basics and they've not even touched what AI can really do.

Noel (17:39)

Yes, exactly. But I think that's also an important point to make. Anybody that thinks, actually, I'm only using the basics — I wouldn't be worried about that, because you're still miles ahead of the majority of people anyway. Even by just doing the repetitive stuff and using AI, you're way, way ahead of most business owners.

Katie (18:07)

But even if you use Canva, Canva now has a lot of AI tools, including a background remover. So people are probably using AI without even realising, which I find quite entertaining at times.

Katie (18:30)

Because my algorithm has really changed on Facebook at the moment. It's a lot of hate for AI, which is kind of really awkward seeing as I'm a co-host of an AI podcast.

Katie (18:50)

But people don't realise that AI is creeping into all aspects of our life, whether we like it or not. It's a bit like, I always go back to it, it's a bit like when the internet came along and people were like, why do I need the internet? I've got a book.

Katie (19:10)

And then it slowly just creeps into everything. And now we can't really live day to day life without using the internet at some point, whether that's to watch something, to do some shopping, to find out something, to send an email, to actually complete our work.

Katie (19:25)

And I think almost this is what's going to be happening with AI. It creeps into places that we don't realise. A lot of online shops have got, you can try this on, and it's like, what do people think they're using to do that? And things like Canva. I think there are going to be more and more apps and websites that are going to have AI, and I think it is going to be more difficult to ignore or avoid. What do you think of that?

Noel (20:08)

Yeah, I do think it's everywhere, isn't it, now, when you look at it more deeply. You do see lots of AI features creeping into things. Even your CRM that never had AI a couple of years ago is probably now really, really pushing it. Whether or not it's any good, or does what you need it to do, is a different thing. But there are lots and lots of things that get pushed in.

Noel (20:40)

One thing with the people on your algorithm, what I would love to poll them for is all of the naysayers, and ask them how many of them either have or had an Alexa in their kitchen or in their house.

Katie (20:49)

Mm-hmm.

Noel (20:52)

I'd love to know what the percentage of that is, because I would imagine it would be quite high. They would happily talk to that, but they don't really realise what's going on, how that's training an AI model. It is learning about you, which is just what ChatGPT and Claude do. It learns.

Katie (21:10)

Or even if you've got an Apple phone, it's Siri.

Noel (21:17)

Exactly. Well, that has changed this week, hasn't it? They are allowing you to switch in OpenAI and Anthropic models, I think, into Siri. That came out with the new Duo phone. But I'd love to get AI to create an app for the Duo phone where you would turn it sideways and it'd be like using a BlackBerry. Let's go back.

Katie (21:43)

Oh gosh, I loved my BlackBerry. I mean, who didn't? I think I was having a BlackBerry moment, but I flipping loved it, it was so good. I mean, it wasn't, but it felt like it was at the time. The keypad.

Noel (22:01)

Exactly, you felt important.

Katie (22:03)

Yeah, the keypad was everything. Do you know, I think I went from a BlackBerry to an iPhone. How did I manage to do that? Having all those buttons on a BlackBerry and then, yeah, none. Or what, it would have been one, wouldn't it? One button.

Noel (22:21)

Yeah.

Katie (22:22)

Gosh.

Noel (22:24)

Crazy times.

Katie (22:26)

See, us millennials, we are just so good at adapting, honestly.

Noel (22:34)

We've seen it all.

Katie (22:36)

We have. We've experienced it all.

Noel (22:39)

Yeah.

Katie (22:42)

Well, we really hope you find this podcast episode helpful and interesting. If you've got any questions or comments, then please feel free to send us an email, hello@makeautomations.ai, or you can come and join our free LinkedIn group where Noel is in there. It's completely free to join, it's called AI Automations for Business.

Katie (23:00)

Everyone is welcome. But for now, thank you so, so much for listening, and we will catch you next time for another episode very soon.