WEBVTT - The secret to smarter AI? Make it disagree with you

0:00:00.640 --> 0:00:03.880
<v Speaker 1>AI is great at doing it what we tell it

0:00:03.960 --> 0:00:08.080
<v Speaker 1>to do, but that can also be a problem. Instead

0:00:08.119 --> 0:00:11.600
<v Speaker 1>of challenging us, it often acts like a polite, even

0:00:11.680 --> 0:00:17.200
<v Speaker 1>sycophantic assistant, doing exactly as asked without pointing out flaws

0:00:17.360 --> 0:00:22.759
<v Speaker 1>or blind spots. Today, I'm joined by Inventium's AI expert,

0:00:22.800 --> 0:00:26.600
<v Speaker 1>Neo Applin, to explore how we can use AI differently

0:00:27.280 --> 0:00:30.680
<v Speaker 1>as a devil's advocate, a creative partner, and even as

0:00:30.720 --> 0:00:34.279
<v Speaker 1>a stand in audience to help spot weaknesses in our

0:00:34.360 --> 0:00:38.919
<v Speaker 1>thinking and uncover new perspectives. In this chat, you are

0:00:38.960 --> 0:00:42.200
<v Speaker 1>going to learn some very practical prompts to make AI

0:00:42.720 --> 0:00:47.560
<v Speaker 1>your best and most useful collaborator, not just your obedient assistant.

0:00:52.440 --> 0:00:56.120
<v Speaker 1>Welcome to How I Work, a show about habits, rituals,

0:00:56.200 --> 0:00:59.960
<v Speaker 1>and strategies for optimizing your day. I'm your host doctor

0:01:00.080 --> 0:01:04.600
<v Speaker 1>to Amantha Imba. Two years ago, I completely overhauled how

0:01:04.640 --> 0:01:08.160
<v Speaker 1>I work with Genai, and I am now saving over

0:01:08.280 --> 0:01:13.760
<v Speaker 1>forty hours every single week. That is no exaggeration, and

0:01:13.840 --> 0:01:17.520
<v Speaker 1>that is exactly why my company, Inventium, created the Genai

0:01:17.680 --> 0:01:21.000
<v Speaker 1>Productivity Upgrade. It's a twelve week course designed to move

0:01:21.040 --> 0:01:26.640
<v Speaker 1>you from Jenai dabbler to productivity machine. There is no fluff,

0:01:26.800 --> 0:01:29.640
<v Speaker 1>just practical strategies that will pay off from week one,

0:01:30.040 --> 0:01:34.520
<v Speaker 1>saving you at least ten hours every single week. We

0:01:34.560 --> 0:01:36.840
<v Speaker 1>are going to teach you how to automate the grant work,

0:01:37.080 --> 0:01:39.880
<v Speaker 1>use AI as your second brain to excel at your job.

0:01:40.360 --> 0:01:43.040
<v Speaker 1>Learn how to create AI agents that we knew back

0:01:43.200 --> 0:01:46.520
<v Speaker 1>hours every time you use them, and so much more so.

0:01:46.600 --> 0:01:49.960
<v Speaker 1>Whether you're a complete beginner or already using AI, we

0:01:50.120 --> 0:01:53.520
<v Speaker 1>have got you covered. We start with prompting fundamentals and

0:01:53.560 --> 0:01:56.080
<v Speaker 1>we go all the way through to advanced automations and

0:01:56.120 --> 0:01:59.920
<v Speaker 1>agentic AI. We kick off on October fifteen and spots

0:02:00.080 --> 0:02:04.160
<v Speaker 1>are limited. Visit inventtium dot com, dot au, forward slash

0:02:04.280 --> 0:02:08.280
<v Speaker 1>genai hyphen cohort to secure your place now And yes

0:02:08.320 --> 0:02:10.600
<v Speaker 1>that's a long link. It's also in the show notes.

