Tēnā koutou katoa. Welcome to this Alan and Clark webinar with Wright on improving communications with the help of AI. My name is Rebecca Grey.
I am one of Allen and Clark's senior consultants in research and evaluation. Fair to say I've written a few reports with a bunch of different collaborations but not many of them have been our new robot friends, giving much help with writing at this stage. Look, we have hundreds of registrations come through for today and we want to give a warm welcome to those who are joining our community for the first time.
It's great to have you with us if it's your first time joining. A little bit of background, Alan and Clark are an Australasian based consultancy dedicated to making a positive impact on communities throughout Aotearoa, Australia and the Pacific. Our areas of speciality include strategy, change management, programme delivery, policy and as I mentioned research and evaluation.
As an organisation we give a damn about empowering you to overcome society's biggest challenges and that's why we regularly run things like these free webinars. We create desk guides, we provide expert advice wherever we can. For today's discussion we are very fortunate to have one of our partners, Wright, come and share their expertise on plain language.
So Wright is a plain language consultancy with a team of trainers and content specialists. We've had some of the team here for a webinar earlier, it was earlier last year, but today we're specifically looking at plain language in the age of AI assistants and in order to help us kind of position what their expertise is in here, the Wright team have given us a quote from some tech experts who've worked with them. This is Tim Clark and Phil Vennell of AI Native and what they said was the team at Wright Group are the best people I know for AI plus human training.
What makes them perfect isn't just their familiarity with AI, it's that they're exceptional writers who understand how to create content that communicates clearly and effectively. AI works best with clear, concise, natural language and Wright has that mastery. Nice one.
So it's my pleasure to welcome Friederike Teig and Una Cox to the studio today. Rieke, Una, can you please introduce yourselves and your work on this topic? Thank you, Rebecca. So kia ora koutou, my name is Rieke or Friederike, I go by Rieke and I'm a plain language consultant and trainer at the Wright Group.
Let me just maybe briefly explain what the Wright Group is. We are a clear communication agency or consultancy and we're one of the biggest in the world and we help clients around the globe to write more effectively and more clearly with a really human or user focused approach. We offer training for that like technical writing training, business writing, government writing and also really precise types of training like creating A3s or proposal writing.
And we obviously also support writing with AI and for that we use a trademarked approach that's called AI plus human. And maybe a little bit about myself. So you can hear I'm not from New Zealand.
I'm originally from Germany. I came to New Zealand around 15 years ago now to do a PhD at Victoria University in applied linguistics. So it's all about language.
Yeah, I've worked at Massey University as well as a lecturer in linguistics and have a pretty strong background in academic writing and teaching and research. But in a different life, I've also been an editor in a publishing house. Yeah, I think that's plenty about me.
A little bit about Una maybe. Thanks, Rieke. Hi, everyone.
My name is Una. You won't need, I don't think, an AI bot or a tool to detect my accent. I'm originally Irish.
I've been in New Zealand my first seven years. Like Rieke, I have a background in linguistics as well. So both of us are all about the words, the words, the words.
In my role as innovation lead and plain language consultant at Right, I love creating new ideas, new concepts, new things, embracing change as opportunity. And even within that, I like to think of it's A plus I, so the artificial tool plus the I ourselves driving it because we bring that human trait of intelligence to the artificial tool as well. So they go hand in hand and fit quite nicely with Right's brand of AI plus human.
Awesome. Thank you. One more note to our audience too.
From these registrations and from previous webinars on AI and writing, we've gathered something like 400 comments about what people find challenging. So what we discuss today will be based on the main ideas there that Right can address with their work. But just to start, and as we mentioned, given our different accents, we're not even just talking about English here.
Could we explain what we mean by plain language in this context? Yes, we can, of course. It's strongest tool to communicate purely and effectively. Ultimately, the reader is at the heart of all of your communication, so very reader focused.
So think of C words, so clear, concise, consistent, well-organised language and design. So you'll see a definition there on the screen from the International Plain Language Federation. So really focussing on the wording, the structure, the design of your information so that your readers can find, they can understand, and they can use that information.
