Think Harder, Not Less: The 'AI Makes You Dumber' Myth

Mark Smith (00:12)
Well, pleased to see you're back again to the AI Advantage. It's our fourth episode, think we're doing now. So four weeks on the trot. I'm here with my co-host Meg. Hey Meg.

Meg Smith (00:22)
Hey, how's it going?

Mark Smith (00:24)
Good. It's good to have you on air again. I think that our viewership seems to be going up now that we've got a female co-host. So extremely, extremely positive on the numbers that the podcast is doing. So that is awesome. And it's awesome that you're back listening or watching us here again today.

Meg Smith (00:33)
It's my...

Mark Smith (00:50)
And really, I'd love some feedback. What do you think of the podcast so far? What can we do better? And feel free to hit us up on WhatsApp. I'll provide the link to the WhatsApp group in the show notes as always. And anything we cover, we will make sure we link in the show notes. Particularly the MIT study we're gonna touch on today and some stuff happening in the healthcare space. We'll make sure those links are available for you.

But we want to keep this show lively and really when we think about the AI advantage, it's about how our human skills, what happens when we augment them with AI, what's the impact that we can potentially have. So it's not about AI replacing us in any stretch, but it's around how could we boost ourselves. And I don't want to just use the word productivity, but how can we boost our creativity? How could we boost our our production, our output, you know, in life, how could we create new things? And, and I feel that some believe that AI is going to create everything for us, but it takes the human at the current point, it takes the human intelligence to bring the seed, into the mix to be able to grow with AI. In fact, just as I'm talking about it there, if you've ever been into gaming and you have a new landscape. built out in gaming, let's say it's a, you know, the game that comes to mind for me is age of empires and you have the landscape built and it uses a seed, which is an algorithmic number, meaning that every time you get a totally new landscape, right? But it needs that seed to actually grow the new thing from. And I think that the way we work with AI many times has got to do with us bringing the seed of our own human intellect into the equation and then mixing that with AI and creating the unbelievable possibilities in the future.

Meg Smith (02:43)
When you were talking about that, reminded me of when our son who's at university came and visited and was seeing what you'd been doing and planning our property and organizing parts of it. And he was like, ⁓ dad, I get it. You're just playing Minecraft in real life.

Mark Smith (03:01)
Yes. And that's, that's a hundred percent. Do you know what the Minecraft is? Cause that's a biome, like I was thinking biome and I'm like, do I bring them biomes there? But that concept and Minecraft of the seed, if you've ever played it, you'll understand it. That's exactly what's going on here. The ability to create whole new worlds, et cetera. and you know, by using that. And so I think that we are in an amazing time of where we can take what we know and create this. unbelievable amount of leverage and I forget some famous dude said you know give me a bar long enough and I can leverage or move anything right I could move the world you know on its filcrum type thing and I think that we have this we're in this point in time we're taking our own unique human skills and our own unique human traits and strengths We can leverage them like never before. If we lean in and go, how can we use these tools to really augment us and step us into the future possibilities of what we can do? and a part of that, I feel that we've got to look at changing some historic thought patterns. So I've had beliefs in my life that have limited me many, many times that started right back in my early education, right?

I've been told I was not so smart or dumb growing up caused me to shy away. So I can remember in high school, example, hating writing. I don't write. And I'd say, I don't write because I can't spell and ⁓ I can't read. that's, and all these can't, can't, can't, can't. And what I have found now in having AI augment what I can do. So many of those things were, I felt I couldn't do. I now am empowered to do. to, you know, where I would shy away. can bring my creative thinking and the way I smash ideas together and the insights I see some of my core strengths and I bring that to AI and I can create something new and fresh and, um, and people go, wow, that's, you know, that's inside. Like that's helpful to me. And I would not have been able to do that. I don't believe in my lifetime without the advent of AI as we know it today.

