Latest / How I Work / Still arguing with AI over a bad first draft? Try these 4 tips
Transcript
- 0:00Speaker 1: Do you still have one of those prompt formulas saved somewhere?
- 0:05Speaker 1: Maybe some acronym, an Instagram reel, swore would get you
- 0:10Speaker 1: better answers from AI. And maybe you still paste it in,
- 0:14Speaker 1: swap a few words, hit send, and what comes back
- 0:18Speaker 1: still isn't what you wanted. Part of the trouble is
- 0:22Speaker 1: that most of us still talk to AI the way
- 0:25Speaker 1: Google trained us to. A few keywords shouted into a box. Today,
- 0:31Speaker 1: Neo and I look at why those formulas caught on
- 0:35Speaker 1: and what has changed since, including how memory now fills
- 0:40Speaker 1: in some of the gaps for you. And today, you're
- 0:43Speaker 1: going to learn how to brief AI properly, what to
- 0:47Speaker 1: ask it so it shows you better options, and how
- 0:50Speaker 1: to get it to challenge you right down to whether
- 0:53Speaker 1: you should be making the thing at all. By the
- 0:57Speaker 1: end of today's show, you're going to have a simpler
- 1:00Speaker 1: way of talking to AI that gets you better work
- 1:03Speaker 1: back with far less back and forth.
- 1:12Speaker 2: Welcome to How I AI with me, Dr. Amantha Imber
- 1:15Speaker 2: and Neo Applin, head of Inventium AI.
- 1:19Speaker 1: Each episode, we share one practical way to use AI
- 1:22Speaker 1: better at work and in life.
- 1:25Speaker 2: No fluff, no tech jargon, just things you can use
- 1:28Speaker 2: straight away.
- 1:29Speaker 1: Prompt formulas used to be really, really big in AI
- 1:34Speaker 1: in like 2023, 2024, maybe marginally in 2025, where, you know,
- 1:41Speaker 1: there were all these fancy acronyms that we had to
- 1:45Speaker 1: remember that the Instagram influencers were telling us in reels
- 1:50Speaker 1: to apparently get better output from AI.
- 1:53Speaker 2: So, I mean, why did we have, like, why was
- 1:57Speaker 2: there such a focus on prompt formulas, Neo?
- 2:01Speaker 3: They were cool. Everyone had one and I wanted to
- 2:04Speaker 3: collect them. No, I didn't want to collect them all.
- 2:05Speaker 3: I found it very difficult to remember. Did you? It
- 2:08Speaker 3: was a prep, edit or goal. I can't remember all
- 2:11Speaker 3: those different things that they had for these acronyms. Yeah,
- 2:16Speaker 3: the reason was first off, these were pretty early models.
- 2:20Speaker 3: They were pretty good. Like if you look back to
- 2:22Speaker 3: AI a couple of years back, they were pretty good,
- 2:24Speaker 3: but they were not as good as what we have today.
- 2:27Speaker 3: And so people were trying to find a way to
- 2:29Speaker 3: get the model to actually work with them in a
- 2:31Speaker 3: way which was a little bit more reliable and actually
- 2:33Speaker 3: would actually be better. And so each of the companies
- 2:37Speaker 3: and different people had different ways to be able to
- 2:40Speaker 3: tell AI what we wanted. But it was a flawed
- 2:44Speaker 3: perspective in a couple of ways. One was what we
- 2:48Speaker 3: were trying to do was send it one command, a
- 2:53Speaker 3: magical command, this big long prompt that had all the
- 2:56Speaker 3: words in it and all the things. And then it
- 2:59Speaker 3: would be like an abracadabra. It would then open up
- 3:02Speaker 3: the thing and then suddenly you'd get a great response
- 3:04Speaker 3: from AI. And maybe that was the case for the day.
- 3:09Speaker 3: And I don't think it was really... true, but I
- 3:12Speaker 3: think a lot of people felt a lot of comfort
- 3:14Speaker 3: in that I've got this magical thing called AI. I'm
- 3:16Speaker 3: not really quite sure how to talk to it, but
- 3:18Speaker 3: here I've got a formula. I'll just grab the formula,
- 3:20Speaker 3: change a couple of things and then send it and
- 3:22Speaker 3: then it will be great. And that's where the kind
- 3:25Speaker 3: of reliance came from. And it does make sense. We're
- 3:27Speaker 3: all getting used to it and people wanted to find
- 3:29Speaker 3: their way through it.
