Australians trust AI advice over friends and family | The Morning Show

[00:00:00]Well, artificial intelligence is no longer just answering our questions.

[00:00:03]It's becoming our advisor. New research reveals 47% of Australians trust AI-generated advice more than that given from friends or family, while many are using it as a coach.

[00:00:15]>> From meal planning to major life decisions, AI is fast becoming our new everyday confidant. So, why are we so quick to believe the bots? And are humans trusting the technology a little bit too much? Change strategist and trend forecaster Michael McQueen joins us live in the studio. Michael, thanks for being Good morning.

[00:00:32]>> Now, the stats are pretty staggering.

[00:00:34]Why are we turning to chatbots over people so often these days?

[00:00:38]>> a lot of reasons. It's convenient.

[00:00:39]They're always there. They're always accessible. There's also no shame, no judgment. You know, there's that that funny like the meme often going around like people take a photo and ask ChatGPT, "What is this rash?" You know, that's that's easier than asking a human. But, the stakes are even higher in some cases. I read a great example recently where in Melbourne there's a homeless support agency that have developed an AI bot. And when they surveyed people who use their services, 84% said they'd rather speak to the bot than a human to figure out what sort of services are available if they're at risk of homelessness cuz there's often shame and embarrassment about that. So, there's that sense of the non-judgy space is a lot of the reasons where we're going to these AI bots.

[00:01:15]>> Are you Are you surprised? Does your head spin about how quickly we're just now all latching on? I mean, you've been You've been across this for a long time.

[00:01:22]But, for just the general population.

[00:01:24]>> 100%. Yeah. How quickly we've just uh assumed that they're like humans.

[00:01:28]>> Yeah.

[00:01:29]>> People have got a name for their AI bot and they they've got to assign a gender to it. Like, they it's so familiar now in a way that's unthinkable 2 years ago.

[00:01:36]>> more likely to believe it because it sounds so confident and it gives us so many options?

[00:01:40]>> 100%. So, confidence is the key because humans tend to associate confidence with competence. And even you think about the number of people on this couch over the years. If you say something with clarity and conviction and with great confidence, everyone just assumes you must know what you're talking about.

[00:01:52]>> can still see through that.

[00:01:53]>> [laughter] >> You've got discernment. But, it is tricky. Like, our brains often go to that. So and AI is always absolutely sure, super confident. It's never like I'm not sure, I don't know. So there's that. We perceive it as being objective as well which is sort of tricky. There's often this sense too of what they call trust spillover. This idea that if if it's gotten one thing right, we'll trust it more next time. There's consistency that counts. So these are all the things that make our brains sort of lean toward trusting it.

[00:02:18]>> Yeah, even when it is quite seriously wrong.

[00:02:20]>> Correct.

[00:02:20]>> You know, in what situations should we be embracing AI and you know, maybe some situations where it really could get us into trouble.

[00:02:28]>> Yeah, there's some good things. I mean it's great for travel planning, great for meal planning, meal ideas. It's a really good brainstorming partner for instance. Some of the things I would steer clear of would be things like legal advice, financial advice, medical advice. It's good for some medical things but other things not so much. In fact, I was chatting with the head of innovation at Zero, the big AI the big tech firm for the accounting sector. And he put it well. He was talking about accounting, you know, getting financial advice from AI. He said this, he said, you know, the reality is AI can tell you what you could do financially. It won't tell you what you should do. That requires human judgment. So we need to use AI cuz it is sensible, it's powerful but use it within within sensible boundaries.

[00:03:05]>> Can you use it as a jumping off point?

[00:03:07]>> Yes. Yeah, I think that's a very smart place. The question then is where do you jump off? Like what's the point at which you go, okay, I need to trust it to here >> Yeah.

[00:03:14]>> but not trust it beyond that cuz you know, I think as as humans often we we even if we know the information's wrong or incomplete, you know, there's this interesting thing in psychology where our brains are allergic to ambiguity. We don't like not knowing. And I I read this great quote recently that if we're lost in the Swiss Alps, most of us would rather have a map of the Pyrenees than no map at all.

[00:03:32]>> Yeah.

[00:03:32]>> This idea that even if we know it's not right, it's like we just want to hold on to something.

[00:03:36]>> Something.

[00:03:36]>> And so AI can often be that crutch.