Otherwise, what is the point of creating information? So clear writing principles and where AI sits in all stages. You'll also see a little AI, not AI, plain language standard on the side of your screen as well. So 10 fundamental principles which really put the reader up front, set you up for success if you use these principles, and we'll send that out after today as well.
And we'll talk about these principles as we go through today's webinar. Great, thank you. So we're talking about producing clear, effective communication, potentially with the help of AI.
Now, as we know, there's an increasing amount of content out there that has AI fingerprints all over it. Not always a good reading experience, right? I mean, it's not just the em dashes, it's the vibe. I mean, some people already complain about AI slop in all sorts of contexts.
Can you talk us a bit more about what that slop idea means in the workplace? Absolutely. Yeah, so you're absolutely right. There's a lot of sort of AI slop being produced, and in the workplace, we prefer to talk about work slop.
It's a bit more precise, focussing on content produced at work, and that creates problems at work. And most of the time, that's because superficially, it looks amazing. And so we get sort of lulled into a false sense of security, easy acceptance, and we don't look under the hood, where if we scrutinised it, we would find issues around faulty logic and not meeting the purpose, and maybe not meeting the standard that we actually meant to meet.
Yeah, we really like this definition that you can see on the slide by Better Ops Labs and also Stanford Social Media Labs, and I'll read that with you. Yeah, work slop means AI-generated work content that masquerades as good work but lacks the substance to meaningfully advance a given task. So it looks great, but it might not be so great.
And I don't know, like, why does it happen? I don't think people want to produce this. So what are the causes, Una? What do you think? I think it happens for multiple reasons. My honest opinion is that people have very unrealistic and high expectations of the tool.
They're not experienced with it. They don't know the difference between different platforms, what they can do and what they can't do, which platform might be best given the task they're doing. And you'll see all of these causes on the left there.
So they're not all in isolation. Some feed into the next one. So those expectations, we want to be clear about what it is we're wanting the tool for, which tool is best, and what that output is going to look like.
Quite often, I think, as a user, we can be tired and we give vague or lazy instructions. But again, hoping for the ultimate result, we get that in all areas of life, not just using AI. And then we tend to do too little refining and editing of what we need as well.
So we can become really lucky sometimes with a very fast prompt or really unlucky and frustrated. So those frustrations then can make us feel like it's not affected or can cause problems down the chain. So there's a nice quote from the same research that Rita quoted before.
So the effect then is it just passes that burden down when it isn't clear to begin with of what you need the tool or what you're asking for. And then that requires more work from the receiver. And that could be a colleague.
It could be senior management, could be ministerial level. But regardless, it means time where you don't have the time. And thanks for pointing out that impact on the reader.
It's just yet another kind of layer of media literacy we need to develop, isn't it? We don't really want to. We don't want this slope. So how can we avoid producing it? OK, we're going to talk about that today, which is nice.
So you'll see on the slide, there are some prompts that were in the workshop and the webinar guide for today. And we're going to address those. And like everything, we love using processes and steps to help participants and people remember how to do these things.
So we're going to talk you down and break it into those smaller questions through eight quick steps to avoid workflow. Step one. All right.
Step one, you know, fix your thinking, right? So AI can't do anything without you. It needs you, right? And so clarify your thinking and be aware of what AI can and cannot do, right? And that AI in general, but also at particular platforms that you're using. And one thing none of them can do is read your mind.
So you really need to communicate clearly what you expect of it to get a good result. And I put a little picture on the slide, you know, and that's a reference to the Hitchhiker's Guide to the Galaxy. You might have read it.
And I think it's a really good analogy because in that story, humans ask a supercomputer, what's the answer to life, the universe and everything? And the computer goes away for 7 million years and thinks hard and then gives the answer. And the answer is 42. And the humans are naturally quite upset because that's not a helpful answer.