Meg Smith (05:19)
I think it's kind of the antithesis or the opposite of what we start to see in the media when you see headlines like ChatGPT is making us dumber, right? So your experience and I know mine too, as I've learned more, I guess taking that seed analogy a bit further and you see it start to grow. There's skills around using AI to extend your skills.

I think harder now than I've ever thought before in my whole entire life. It's making me, you know, really stretch my own understandings, being super motivated to learn more about certain topics. And we're going to talk a little bit in a minute about some of the ways we can use AI to research and draft and analyze as sort of specific skill or actions or things we can do. But yeah, this idea that it's making us ⠓ dumber comes from a lot of headlines. There was one based on the study from MIT called Your Brain on Chat GPT. And as Mark said, we'll put the link in the notes. I'd seen a lot of headlines or all kind of conclusions based on this. A few people, you like, raising their hands and going, aha, we knew, we knew this was bad. But if you go and actually read the page that the researchers have put up on MIT, and if you look at the study, ⠓ they admit there are massive limitations to this, right? And they even ask in a little footnote and they frequently ask questions. ⁓ don't say things like brain rot. Don't say things like chatGPT makes you dumber. that's not what they, as researchers say their, their, study show their limited size study on a particular type of person. I think it was less than 50 people of a very similar, education. background, socioeconomic even, think they touched on. So they might say that, but as we know, is it? Lies will travel the world before the truth has its chance to get his pants on or something. ⁓ So it doesn't matter now because it's in the headlines. You know, it says that and then right at the end, right at the bottom of the page, they have a video embedded from CNN where the headline is ChatGPT is rotting your brain or something like that.

Mark Smith (07:14)
Yeah, yeah.

Meg Smith (07:28)
So I look at this and I go, what's the so what? Like if I was, if I was not using AI myself, which I know a lot of people are, talked in our last episode about, you know, still the majority of people are not regularly using AI. So if I looked at that, I would be going, no, like ⁓ I don't have time for this. This is just, I was right. It's not worth my time. I'm not going to put any effort into it. And that myth really frustrates me because Who benefits from that, right? Who benefits from me staying not engaged? And it's a complex topic. I'm not pretending it's not. There's a lot of things we've got to think about in terms of ethics with AI, impact on environment. think though you have to be engaging in order to be informed on those conversations. And the first starting point is getting a bit of knowledge and then having a play. And my favorite feedback I've had from people listening to this podcast is people that I've met in person or run workshops for or met at events saying to me, I tried that. I tried that prompt and actually it was really helpful. So yeah, the kind of the myth of, we're all just dumb and AI is going to do it for us. It's not. Anyone that, well, if you have a play and maybe we can talk about some of the ways that we've tried to write with the drafting.

Mark Smith (08:36)
Yeah.

Meg Smith (08:45)
If you, can't just do it for you. It's not just going to write an essay that's good quality. You have to work really hard and actually put some thought in.

Mark Smith (08:52)
Yeah, yeah, 100%. I've been reading this book here and it's Calling Bullshit. And it's the art of skepticism in a data-driven world. And when you see articles like this, right, which are, what I was frustrated with this MIT article, one, it wasn't peer-reviewed. It was quickly put out because it was sensationalization, right, of it. And... although they put down in the footnotes, don't use these things and it's not this and it's not that, they knew that by getting the idea out there, it would have a life of its own. And this book amazingly is about examples right through history. This PhD who wrote this is how data is used to manipulate us in our thinking and that a story, doesn't matter if the story is wrong, as long as the story is constructed in the right way, It will create a life of its own and get out and run around the world in no time at all. And he gives examples up that I've seen in my life where the news has influenced me and then he unpacks it. And I tell you what, more than ever in the age that we are in, we need to have our bullshit radars on all the time and be able to validate what we're listening, what we're hearing, what we're reading, be on the headlines, right?