0:02:10.600 --> 0:02:12.960
<v Speaker 1>If you're looking for it and you've literally got nothing

0:02:13.000 --> 0:02:15.919
<v Speaker 1>to lose, there's a seven day money back guarantee. So

0:02:16.240 --> 0:02:18.120
<v Speaker 1>go to the show notes, hop on the link and

0:02:18.240 --> 0:02:22.639
<v Speaker 1>join our program today. So why do we need AI

0:02:23.080 --> 0:02:26.919
<v Speaker 1>to help us point out our blind spots? Like, what's

0:02:26.960 --> 0:02:28.239
<v Speaker 1>the problem that it's solving here?

0:02:28.800 --> 0:02:32.400
<v Speaker 2>We as humans, we're lazy. We were supposed to be

0:02:32.440 --> 0:02:35.400
<v Speaker 2>cognitively lazy. What that means is if we saw everything

0:02:35.400 --> 0:02:37.200
<v Speaker 2>and we did everything that we needed to do, we

0:02:37.240 --> 0:02:39.640
<v Speaker 2>would run out of time and energy. So what we

0:02:39.680 --> 0:02:43.359
<v Speaker 2>do is we fall into patterns, and those patterns might

0:02:43.360 --> 0:02:45.640
<v Speaker 2>be we do the same thing every single day, the

0:02:45.720 --> 0:02:49.320
<v Speaker 2>same way, and that's fine for some things. But when

0:02:49.360 --> 0:02:52.040
<v Speaker 2>you're doing things like you know, strategy or reports or

0:02:52.680 --> 0:02:56.240
<v Speaker 2>things that are different than having a different perspective and

0:02:56.280 --> 0:02:59.760
<v Speaker 2>being challenged on those things is really valuable. Because AI,

0:03:00.160 --> 0:03:03.239
<v Speaker 2>believe it or not, it's been built as a helpful assistant,

0:03:03.560 --> 0:03:06.240
<v Speaker 2>so for some reason, it always then becomes a little

0:03:06.240 --> 0:03:10.040
<v Speaker 2>bit like a cheerleader, sometimes a bit sycophantic, but in

0:03:10.080 --> 0:03:13.800
<v Speaker 2>reality it'll go along with any pattern you want. So

0:03:14.000 --> 0:03:16.480
<v Speaker 2>getting AI to instead of going along with those patterns,

0:03:16.560 --> 0:03:19.240
<v Speaker 2>to actually challenge you and ask questions, it becomes an

0:03:19.280 --> 0:03:22.359
<v Speaker 2>awesome creativity tool and helps you get out of those

0:03:22.480 --> 0:03:23.919
<v Speaker 2>ruts a little bit unstuck.

0:03:24.520 --> 0:03:29.280
<v Speaker 1>So how do we turn AI from being this sycophantic,

0:03:29.480 --> 0:03:34.800
<v Speaker 1>obedient assistant into a collaborative kind of sparring partner.

0:03:35.600 --> 0:03:39.840
<v Speaker 2>You tell it too, believe it or not, So there's

0:03:39.840 --> 0:03:41.960
<v Speaker 2>different ways you can do it. By one, you can

0:03:42.040 --> 0:03:44.640
<v Speaker 2>just say, hey, I want you to challenge me in

0:03:44.760 --> 0:03:48.280
<v Speaker 2>the ways that I'm thinking things like that. I love

0:03:48.480 --> 0:03:51.320
<v Speaker 2>using it as a devil's advocate. So it might be

0:03:51.360 --> 0:03:53.920
<v Speaker 2>for a particular piece of work. You might say, Hey,

0:03:53.960 --> 0:03:57.000
<v Speaker 2>I'm doing this. Here's the work. I want you to

0:03:57.040 --> 0:04:01.000
<v Speaker 2>review it from this perspective. What ways can it be improved?