But of course, it's the answer they deserve, right? Because their input was so vague, they can't expect a clear output. And we put a bit of a sort of gold nugget of wisdom on the slide, good in, good out, right? So you need to communicate clearly and clarify your expectations and what you want to get to actually get that, right? In the middle of the slide, I put a good tip. Think of your AI as an enthusiastic new graduate you recently hired.
So they have a lot of knowledge and a lot of enthusiasm, sort of book knowledge, but no knowledge about your particular situation yet, right? Your organisation, the subtleties of your job and your role. So those are all things you need to clarify for them to do a good job. Same for AI, same for human.
Sure. So you've described your AI as this capable graduate and hi to all the capable graduates in the audience. I would say it's an alien who's masquerading as a graduate, but that's fine.
Synergics. But yeah, the principles, the same principles apply, right? You know, how do we approach them? How do we advise them? Perfect. And that brings us into the second fix, the second step, which is good planning, which goes hand in hand with good thinking.
We can't really separate them. If we invest time in our thinking and planning, then I often think that sets us up for success. And it'll give us a much more realistic expectation instead of that disappointment that we often face.
So we don't want to rush into the job and just use AI for the sake of using it. It's trendy or it's going to be cool. That can often cause frustration as well.
And quite often, there's a false economy or economy of the truth in the idea that AI is going to save me time, but it'll only save you time if you invest that time upfront. So just be aware of any fallacies, any flaws in your thinking. You do have to invest time to gain time at the end.
So really invest it. Plan your output. So you'll see some steps there on the screen.
What should it look like? Okay, what format do you want it in? Is it a memo? Is it a report? Who is the audience? Again, is it going to cabinet perhaps? Is it going to senior leadership? Is it going to your staff meeting on a Friday? Think then about the planning as part of that. Who are the readers again? Are they colleagues? What process? Where could you put AI into that workflow process? And we'll talk about that a bit more in the stages later. And just be aware, again, platform awareness.
Is it the best tool for the job you're doing? So a calculator is better for a quick sum rather than popping that into a tool. And I'll just mention again that that idea of thinking and planning really kind of is at the foundation of our plain language standard that you see on the screen. And I probably should mention that that standard's also aligned with the ISO, the international ISO standard for plain language as well.
So that's good to know, because that's operational in 40 plus countries and more countries are buying into it. So not just in English, in other world languages as well. And I think at this moment, maybe translated in 20 plus languages as well.
So it's a real good tool to have in your toolkit for AI. Cool. So what's this step about? Okay, so this is the five steps that we talk about in all of our workshops.
So they're really our cornerstone in terms of good thinking and planning. We've talked about our step one and step two. They were our strong fixes.
Then we have the writing stage, the editing stage, and the proofreading stage. So there's an argument and a case for AI at all of these stages. I want you to think about where can I use AI in the loop, just as much as where can I use myself, the human? Because it'll add value in certain stages, depending on the type of role or job you're giving it to do.
So context is key. So you might decide to use it in thinking. I'll not talk through all of those examples there.
But really, it can massively help find your blind spots in your thinking. So maybe you're very aware of all of your readers or you think you can confirm that for you or can find any blind spots and do some scenario building, do some analysis, do some research at that high level for you. But definitely, you need to plan at the start what stage it will add most value to your work as well.
So you'll see arguments for it right through. And we'll talk about some of those as we go through the other slides as well. Okay, great.
So next audience question theme, what should I include in a prompt if I want to get the best response possible? And so I think I'll take that. So you've done some really good thinking and really good planning. Now you really also want to convey that to the AI, right? You don't want to keep that to yourself.
And so a prompt is just your way of communicating with the AI. So make that count. On the slide, you might see, talk to the AI like you would talk to that newly hired graduate who doesn't know the situation yet.
So what would you share with them to allow them to do a good job and not break out in sweat, right? So we've put a little prompt here and you can see how nervous he gets, right? Not just because the language is difficult here, but also because that's very limited information for the graduate to do a good job. You're not telling them what document you expect, who this is for, how long it should be, what information to include, what purpose it serves, right? So you're kind of setting the graduate up to fail and do the same or don't do the same for AI, right? Give them both more information. And sort of generally, there's a strong agreement that these five elements that you see on the slide should be in a good prompt, in good communication.