[00:03:38]>> you should question what are the con If I if AI gets this wrong for me, what are the consequences? I used it yesterday for a recipe. I was in the supermarket, I couldn't find the recipe, I asked it.

[00:03:47]I've got mince and I've got this, what should I do? I trusted it 100%. I then asked it for a travel and airfare and I kind of used it as a bit of a okay, now I'll go and investigate that myself.

[00:03:57]>> That's smart. Cuz you know what it's like, the confidence thing is interesting cuz you'll ask it and it'll give you a confident answer and sometimes you go, "That's not true." And you actually say that and go, "Oh, you're correct. Yes, that wasn't true.

[00:04:06]This is now true."

[00:04:07]>> But we know that as people of a certain age, 19-year-olds may not necessarily have that discernment. So what's what do you advise in sort of younger people?

[00:04:16]>> Well, I think teaching those skills of critical thinking and curiosity is vital, especially for teenage teenagers cuz a lot of them there is that sense that again, if it seems confident and it seems to have all the answers, we trust it and I think we're over trusting in many cases right now.

[00:04:30]>> We have a a graphic here. Take us through that.

[00:04:32]>> And so yeah, these are again some of the things that you might want to trust it for, but I think there are certainly boundaries. Explaining complex topics is interesting. I use a thing called Notebook LM a lot and Notebook LM is great cuz I can put a large document into it and actually create a podcast episode 20 minutes long of an AI bot explaining that really large complex document to me. Um and so it's really good for synthesizing complex ideas, but then don't trust it entirely cuz it can still hallucinate more than I think many of us realize.

[00:05:00]>> So looking ahead to the future, >> Yeah.

[00:05:02]>> do you think we're in on track for AI to become a trusted confidant, a trusted advisor, or are we potentially heading down the Terminator type territory?

[00:05:12]>> Well, there's often that that that fear and I think it's a reasonable fear. In fact, even the Prime Minister's announcement yesterday about having guardrails and regulation is sensible for this cuz it is a confidant. In fact, I was reading a report last week, 62% of teenagers who use AI treat it as a friend. It's a companion, which is sort of understandable but a bit. I mean, even your face speaks volumes, Carly.

[00:05:31]It's a bit uncomfortable. It's like, really, is that a good thing? Um but from a business perspective, it's massive. The big shift I'm seeing for corporates right now, we've gone from SEO, which is search engine optimization, you get to the top of the Google search, to a thing called AEO, which is when someone's using AI to do research, how do you show up in an AI response on chat GPT or on co-pilot?

[00:05:52]This is like the way we search and do research is thoroughly shifting because of this.

[00:05:56]>> Wow.

[00:05:57]>> And so businesses are rapidly trying to figure out what what does marketing now look like?

[00:06:00]>> Can I ask you one quick question? You mentioned something in one of the programs, the AI helpers that you use.

[00:06:06]Um, all these come with a cost though, don't they?

[00:06:09]>> Some of them.

[00:06:10]>> So is that where also we have to be just a bit careful?

[00:06:13]>> You do, yes.

[00:06:14]>> are free, some you have to pay for.

[00:06:16]>> Yes.

[00:06:17]>> And and much like we've been collecting TV subscriptions that we probably don't need because [laughter] free to air is fantastic. Um, yes. Um, do you have to be wary of that that you're not collecting like how much is an expensive AI AI subscription?

[00:06:30]>> Yes, so if you're spending 20 to 25 bucks a month for some most of these apps, that's good. Choose one and just use that one I would say to begin with.

[00:06:37]>> Claude.

[00:06:37]>> Claude is very good, but sometimes the cost of that can be more cuz often you're paying per token. Um, but we code with chat GPT, you need to have like a not a base version I would say which is the free version largely because they're now selling ads. That influences what you're getting back when you do a search on chat GPT. Yes, they are. So we've got a profit motive there which we've got to be mindful of.

[00:06:59]>> That's interesting.

[00:07:00]>> Correct. Yeah, but co-pilot which is the Microsoft one, if you've got a Microsoft account that you're paying an annual subscription, you're getting that for free anyway.

[00:07:07]>> Okay, maybe we should look into that too cuz I'm a little bit unsure about how that works. Thank you, Michael.

[00:07:11]>> You're welcome. Thank you so much, Michael.