Right back from early days of newspapers, they know a lot of people skim the headlines. They only read the headlines. If they go a level deeper, they might read the opening sentence in the first paragraph, but the full context is, you know, they know that maybe a high percentage, and I'm not gonna give a percentage, because it would have been made up on the spot, are not actually gonna read the whole thing. And then in that example with MIT, see at the bottom that they're saying it's not that all the things that they say don't talk about it, but they know that people are not gonna get down there and they know that they're gonna use those terms and that's gonna get the story into the limelight and around the world. When I'm looking, say that again.

Meg Smith (10:51)
It's even worse now, Sorry, I was gonna say it's even worse now because we're not just skimming the headlines. We're getting 30 second reels. The whole study, the whole point of view in 30 seconds, you know, they know that's where most people are getting their news from now.

Mark Smith (11:05)
Yeah, crazy, crazy. The... You know, I mentioned in a couple of episodes ago that I believe AI is a mirror and the amount you put into it, the effort you put into it to understanding, to working with it, to a known nuance. Like one of the things I've said in the past that I'm getting into JSON prompting and using JSON a lot. Well, I just found out the other day that the data shows that JSON is one of the the less effective methods of prompting with AI in how it uses tokens and how it actually interacts with the AI model underneath them. The best way that they've found, if you're like the one that gets the best results consistently, optimization when it comes to token usage, et cetera, is actually using Markdown.

And so Markdown, the beauty of Markdown, keeps it very human legible. Everything that you do. Where I found JSON effective is when I know a prompt is going to run out in my current context window. And I've had this particularly happen with Microsoft Copilot. And then I'm going to start a new prompt to have all the data that's captured in a JSON and then use it as a starting point for my next prompt is amazing. Now under ChatGPT, they brought out a feature this last month. that allows you to do branching of an idea. So you can actually start a new chat by branching off an existing chat. And I tell you already that is saving me so many times where I've gone down a rabbit hole in a conversation and realized, no, the mistake is the algorithm went the wrong way here. And so I go back up to that point and I start a new journey out. But it has all that context. I don't have to pre-prompt with all the stuff that ⁓ went before.

Meg Smith (12:49)
Let's talk about research. ⁓ Because I have been using Microsoft Researcher a bit. So if you have the paid version of Microsoft Copilot, one of the built-in agents that you get is Researcher. And this is something I've been having a play with lately. It's been really helpful as a way to pull together a lot of sources around a particular topic that I might not have the time.

Mark Smith (12:50)
Okay.

Meg Smith (13:15)
the knowledge of those sources to pull together quite a detailed but high level overview. And I can then go into any of the sources it finds and go deeper. But the best thing about it for me in the context of my work is that it looks at my work as sources. So it looks at my emails, it looks at transcripts from recordings, it looks at documents on my SharePoint, documents on my OneDrive, as well as external sources. â “ And one of the ways that I used it recently was I was looking at doing a new type of work I hadn't done before. It was a different type of contract for a new customer of mine, And I had no context of how they typically created content, but they were working in a particular way. So I use research to say, hey, this is the company. They're looking to bring me in to do a project that they've run many times before. Tell me how they typically run a project like that. What are other companies like them doing in this space? What are the things they look for from subject matter experts? And I use researcher to give me an overview of my, know, something I would probably have only been able to go and ask somebody who had previously done that type of work before. Now I did both. I did the researcher and I did, I spoke to someone who had done that work before, but what it meant that My questions were way more informed because I'd already had basically a whole bunch of context given to me by researcher, by being able to go out and get a whole bunch of different information from both the internet and internal documents.