So one thing here that you can see is clarify the output. I've already said that, Una has said it. Tell them what is it you are expecting.
Is it a brief? Is it a memo? Is it a casual email, right? And also clarify the purpose. Will someone make a decision based on this? Will they just be informed? Is it supposed to be persuasive, right? That'll require very different work. And then you can see the next step is for AI specifically, clarify the role, right? Like tell it, hey, for example, act as an experienced risk analyst, you know, or something completely different because it'll change the tone, it'll change the information, like act as a primary school instructor, you know? And in that same realm, clarify the context they need, right? Anything that they need to know, any additional information they need to have to do a good job about the information, about the project, about the document.
If you have a pretty complex task, it might be a good idea to break it down step by step. You might do that for a human graduate, right? You're like, oh, I don't think you're as experienced yet. How can I mentor you through this, right? Like first do this, then do that, then do the next step.
And that can be a really useful thing for AI as well, because it can itself also run into the wrong rabbit hole and sort of get confused if the task is very complex. And then it's a good idea to break that down. Next step here on the slide is tell it what tools to use.
I'm still on the previous slide. Oh, sorry. That's all right.
You know, like you would tell a human, right, please use that database, please refer to those articles and so on. So tell the AI as well. Again, I can't really read your mind.
And finally, everyone loves a good example. You know, we all like to say like, oh, look, here's a good report, and please make the next one look like this. Yeah.
So save yourself a lot of hassle and save some good examples and give them to the AI, because it's very good at looking at it, analysing it, and then taking it to the next report and writing it in the same style. Right. Yeah.
Let me just show you. And because I feel this feels really quite complex, right, and you feel like maybe that this is going to get really long and excessive. And that can be for a really complex task, right? But it can also be something like what you see on the screen now, right? You can see it's still not terribly long, and it has all the relevant elements.
Like we're clarifying the output two-page report, and we are clarifying who this is for, senior leadership team, and what they'll do with it, right, the purpose or the outcome. You know, they'll make a decision in a meeting based on this. It gives more context, like the time frame.
It clarifies tools, like draw on this database and so on. It gives an example, and it does a little breakdown as well of more complex steps. First, you know, assess the risks here, and then start writing.
And please also include actionable recommendations. So it doesn't have to be extremely complex to include all these five steps. Right.
And this is just one prompt, though, with five steps in it. Yeah. Yeah.
So using all the elements that we've just discussed in a prompt, is this where we also want to think about plain language principles in order to help our prompts get clearer outputs? Absolutely. Yeah. Okay.
Yeah. Yeah. I can talk to that now.
So we'll highlight a few of these plain language concepts that will for sure improve your prompt. So we've heard Riga talk there about that essential context. You know, it's not psychic.
We have to tell it what the context is, and sufficient context as well. So try to keep out irrelevant context. And that in itself is a plain language principle as well.
So the most important and then the nice to know in that type of order. So be specific and avoid jargon is also something to remember as well. Why? Jargon is ambiguous.
It doesn't often mean the same thing across different fields or industries. For example, what does it mean to leverage synergies? All of a sudden people are agile. What does that mean? And you'll see, or will have seen in the previous quote, and how not to prompt some examples of jargon in there, which wasn't very clear.
So unclear in will be unclear out. So good in, good out. So we always at right talk about the power of the verb.
So what is the verb? It's your action word, your call to action, your ask. So instead of putting out like that wish list again, I want to be more specific. So create a, produce a, outline, dot, dot, dot.
And I think we'll hear lots of arguments for how important the verb is. Keeps it active and keeps the I, us as the human, the agent at the front. So natural tone as well.
Sometimes they are, well, chatbots have been trained to interact in natural language. So again, talk to it like you would just expect it to talk back to you too. So be conversational just as you would with a graduate or a colleague as well, where you might ask it questions, correct it, improve it, repeat it.