0:04:01.480 --> 0:04:03.560
<v Speaker 2>What am I missing? Are there any things that I've

0:04:03.600 --> 0:04:07.440
<v Speaker 2>gotten wrong here? Getting it to really review and critique

0:04:07.440 --> 0:04:11.000
<v Speaker 2>what you're doing and provide suggestions is an excellent, excellent

0:04:11.040 --> 0:04:13.560
<v Speaker 2>way to do it. Also, other ways just might be simple,

0:04:13.560 --> 0:04:15.240
<v Speaker 2>which is is there a better way to do this?

0:04:15.800 --> 0:04:18.080
<v Speaker 2>Or Hey, for this audience, what would they want to

0:04:18.120 --> 0:04:20.280
<v Speaker 2>see that's not there? Are there other ways I can

0:04:20.320 --> 0:04:24.279
<v Speaker 2>approach the task. You can ask these questions. Now, it

0:04:24.320 --> 0:04:26.640
<v Speaker 2>doesn't always come up with gold every time because it

0:04:26.680 --> 0:04:29.279
<v Speaker 2>doesn't know all the background and all those kind of things,

0:04:29.400 --> 0:04:31.920
<v Speaker 2>but it does have different perspectives and it's up to

0:04:31.960 --> 0:04:33.599
<v Speaker 2>you to go, Wow, that's a really good one. I

0:04:33.640 --> 0:04:36.160
<v Speaker 2>didn't think of that, and then you can incorporate that

0:04:36.360 --> 0:04:37.039
<v Speaker 2>in your work.

0:04:37.640 --> 0:04:40.239
<v Speaker 1>Yeah, I might say that's something I love about AI

0:04:40.400 --> 0:04:43.880
<v Speaker 1>that it can help me see things from completely different

0:04:44.040 --> 0:04:48.200
<v Speaker 1>perspectives and you know, put myself in people's shoes that

0:04:48.600 --> 0:04:53.480
<v Speaker 1>I would you know very little about. So what tips

0:04:53.480 --> 0:04:56.320
<v Speaker 1>do you have on how we can get the AI

0:04:56.880 --> 0:05:00.320
<v Speaker 1>to help us really easily see things from are the

0:05:00.640 --> 0:05:02.840
<v Speaker 1>people's customers perspectives.

0:05:03.680 --> 0:05:08.160
<v Speaker 2>We do things for people always with clients or customers

0:05:08.440 --> 0:05:12.520
<v Speaker 2>or staff members or other team members. So you can

0:05:12.520 --> 0:05:14.760
<v Speaker 2>get AI to step in the shoes of these people

0:05:15.000 --> 0:05:17.520
<v Speaker 2>just by telling AI who those people are and what

0:05:17.560 --> 0:05:20.160
<v Speaker 2>their perspectives are. So it might be a customer, it

0:05:20.240 --> 0:05:23.120
<v Speaker 2>might be a client, might be a boss, it might

0:05:23.160 --> 0:05:27.120
<v Speaker 2>be a prospective persona that you might have, so like

0:05:27.640 --> 0:05:31.359
<v Speaker 2>a busy person with these particular needs, a downstream user

0:05:31.440 --> 0:05:34.640
<v Speaker 2>in the process. So I want you to say to AI,

0:05:35.000 --> 0:05:37.960
<v Speaker 2>I want you to be in the shoes of this person.

0:05:38.720 --> 0:05:41.960
<v Speaker 2>Want to look at it from their perspective? How would

0:05:42.000 --> 0:05:44.040
<v Speaker 2>they look at this, What would they want to see

0:05:44.080 --> 0:05:47.120
<v Speaker 2>that's not there? Which bits would they get worried about,

0:05:47.360 --> 0:05:49.960
<v Speaker 2>which questions would they have, those kind of things, But

0:05:50.240 --> 0:05:53.039
<v Speaker 2>really be clear about the kind of person and the perspective,

0:05:53.120 --> 0:05:56.040
<v Speaker 2>and AI can really jump in those shoes very effectively.