Mark Smith (14:49)
it's, it's often my default setting in ChatGPT is this, you know, when I think of thinking, I use the pro version of thinking and Which means it takes, you know, a long time, like 10, 15 minutes to come back with its findings when I'm in that pro-level point of research. So deep research on a topic. And I use that as a starting point for any time I'm kicking off something new that I go, hang on, I need, I'm not as knowledgeable. I know this, I know this, I know this, and I know there's these people and these resources and things. And I put that all into my prompt. And then I get this, really rich tapestry of understanding of the topic I'm going to explore. I'll give you a simple one right now. On my written card here, which is my list of things to do today, a fantastic little tool I keep on my desk, which yes, it's not digital. And what it allows me to do, is like I've got here is my children's education here and for me to explore it from an AI perspective and what the data is showing out there remember cool bullshit whenever somebody says someone says data like what a lot of the research is showing is that where education has been implemented in conjunction with AI and assisting a child to learn is that they're getting three X or three X time throughput that a kid can learn something and it's really tailored to their gaps in education and so for each of my children I will I'm doing this kind of deep research piece around based on age appropriateness what should I be teaching them to a kind of core level of competency so let's say math â “ English, et cetera, science, that type of thing. Art, you know, what should I be getting? What should I be encouraging at this point? They're going to school, they're learning the traditional school curriculum and their daycare, but they'll be going into school. And I'm not going to fight against what the school is going to teach them. But what I want to do is augment their education with what I know is the advancements that happening in this space. therefore, you know, what can I do to kind of guide and make sure that they're going to be able to maximize this new era that we're going into and future getting them into a future-ready state for whatever life you know brings across their path. I'm not just going to rely on the education system or their teachers knowledge to teach them at this point. So yeah that's what I use researcher for.

Meg Smith (17:33)
think we also talk a bit about how you can. kind of fall into a trap of thinking that the research is the output, like the research is the so what, but I loved how you kind of talk about in those examples how it's really teaching you, it's how you're learning something in a new way. And I know that tailoring the learning is such a big opportunity that our kids will have that again, going back to you as a high school, imagine if you'd had that then, you would have come out with completely different views of yourself and what you're capable of. So I love that that's something that we can do. with our kids and for our kids. And I keep coming back to this, like this check to go, it's not just, ⁓ AI will do it. Like that's also a myth, right? It's everything we've learned before and this is a new tool. So one of the biggest things, my mom's a primary school teacher and I asked her, I've always been interested in learning and development for kids. And I, before we even had kids I was like, ⁓ mom, how do you like? thinking about all the things, the daycare, the school. And I was like, tell me like, you know, is it better if they stay at home or is it better if they go to school? And she had always said consistently, the best thing for kids, the difference between kids who come to school and do really well and the ones who need to adjust. She said, we can tell the kids that have been talked to if they've had someone that's just talking to them. So often that is at daycare, right? Or it's a stay at home parent, whatever it is, it's. They're talking to the kids and reading them books and engaging with them. So as we do all of this, of course, like that continues to be a central theme, right? And I think as adults, we forget that too, as we learn about AI, we're so, you know, we have to keep practicing the talking to each other about it and kind of sounding out for bullshit together.

Mark Smith (19:22)
Yeah, I like it, like it. It can over emphasize the need for critical thinking. As always, we've had a lively conversation on the WhatsApp channel in the past week. Alex shared something Meg, what was that that was shared in the group?

Meg Smith (19:38)
Yeah. So we were chatting about AI and health and actually a couple of different people shared some of the ways they've been using AI to sort of navigate their personal health. Right. And I remember seeing, I was trying to find it now. I'll find it afterwards. I remember seeing a Ted talk many years ago by a woman who was a researcher and she took a research approach to her pregnancy. And she kind of ⠓ showed how she was able to, because she had her own health data so clearly mapped and so clearly understood, she was able to work with her healthcare providers to understand sort of the best practices or the kind of general recommendations they would make at different stages of the process, but overlay it with her own data. And her kind of takeout from that talk was general medicine is great. It's an average, right? It's a general thing. So when you're working with healthcare providers, they are going to do their part really well. But what we need to do is be experts in us, because then if they're the expert in general medicine and we're an expert in us, those two things together can give you a really good health outcome. And I think AI has been enabling that, ⁓ you know, disclaimer, people have different appetites for how much they want to share with AI. I've heard people go, you know, right the way through from happy to give full access to their health data through to I'm more on the other end where I would like try and anonymise or hide or make it not about me. You need to figure out what you're comfortable with. I think it also depends on the risk level, right? You know, if you've got something to gain, deeper insight. I personally like that it gives me a little bit of ⁓ reassurance. If I'm worried about something and I feel like the channels like my GP, can be really hard to get a GP appointment or going to after hours care, they can kind of only give you so much.