Don't just become passive consumers of the information. Remember to be active and keep challenging the turtle as well where needed. Do you say please and thank you at the toilet? I say please, because it doesn't extend, you know, it doesn't add more prompts.
I wouldn't say thank you because that actually creates a new prompt and a new reaction that uses a lot of energy and water. So I would say don't do that. I do say please, because I'm scared they're going to take over the world and be our overlords and they might kill me last.
Because I was really polite, but that's just me. Let's see how that works out for you. Cool.
So we've got an output. Sometimes though, it still doesn't feel like you're saving much time if you have to keep going back and going back and redoing things and editing them. Like how do we avoid too many time-wasting iterations? Well, we can apply all of the steps we've talked about so far, particularly really good solid thinking and planning to begin with.
That will get rid of a lot of some of these iterations, but of course it will be hard to get rid of 100% of them. So you'll see here clear prompting, clear wording in those prompts, active, the verb up front, what is it you're looking for, will all result in a better result with less ambiguity. So get the ambiguity out of your prompt and that will make hopefully that output more to your expectations if you've been clear about the context.
But some tips in addition to what we've already discussed is don't make your life harder than it needs to be. Don't reinvent the wheel. One thing we all have in common is the fact we're probably time poor and have limited time.
So save the good stuff. Really put that into either a prompt library where you can just swap out variables, maybe just bracket out certain things like maybe the role or what types of questions you want to add in or what it is being produced, what format it's coming out in, what style, an example you're putting in, and also maybe save an example too can be quite useful. So if you have what a good report or a paper looks like, a good piece of communication depending on your world and your work, then save that.
Because remember, if you don't guide it, it's often going to this big infinite library and will imagine or create what it thinks you think good looks like. And good in anything in life looks very different to lots of people. So make sure you prompt it clearly.
So save those things, maybe style guides and different things as well. Put in that you want an em dash with the space either side of it. We're going to argue for the em dash, Rebecca.
We are fans. But yeah, save time. Remember, invest time to save time, but don't make it harder for yourself.
If you've got a really good prompt that works well, keep it somewhere handy. It could even be in a folder in a desktop if you don't want to put it into your tool. Yeah, I think we have an example here.
So I really love how you're saying like, don't send it to the infinite library if you actually want it to come to your library. So save things in your prompt library, in your example library. And if you find yourself repeating instructions over and over again and you're writing them fresh again and again, it sort of defeats the whole point of using AI.
It's supposed to save you time. So for example, here on the slide, you see if you keep telling it again and again, these are my style choices. Please write in New Zealand spelling or please use sentence case in my headings.
Don't write that again and again. Save it as a template somewhere and then either upload it or copy paste it in. And can you include this as part of your big multi-prompt as well? Oh yeah, you can just- Like producer report in New Zealand English, blah, blah.
Yeah, I would put it right at the start. Don't use it only when you're rewriting, when you're already unhappy. You know, your prompt window can take quite a lot of information.
Yeah, that's good to know. Yeah, because that point about New Zealand spelling reminds me of something that came up in some of the comments from these webinar registrations. So hi everyone who raised this because I agree with you.
People aren't comfortable when they ask for an AI-assisted script or a summary, whatever, and try to incorporate that in their work and what they get back is something that sounds nothing like what they themselves would say, right? And some people have even pointed out the voice of it often seems maybe a bit flashy, maybe a bit sales pitchy to have come from a New Zealander. Like, do we want a humble deadpan AI just for us? I don't know. But that culture thing, you've got to be quite self-aware, don't you, about what you actually want it to sound like? How can you get it to sound like you?
I, you know, you get something back and it's a structure of what you might say, but then you've got to maybe put in some work and I've called it de-clawing in the past, you know, someone gives you a script and you're like, well, yeah, I'm going to de-claw this before I sign it. AI was not involved with that. Anyway, yeah, look, how do we do it? How do we create something that sounds like us, the individual, or perhaps us as the organisation? I'll stop talking about M-dashes.