- 3:31Speaker 1: So what has changed so much that really these prompt
- 3:35Speaker 1: formulas are really don't matter in the way that they
- 3:39Speaker 1: did previously. So what are the big changes that have happened?
- 3:43Speaker 3: The first is that these models have gotten better. So
- 3:46Speaker 3: the AI companies, Anthropic and OpenAI and all the others,
- 3:51Speaker 3: they've done a better job with these models. And the
- 3:53Speaker 3: models are better at understanding what you really mean, not
- 3:57Speaker 3: what you say. The problems that we as humans have
- 4:00Speaker 3: are we're really lazy with our language. And so AI
- 4:04Speaker 3: can But earlier, it was very literal. And so it
- 4:07Speaker 3: would go, oh, you want this? And then you go, no,
- 4:09Speaker 3: it's not really what I meant. It's what I said, sure,
- 4:12Speaker 3: but it's not what I meant. Now, AI has a
- 4:14Speaker 3: much better understanding of what good looks like. Sometimes it
- 4:18Speaker 3: has a better idea of what good looks like than
- 4:20Speaker 3: what we do. And so then it can kind of
- 4:23Speaker 3: read between the lines. And there is really the big
- 4:26Speaker 3: benefit of it. So rather than us having to teach
- 4:29Speaker 3: AI what good is, because they've elevated their models, they've
- 4:33Speaker 3: something a lot smarter than they used to be, AI,
- 4:36Speaker 3: and here's actually probably a problem, is AI thinks it
- 4:38Speaker 3: probably knows better than you do in a lot of ways.
- 4:40Speaker 3: And so we don't need to use magical phrases or
- 4:44Speaker 3: cut and paste major prompt libraries to be able to
- 4:47Speaker 3: get a pretty good output out of it.
- 4:50Speaker 1: So Also, memory, I believe, is something that has changed
- 4:56Speaker 1: as well that impacts on how we have a conversation
- 4:59Speaker 1: with the AI in a way that gets us really
- 5:02Speaker 1: good output.
- 5:03Speaker 3: Yeah, I should have mentioned that. That's the other part.
- 5:05Speaker 3: So the memories in these models, AI does know a
- 5:08Speaker 3: fair bit about you. It depends on the different tools
- 5:10Speaker 3: you've got. There's different types of the memories there. But
- 5:13Speaker 3: in a broad kind of framework, Think about AI having
- 5:18Speaker 3: a little notebook about you. And it's about three pages-ish
- 5:21Speaker 3: of notebook about you. And it depends on the different
- 5:25Speaker 3: models as to how they do this. But generally, it's
- 5:27Speaker 3: like who you are, what you do, what you're generally
- 5:30Speaker 3: working on. And it does really like to know about
- 5:34Speaker 3: the things that you care about. So if you think
- 5:36Speaker 3: like your children, your dogs, your pets, your hobbies and passions,
- 5:41Speaker 3: it certainly likes to scribble those away. also how you
- 5:44Speaker 3: try to talk to it. And then it starts to
- 5:46Speaker 3: get an idea of the human that it's dealing with
- 5:49Speaker 3: so it can more likely personalize things for you. And
- 5:52Speaker 3: so it's then kind of guessing based on your role,
- 5:55Speaker 3: your job, what you've done with it before, I got
- 5:58Speaker 3: an idea of what you're trying to achieve here. So
- 6:00Speaker 3: it'll help you do it again. So you don't have
- 6:02Speaker 3: to be as explicit. But then again, I'd also argue
- 6:06Speaker 3: that can be a bit dangerous because Just because I
- 6:08Speaker 3: did something once or I've worked in a particular area
- 6:11Speaker 3: doesn't mean that I'm going to do that kind of
- 6:14Speaker 3: thing again in the future. And so I find the
- 6:16Speaker 3: memory is a bit of a two-edged sword. For most people,
- 6:19Speaker 3: it's magnificent and you don't have to prompt as hard
- 6:22Speaker 3: with those magical prompts because it already kind of understands
- 6:25Speaker 3: what you're after. But if you're doing lots of new
- 6:26Speaker 3: or different things, then how you worked yesterday doesn't help
- 6:29Speaker 3: you today.
- 6:31Speaker 1: So given all these, given all these changes, how should
- 6:34Speaker 1: we be prompting AI now?
- 6:37Speaker 2: What's the best way to have a conversation with it?
- 6:38Speaker 2: All right.
- 6:39Speaker 3: I'm going to give you four tips. Tip number one,
- 6:43Speaker 3: we've got to forget what we've learned from Google. Google's great.