0:05:56.760 --> 0:06:00.720
<v Speaker 1>Now, a lot of listeners would be giving presentations as

0:06:00.760 --> 0:06:03.360
<v Speaker 1>part of their work and trying to come up with

0:06:03.800 --> 0:06:06.680
<v Speaker 1>new ideas whether that be products or services or all

0:06:06.720 --> 0:06:11.320
<v Speaker 1>sorts of things. How can AI help test these things,

0:06:11.320 --> 0:06:14.560
<v Speaker 1>whether it be a presentation that's pretty high stakes, or

0:06:14.600 --> 0:06:17.000
<v Speaker 1>maybe some kind of an idea that you're thinking about

0:06:17.040 --> 0:06:17.839
<v Speaker 1>for your customers.

0:06:18.960 --> 0:06:22.000
<v Speaker 2>First, I got to say that AI can't actually be

0:06:22.120 --> 0:06:26.240
<v Speaker 2>that person, can't actually be that audience member, but it

0:06:26.240 --> 0:06:28.320
<v Speaker 2>can be a clue. It can be a hint as

0:06:28.360 --> 0:06:30.960
<v Speaker 2>to what that person might think. So think of this

0:06:31.000 --> 0:06:35.200
<v Speaker 2>as like a pre test or a sparring partner as

0:06:35.240 --> 0:06:39.320
<v Speaker 2>opposed to the final boss. But you can give it

0:06:39.320 --> 0:06:42.400
<v Speaker 2>that perspective. You say, here's the audience, here's a perspective

0:06:42.640 --> 0:06:45.839
<v Speaker 2>they'll probably have, and then ask it where are the

0:06:45.839 --> 0:06:48.279
<v Speaker 2>bits that they might find too risky, Where are the

0:06:48.320 --> 0:06:51.560
<v Speaker 2>bits that in my presentation they might find boring or

0:06:52.240 --> 0:06:54.359
<v Speaker 2>they might lose interest, or where are the bits that

0:06:54.400 --> 0:06:57.560
<v Speaker 2>they might find surprising and they may want more information

0:06:57.760 --> 0:07:01.040
<v Speaker 2>those kind of things. Also, you can ask then ask

0:07:01.160 --> 0:07:04.480
<v Speaker 2>AI what the kind of perspectives are of these people,

0:07:04.480 --> 0:07:06.839
<v Speaker 2>What are the jobs that they are trying to achieve,

0:07:07.200 --> 0:07:09.159
<v Speaker 2>what are their needs, what are their desires, what are

0:07:09.200 --> 0:07:11.520
<v Speaker 2>their goals? And then you can figure out whether your

0:07:11.560 --> 0:07:14.320
<v Speaker 2>work will fit within those as well. So it's really

0:07:14.360 --> 0:07:18.480
<v Speaker 2>like an empathy tool from a robot, which sounds ridiculous,

0:07:18.760 --> 0:07:21.200
<v Speaker 2>but it's great as a first step before you then

0:07:21.280 --> 0:07:24.160
<v Speaker 2>go to the actual human out there, so you can

0:07:24.160 --> 0:07:27.600
<v Speaker 2>then potentially something have something which lands better for you

0:07:27.720 --> 0:07:28.240
<v Speaker 2>and for them.

0:07:29.280 --> 0:07:31.400
<v Speaker 1>Yeah, I think it's really interesting in the context of

0:07:31.920 --> 0:07:35.520
<v Speaker 1>innovation work, which obviously Inventium does a lot of, and

0:07:35.680 --> 0:07:40.160
<v Speaker 1>previously to really get any idea or any kind of

0:07:40.160 --> 0:07:44.080
<v Speaker 1>feedback around an idea of what customers think, you'd have

0:07:44.160 --> 0:07:47.960
<v Speaker 1>to spend quite quite a bit of effort finding customers,

0:07:48.240 --> 0:07:52.040
<v Speaker 1>talking to them and so on. And what I really

0:07:52.160 --> 0:07:56.840
<v Speaker 1>like about AI is you can almost get that first pass. Like, yes,

0:07:56.880 --> 0:08:00.800
<v Speaker 1>as you say, it doesn't replace human feedback, but it

0:08:00.920 --> 0:08:05.280
<v Speaker 1>is better than just going blindly into creating a business

0:08:05.360 --> 0:08:09.440
<v Speaker 1>case where you haven't really put yourself in the customer's shoes.