Well, I feel like absolutely AI is capable of doing this, but of course you need to tell it, right? Like when I said, you're right now sending it to an infinite library, which is probably very American influenced, so it'll pick up on that tone. And if you have a tone in mind that you want to sound like, of course you need to tell it. We already said, it can't read your mind, right? And so one way of course, is to just actually tell it, put it in the prompt, you know, I want to sound like this, more formal, but friendly, avoid those pesky M-dashes, you know, and so on.
So you can put that in. But of course, for that to be successful, you need to actually know what you want, right? So you need to know what a Kiwi sounds like compared to an American. You need to know what the branded voice of your organisation is, and also how you could describe your own voice, right? Sometimes it's already done for you, right? I'm pretty sure most of you in your organisation have a style guide or a brand guide, and it's all distilled there.
So you could just upload that, keep it handy, upload it or copy-paste it into the prompt, and then just tell the AI to follow that. And you can also create for yourself a voice guide for your personal voice, maybe several for different purposes, right? You might want to have like a casual email voice guide or a, you know, more formal voice for external stakeholders and so on. But again, that probably takes a bit of work initially to then save time later, because you're building it over time and see what the result is, yeah? You might create a template that says, I always want to have short sentences, I don't want you to ever use the word leverage or undertake, because those are awful, and just my personal preference here.
But, you know, and slowly you'll realise, okay, this is coming closer and closer to what I want, and you'll still refine it. Yeah, yeah. Of course, if you don't want to do any of these kinds of work, you could just use a really good example of something that you like, that sounds like you, that sounds like your organisation, and then upload it or, you know, copy-paste it into the prompt and say, please analyse that for me.
And I have a little bit of an example here. And it looks like it, right? So if you find that you have a piece of writing that you don't like at all, you know, it doesn't meet your standard of what the voice is, maybe for your organisation or yourself, then get something that's really good and do a rewrite with that. So take the good one, put it into the AI and ask it, please analyse it and describe the tone.
And it'll give you a really good analysis and a good description. And then you use that description, you copy it out, and you start a new prompt saying, now rewrite this other text using this, following this description, and then see what you get out of it. Usually, it's quite a good result, I would say.
You might learn a few things about yourself in the process. Absolutely. Cool.
Those are some great tips. Hey, we're nearly at the final step. So we've got a product, we've refined the voice it's using.
But we still can't use it just as is, right? We still need to give it a pretty critical edit. We do have cautionary tales piling up about what can happen if you let AI a bit too loose, starts making stuff up. If we're using this to help us, we want to ensure accuracy.
But how do we do that efficiently? So yeah, I kind of have to be honest here, like you cannot get around editing, right? Like, and you don't want to because you are the relevant knowledgeable factor in this whole process, right? You don't want AI to entirely take over. Like, it shouldn't do all of your work. Otherwise, we are, in fact, redundant, right? So I put a question, like a gold nugget on the slide, and it says, am I comfortable taking responsibility for this? Yeah, always ask yourself that because you are responsible for the result, the AI is not, right? And so, yeah, always think, like, for a human, you would always check, you would do a peer review, you would do an edit, and then only would you sign off and publish.
So do the same for the AI, yeah? One way where you could save a bit of time is you could first ask the AI to edit for you. Yeah, not the final step, but one step in the process, right? So you can do that in different ways. One way would be a self critique.
You could take the text that AI produces, and say, like, could you now self critique that? And maybe on a scale from one to 10 on different sort of categories or aspects that are relevant to you, like accuracy of facts, tone, plain language use, and so on, meeting the purpose I defined. And then say, give me a rating from one to 10. And then you can also ask it to say, like, what are the problems that kept you from giving this a 10? And then you can use those to say, now fix that.
And then you'll get much closer to something that you actually want. But that won't stop you from also doing, ideally, your own edit, right? Like, so your human eyes are the final stop, right, and the final sign off. And I wrote on the slide, with AI, you must be an astute editor with really good critical thinking skills.