- 6:47Speaker 3: Don't get me wrong. Love Google. Google's awesome. But what
- 6:49Speaker 3: we've learned from Google over all of these years is
- 6:53Speaker 3: I will shoot it a, I call it a caveman
- 6:55Speaker 3: phrase of keywords. So it might be a best washing
- 7:00Speaker 3: machine 2026 or something like that. And then you'll get
- 7:03Speaker 3: a pretty crap response. And so you might go best
- 7:05Speaker 3: washing machine, family, grass stains, 2026 or something like that.
- 7:10Speaker 3: And so we did like yelling at our AI at
- 7:13Speaker 3: Google and we've got like four or five words and
- 7:16Speaker 3: then we'll get this response back. Now I do all
- 7:17Speaker 3: the work here because I'm hitting another word. I've got
- 7:19Speaker 3: to read another eight pages and all those kinds of things.
- 7:22Speaker 3: And what I found watching people is they will effectively
- 7:26Speaker 3: put a couple of words around their Google phrase and
- 7:28Speaker 3: give that to AI. So it's the equivalent of what
- 7:31Speaker 3: is the best washing machine I should get for my
- 7:33Speaker 3: family in 2026? It's still pretty the same thing, you know,
- 7:37Speaker 3: and people do shout their commands at AI, which is
- 7:40Speaker 3: write me a report that does blah. And of course,
- 7:42Speaker 3: then you get a report and then you go, that's
- 7:44Speaker 3: not a good report. That's a terrible report. And then
- 7:46Speaker 3: you're arguing with your AI and you're spending a lot
- 7:48Speaker 3: of time and energy of your emotional energy and also
- 7:52Speaker 3: tokens as well to be able to get it. So instead,
- 7:55Speaker 3: tip number two, tip number one is forget what we've
- 7:58Speaker 3: learned from Google. Tip number two is treat it like
- 8:01Speaker 3: a consultant. So you can imagine any of the kind
- 8:05Speaker 3: of consultant you would love to be able to hire.
- 8:08Speaker 3: They can be $ 3, 000 a day consultants if you want to.
- 8:12Speaker 3: When you bring someone into your business as a consultant,
- 8:15Speaker 3: you won't just go, write me a report or give
- 8:17Speaker 3: me the analysis. Instead, you'll be going, okay, so here's
- 8:20Speaker 3: what we've done before. Here's what I'm going to do.
- 8:22Speaker 3: Here's where I need to go. What are your thoughts
- 8:25Speaker 3: on this? You're going to have a conversation. do the
- 8:28Speaker 3: same thing with AI. And so use direct natural language.
- 8:33Speaker 3: This is where you and I always recommend dictate to AI.
- 8:37Speaker 3: You're more likely to deal with it like a human
- 8:39Speaker 3: by giving it the background, the context, what you're trying
- 8:42Speaker 3: to do, what you've done before, all of those kinds
- 8:44Speaker 3: of things. And if you're dealing with it like that consultant,
- 8:48Speaker 3: you're more likely going to give it more background, more context,
- 8:52Speaker 3: and help it to understand what you're truly after.
- 8:56Speaker 1: And I should clarify, by dictate, you don't mean be
- 8:59Speaker 1: a dictator, but press the little microphone button that is
- 9:05Speaker 1: in all the major AI platforms. It's so much easier
- 9:08Speaker 1: to speak these things than type these things.
- 9:11Speaker 2: Sorry, Neo, continue.
- 9:13Speaker 1: What is tip number three?
- 9:15Speaker 3: Or just on the treat it like a consultant, I
- 9:17Speaker 3: would also give a consultant documents. And so rather than
- 9:21Speaker 3: just say, do this thing, give it the background, the
- 9:24Speaker 3: context in a document. Say here, read section five, they'll
- 9:26Speaker 3: give you a nice bit of background. We would do
- 9:28Speaker 3: that with people, but we'd never do that with Google.
- 9:31Speaker 3: And so we've got to start giving it much more
- 9:33Speaker 3: background and context, and it does a much better job
- 9:35Speaker 3: that way. Tip number three is be inquisitive. So if
- 9:40Speaker 3: I've got a consultant, they've got a brain, they're really smart.
- 9:43Speaker 3: That's why I'm paying them to come in to help
- 9:45Speaker 3: me with these things. And so I'll go, hey, what
- 9:47Speaker 3: do other clients do? What do other industries do? Is
- 9:50Speaker 3: that a better way to do that? Can I achieve
- 9:52Speaker 3: this in a different way that'll land better for my
- 9:55Speaker 3: customers or clients? So being inquisitive, because it can actually
- 9:59Speaker 3: show you and teach you ways of doing these things.