What I like is to be able to, in that particular scenario, say like, hey, this is what's happened. This is the advice I've got. What would I be looking out for if something changed and what might be a next action I might take? So that's just something I've done. But Alex shared a podcast that I think is worth going to have a look at if you're interested in AI and medicine. It's the Microsoft Research Podcast and there's researchers talking about â “ what's happening with AI and medicine. And the latest episode I really enjoyed because it was actually critiquing their own work from sort of two years ago, was 2022, where they predicted what AI would mean for medicine. And particularly in that patient healthcare provider scenario. And they're really honest about, you know, where maybe they had predicted or had hoped it would be and where we're not quite there yet. So I found it super interesting and I loved the honesty, but also I think we can also enjoy or benefit from or most appreciate how AI is augmenting medical research and enabling what, you know, was, it's really those discoveries or those advancements are happening so much faster because we used to have to have people spending thousands of hours and now through things like particularly Google's Alpha Fold comes up a lot, right? Or DeepMind's Alpha Fold comes a lot.
as an example of, we can now with the help of AI take what would have taken years in research and make it happen in this, by comparison, a very short space of time.

Mark Smith (22:58)
Love it. Join the WhatsApp group. The links are in the show notes for that. Here's a QR code that you can also use to access that. Remember, feel free to subscribe on YouTube. It's under my handle, nz365guy on YouTube, where you'll see the video version of this podcast. If you are listening to an audio, there is a video version, both on YouTube and Spotify.

If you have ideas that you would like us to explore and uncover and unpack on the show, please hit us up in the WhatsApp group with those. If you're happy for us to share it on air, we will also share them on air as part of this. record this every Monday morning, New Zealand time, with the aim of getting it published by the Wednesday of that same week.

Meg Smith (23:38)
It's because we live in the future here in New Zealand. It's usually Sunday for the rest of the world. What are we talking about next week, Mark?

Mark Smith (23:44)
Yes. Critical thinking in the intelligence age. The intelligence age, I love this concept and this is something that and I have riffed on because I didn't want to talk about the AI age because the AI age I don't think is the full story of what's going on now. The intelligence age in my mind covers things like that I think robotics is going to become a big part of our daily world.

Meg Smith (23:49)
That is my favorite topic.

Mark Smith (24:07)
â “ whether they would be, autonomous drones, whether they be, you know, autonomous driving, robotics, AI, I think it's a fusion of all these things coming together. And I feel we're on this age of intelligence. but there is a need more than ever for critical thinking because what could be considered as marketing to the masses. And it's used the word propaganda to the masses, now can be.

Meg Smith (24:20)
It'll-

Mark Smith (24:31)
targeted at individuals with an alarming level of accuracy when you bring AI into the mix. And there's this need for us to develop our spidey senses, right? To really go, is this the truth or who's wanting me to believe this? Right? Who's playing the sleight of hand? Who's playing the magic trick with me and wanting me to believe this narrative?

Meg Smith (24:50)
Yeah, and

Mark Smith (24:58)
And because AI enables it so simply, I think it's a critical skill to be learning.

Meg Smith (25:04)
Yeah, 100%. And I would go further on the intelligence age that for me, it's a reminder that it's our human intelligence, it's our emotional intelligence, it's our ancestral intelligence that we need to be continuing to grow and protect and cultivate alongside these advances in technological intelligence.

Mark Smith (25:25)
Thanks very much for joining us. Till next time. Ciao ciao.