Because I think one of the problems that sort of occur with AI users, we sort of start to disengage. And we just sort of go like, oh, you do it. And I'll hope for the best.
And then I'll be, I'm disappointed. Or I'm impressed by the slick surface. And it looks so good, that I won't look more closely.
And I don't scrutinise. And I don't discover that actually, there's a lot of problems under the surface, right? Like we're under the hood. And like faulty logic, you know, more complex problems than, you And really look for that, because the AI needs you there.
And that really is a positive message, right? You're still relevant, and you want to stay relevant, right? Like, you don't want AI to take over your job. So good news, you are still very relevant there. No, that's great, though.
Thank you. Yeah. And I think one of those things that it made me reflect on too, is a first draft from a human looks like a first draft.
You know, it's got that saying, not quite sure about this, or, you know, need to fact check that. Whereas the first draft from an AI often looks like a final report. Which is why people trust it too readily.
And it's like, no, it doesn't. They haven't checked the logic yet. Yeah.
Cool. All right. So we have on to our last summary slide, I think.
Yeah, thank you for that. So you'll see this fabulous pot of gold has appeared on your screen. And the idea of this pot of gold is it's now full of golden nuggets, golden tips, steps, processes, whatever word you want to put on it, to really help you avoid AI generated work slop.
Key hand in hand, first two steps as a quick summary, are really effective thinking and planning, even in our right workshops, we often say that's 40 to 50% of your time. So quite often, if you want good writing, even at the end, you need to really invest and get no surprises at the end. Because if you rush up it, then you'll be surprised.
And that surprise isn't often, oh, wow, it's quite often frustration, and a bit of wow as well, probably, to what it has produced. Then the step three, the crucial part as well, really put AI in the loop at the right stage. So some research coming out quite recently has shown that less people, the trend is less people are using it fully for their writing, more for research and other types of things.
But overall, the hybrid is where you want to get it right. So use the human plus the AI, because we still have to prompt it for research, prompt it for what we want it to do. And these nuggets will all apply to producing really good output.
So that workflow, it could be in thinking and planning, always, always, always at the editing and proofreading, as Rikke said, just don't be tempted because you're tired to passively accept or to readily accept. We think we've all seen cases in the media where companies have readily trusted it. Probably, I often think it's just time per has caused these things.
And trust, it has to earn your trust, like anything in life, let the machine earn your trust, but just don't always give it too much credit. And then communicate clearly, what is it? Like you would if you're briefing someone on a briefing paper or an aid memoir, what is it you want from that aid memoir or policy document? Same with AI, be very clear verbs in terms of step five as well, or your kind of golden nugget, really put crisp, short, concise verbs in there. Remember, you're trying to express clearly.
You're not trying to impress the AI tool, I should say, with your language as well. You don't want to go, I know all these big verbs and big words. No, you want it to do the job quickly and promptly and effectively.
That will help them reduce unnecessary iterations. So save time by saving those things in your library, wherever it might be, in your prompt library or on your desktop. Customise your instructions.
Rikke talked about that quite a bit. If there's a particular voice, a style, a pattern, a good prompt you've got, then don't reinvent the wheel. Remember, that would, what would I say, that wouldn't be the point of using the AI at all.
And then always, always, always edit the output. Just don't think that it's going to be 100% right, as it would be with a colleague. You would still probably review that and assess it before you put it on your website.
Do the same with your AI tool as well. So I really think those eight steps will do it. You will know where your own weakness is within that step.
Maybe you only need to start doing three, four and five. Maybe some of you are new to it and need to really give consideration to all eight. But you will know yourself what's best.
Okay. Do you want to talk a little more about what we're going to be directing people to in the resources? I can do that. So if you are keen for more or have more questions after today, if this has whet your appetite a little bit, we do have lots of workshops.
We have some 60-minute micro sessions coming up next week. So we have AI Prompting 101 and we have Advanced AI Prompting. So just really short because we know people are time poor.
So how can we give them the skills to set them up for success? Then we also have other workshops like Critical Thinking. To me, you can't use the machine without good critical thinking skills. So it's a full day workshop online and in person, business writing, all of these and a half day AI writing one as well.