0:08:10.640 --> 0:08:13.480
<v Speaker 1>What like, do you have some go to prompts that

0:08:13.560 --> 0:08:19.640
<v Speaker 1>you use when you're trying to get AI to adopt

0:08:19.680 --> 0:08:25.480
<v Speaker 1>a certain persona or customer profile, Like how do you

0:08:25.480 --> 0:08:26.480
<v Speaker 1>typically approach that?

0:08:27.480 --> 0:08:29.400
<v Speaker 2>It depends on what you're doing, of course, but it's

0:08:29.440 --> 0:08:33.080
<v Speaker 2>about giving AI the brief so this is my customer,

0:08:33.160 --> 0:08:35.160
<v Speaker 2>these are the kind of needs that they might have

0:08:36.040 --> 0:08:38.880
<v Speaker 2>if they've got a particular role. So I'm doing a

0:08:38.920 --> 0:08:42.000
<v Speaker 2>presentation to the senior executive team. Don't just say senior

0:08:42.080 --> 0:08:46.000
<v Speaker 2>executive team. Say to AI, who's on that team? Say

0:08:46.040 --> 0:08:49.200
<v Speaker 2>to AI all the things you know about those people

0:08:49.280 --> 0:08:52.000
<v Speaker 2>and what their jobs are and what they're trying to achieve.

0:08:52.400 --> 0:08:54.439
<v Speaker 2>If you even add in things like I don't know

0:08:54.559 --> 0:08:57.240
<v Speaker 2>strategies and things like that for the senior executive team,

0:08:57.400 --> 0:09:01.120
<v Speaker 2>then we'll know that as well, but also stomers and clients.

0:09:01.280 --> 0:09:03.880
<v Speaker 2>So you can say, these are the types of customers

0:09:03.880 --> 0:09:06.720
<v Speaker 2>that we've got, Here's what they've done in the past,

0:09:06.840 --> 0:09:08.880
<v Speaker 2>these are the kind of problems they've had in the past.

0:09:09.679 --> 0:09:13.160
<v Speaker 2>More information you can give AI the better. But even

0:09:13.200 --> 0:09:16.520
<v Speaker 2>with that, AI can still come up with some other

0:09:16.640 --> 0:09:18.880
<v Speaker 2>ideas for you, which is great. So what you could

0:09:18.920 --> 0:09:22.079
<v Speaker 2>say is I know very little about these people AI,

0:09:22.320 --> 0:09:24.160
<v Speaker 2>what are the things that they would want? What are

0:09:24.200 --> 0:09:27.680
<v Speaker 2>their drivers? We are their goals and get AI to effectively,

0:09:27.720 --> 0:09:30.520
<v Speaker 2>you know, put itself in those shoes and it can

0:09:30.559 --> 0:09:33.400
<v Speaker 2>come up with some amazing things. So instead of being

0:09:33.920 --> 0:09:38.600
<v Speaker 2>cold and not being prepared here you're pretty warm and warm,

0:09:38.920 --> 0:09:41.560
<v Speaker 2>warm to medium warm and hot before you even get

0:09:41.559 --> 0:09:44.800
<v Speaker 2>in front of their customer because you've been practicing with AI,

0:09:45.040 --> 0:09:48.720
<v Speaker 2>and AI is effectively giving you different perspectives from different

0:09:48.720 --> 0:09:53.600
<v Speaker 2>people before you get to that final meeting, that final presentation,

0:09:53.920 --> 0:09:55.160
<v Speaker 2>that final sales pitch.