- 10:04Speaker 3: And you might find there's a better way to get
- 10:06Speaker 3: these results you're trying to do and try to get
- 10:08Speaker 3: out of AI.
- 10:09Speaker 2: What is the final tip, Neo? Tip number four.
- 10:13Speaker 3: Ultimately, challenge it. So if you're a customer of one
- 10:18Speaker 3: of these consultants, Yeah, they'll come in, they'll tell you
- 10:20Speaker 3: what different industries and different clients and different companies are doing.
- 10:24Speaker 3: But you then have to go, yeah, that's great, but
- 10:27Speaker 3: I'm not those companies or that's not my industry. And
- 10:30Speaker 3: so it doesn't really apply to me. And so absolutely,
- 10:34Speaker 3: when you have this conversation with the AI, say, here's
- 10:37Speaker 3: my instructions, what I'm trying to do. Before you get
- 10:39Speaker 3: to do the thing, you might actually go, hey, what
- 10:42Speaker 3: a better way to do it. And it might come
- 10:43Speaker 3: up with something and you go, that may not land
- 10:46Speaker 3: for me. There's something you need to look at and
- 10:49Speaker 3: kind of really judge if you'd like, is that right?
- 10:53Speaker 3: And if it's not right, challenge it and tell it
- 10:55Speaker 3: why it's not right so it can then understand what's
- 10:57Speaker 3: not right. And so this way, you're not just accepting it,
- 11:01Speaker 3: you're more likely going to understand or get it also
- 11:04Speaker 3: to understand what you're trying to achieve. Another part there
- 11:07Speaker 3: is also get it to challenge you as well. So
- 11:10Speaker 3: I get mine to challenge me not only just on
- 11:12Speaker 3: how I can create the thing, but also should I
- 11:16Speaker 3: be even creating that thing in the first place? Are
- 11:18Speaker 3: there even better ways to achieve that? And so get
- 11:21Speaker 3: it to challenge you as well as you challenge it
- 11:23Speaker 3: when it gives you the response back or the document
- 11:28Speaker 3: or the output or the research or whatever it is,
- 11:31Speaker 3: and you'll find that you actually get a much better result.
- 11:33Speaker 3: But ultimately, out of those four tips, it's really deal
- 11:36Speaker 3: with it like a human being insofar as... It doesn't
- 11:39Speaker 3: know enough about you. You've got to brief it. It
- 11:41Speaker 3: can do a good job. It could also misunderstand things.
- 11:44Speaker 3: You can learn from it. It can learn from you.
- 11:46Speaker 3: But ultimately, you are the customer of this. And so
- 11:49Speaker 3: it's got to make sure it lands with you. So
- 11:51Speaker 3: check those things off.
- 11:52Speaker 2: Amazing. Thank you for sharing, Neo.
- 11:54Speaker 1: And hopefully our wonderful listeners have got some very practical
- 11:59Speaker 1: tips to talk to AI, perhaps in a slightly different
- 12:02Speaker 1: way today. Thank you so much for listening to today's
- 12:05Speaker 1: episode of How I AI. If this show got you
- 12:09Speaker 1: thinking about how you could use AI better, that is
- 12:12Speaker 1: exactly what we help people do at Inventium AI. We
- 12:16Speaker 1: run AI training programs for organizations and also for individuals.
- 12:21Speaker 1: So if you want to upskill your entire organization or
- 12:25Speaker 1: maybe just yourself, you can find out more at inventium.ai. Hey,
- 12:30Speaker 1: and once more favor before you go, if you've got
- 12:32Speaker 1: a moment, like just five seconds, I'd love it if
- 12:35Speaker 1: you could leave us.
- 12:36Speaker 2: A review and hit follow wherever you're listening.
- 12:39Speaker 1: It really helps the show reach more people and it
- 12:41Speaker 1: helps us create even better episodes for you.
- 12:45Speaker 2: Thank you so much for listening and we'll see you
- 12:46Speaker 2: next time on How I AI.
- 12:49Speaker 1: How I AI was hosted by me, Amanda Imber and
- 12:52Speaker 1: Neo Applin. A big thank you to Martin Imber who
- 12:55Speaker 1: does our sound editing and Jem Rubio for production support.
- 12:59Speaker 1: And thank you to John Kilby who composed the theme music.