So write smarter with AI. We have tonnes of free resources. In today's world, everyone wants to know what's free.
So they're on our website and I think they'll be sent out after today, Rebecca. And yeah, we'll send the plain language standard that we talked about at the start because that's crucial. Those 10 principles of plain language, everything from your purpose, your content, your structure, to your paragraphs, your sentences, your word, and your layout, and your tone overall, and your editing, just really sum up and sit very nicely with the eight steps we've shared today.
Great. Thank you, Una. And I believe there'll be some links to some of Ellen and Clark's resources there as well, which some take a slightly different tack, but same sort of end game involved.
So yeah, have at it. Look, I think we do have a couple of questions of time, minutes left for questions. So shall we just have a look at what some of these audience members have said? And we'll see if we can address this with what we've just discussed.
Thomas has been asking how to encourage a team to use AI, but still feel assured it is their own work. I think with that, show them a good result with it, show them something really effective that has saved you time. You know, a real life example that people can relate to, whether you're updating maybe your work from home policy or something else, it was too long, you've drafted it, you've had iterations, and then boom, it's done that.
If you show them where it'll fit in their workflow, in their working day, where it'll save them time most effectively, if you have a group meeting around that, really get people to air the problems, first of all, define the problem. What are the challenges? What's eating up our time? From there, define that problem, and then see where AI could fit in that five step writing process in terms of thinking, planning, writing, editing, and proofreading. So Thomas, I think if you show a great example, a real example that's worked well, and assure them then how, as Rika talked about, how they can get their voice into that, how they can still make sure they're the driver of the output, and they've created the output, that can help people understand when it's real and relevant, I would say.
I think, yeah, maybe for Thomas as well, give people time to play, right? So give them time to try out AI and see the results during work time maybe, and maybe also as a team, set some time aside where you all try out something together and then compare the results. So just give people more confidence to work with it and share results, and maybe also give them some guardrails, like, I hope you have an AI policy, right? So they know comfortably where, how they can use it and how they shouldn't use it. So research shows a lot of staff still actually use it without telling you, and often in a way that you might not appreciate.
So help people use it in the way you want them to use it. Sure. And we had some other questions about the types of tools or perhaps types of plans that specify AI involvement.
I think that's possibly where it comes back to your organisational policies and guidelines. But look, we're going to have to wrap up. We've got about five minutes remaining on the screen.
So did you want any last things to say about that, or shall we? About the questions, Jo? About organisational plans, or is this...? Do you find some have less iterations? So Esther's asked if some AI tools need fewer iterations than others. There can be a wee bit of a change. I know you've experienced some, Rika.
I mean, get familiar with different platforms. A lot of people have their preferred platform, right? Like Cloud to me is very casual. I'm a Gemini girl right now, but that might change, right? And it also changes for different platforms, right? For example, Copilot is a lot more linked in with your Microsoft suite.
So it can do a lot more that way. So get familiar with different platforms and see what they can do for you in your particular work. And of course, secure versus free versions as well.
So if you're using the work one or you're using it at home and being curious. So just remember your privacy policy in addition to your AI policy will keep you safe for a reason. I think we have some webinars about that, actually.
So look at that. Hey, look, thank you so much for all those tips on steps of playing the language in the age of AI assistance. Thank you for getting these.
Look, we've answered a lot of detail about some of those questions in advance. I'm sorry that we haven't maybe covered all the live questions that just came through, but we are happy to catch up with you in the audience at any time to answer more questions, discuss some of the potential ideas that this may have raised in ways that either of us could support you further. So there may be a button on the screen where we can be in touch there.
And after the webinar, we'll be forwarding some more links to some of those resources that we mentioned. So look out in your inbox for that. Hey, it's been fantastic having Rika and Una in here with you, with us.
So on behalf of our audience and also our team here at Allen & Clark, thank you so much for coming in. OK. Thank you.
OK, thanks for watching, everyone. Ka kite anō.