0:09:56.000 --> 0:09:59.559
<v Speaker 1>Okay, so a lot of different ways to use AI

0:09:59.760 --> 0:10:04.199
<v Speaker 1>to help us point out our blind spots. If a

0:10:04.240 --> 0:10:07.559
<v Speaker 1>listener was going to just do one thing differently today

0:10:07.840 --> 0:10:10.679
<v Speaker 1>in how they work with AI, what would you recommend

0:10:10.720 --> 0:10:11.080
<v Speaker 1>that be?

0:10:12.480 --> 0:10:15.080
<v Speaker 2>First off, don't let AI be you me yes man,

0:10:15.520 --> 0:10:19.679
<v Speaker 2>so tell it work with me, be a critical, inspiring partner,

0:10:19.840 --> 0:10:23.160
<v Speaker 2>help me improve my things, and then get it to

0:10:23.200 --> 0:10:26.000
<v Speaker 2>help you to review your things, just as if it

0:10:26.120 --> 0:10:30.320
<v Speaker 2>was your customer, your client, your audience, and you'll be

0:10:30.400 --> 0:10:33.240
<v Speaker 2>surprised of what it comes up with. Now, don't accept

0:10:33.320 --> 0:10:36.040
<v Speaker 2>every idea that it has, don't accept every critique that

0:10:36.120 --> 0:10:38.840
<v Speaker 2>it has because it may not be appropriate for you,

0:10:39.360 --> 0:10:42.040
<v Speaker 2>but you might find that there's some amazing things in there.

0:10:42.040 --> 0:10:44.400
<v Speaker 2>If you do want to incorporate with your work, you do.

0:10:44.559 --> 0:10:46.400
<v Speaker 2>So don't let it be your yes man.

0:10:47.120 --> 0:10:47.560
<v Speaker 1>Let it.

0:10:49.200 --> 0:10:51.719
<v Speaker 2>Know what your clients and your audience are, and then

0:10:51.840 --> 0:10:54.280
<v Speaker 2>take some of those pieces of advice so that you

0:10:54.280 --> 0:10:55.720
<v Speaker 2>can actually improve your work.

0:10:56.160 --> 0:11:00.800
<v Speaker 1>Awesome, Thank you so much for sharing such great advice. Again, Neo,

0:11:01.320 --> 0:11:05.600
<v Speaker 1>thank you. I hope you are now feeling inspired to

0:11:06.160 --> 0:11:09.200
<v Speaker 1>use your AI in a different way today and turn

0:11:09.280 --> 0:11:13.520
<v Speaker 1>it from a sycophantic assistant into a true thinking partner

0:11:13.600 --> 0:11:16.600
<v Speaker 1>that actually critiques your thinking rather than just agrees with

0:11:16.640 --> 0:11:20.080
<v Speaker 1>everything that you say. If you are looking to take

0:11:20.120 --> 0:11:23.240
<v Speaker 1>your AI skills up to the next level, Inventium's GENAI

0:11:23.360 --> 0:11:27.199
<v Speaker 1>productivity upgrade is kicking off on October fifteen, and there

0:11:27.280 --> 0:11:29.840
<v Speaker 1>is a link to more info and bookings in the

0:11:29.880 --> 0:11:33.960
<v Speaker 1>show notes. If you like today's show, make sure you

0:11:34.080 --> 0:11:37.080
<v Speaker 1>hit follow on your podcast app to be alerted when

0:11:37.160 --> 0:11:40.439
<v Speaker 1>new episodes drop. How I Work was recorded on the

0:11:40.480 --> 0:11:43.240
<v Speaker 1>traditional land of the Warrengery people, part of the Cool

0:11:43.280 --> 0:11:45.800
<v Speaker 1>and Nation. A big thank you to Martin Nimber for

0:11:45.880 --> 0:11:46.880
<v Speaker 1>doing the sound mix.