Loading summary
Transcript69 lines
- [00:02]
A
You're listening to ASBO International's School Business Insider. I'm your host, John Brucato. Each week on School Business Insider, I sit down with school business officials and industry experts from around the world to share their stories and explore the topics that matter most to you. Find out what it means to be a school business official and get your insider pass on all things school business. Hi everyone, and welcome back. Today's guest is no stranger to the show. Aziz Agaev, CEO of Flow List, returns for a deeper dive into using AI effectively in the school business office. This time we're going beyond the basics. Aziz walks us through how school business officials can create their own custom GPTs and including tools like Prompt Assist and prompt Engineering to turn everyday thoughts into structured, high performance AI queries. We also explore Azee's signature framework, the six phases of AI adoption, which helps SPOs move from the curiosity to confident use while staying focused on real impact and responsible practices. Whether you're writing a board summary, refining your capital planning narrative, or analyzing staffing costs, this episode will show you how to get AI to work with you, not, not instead of you. Aziz, my friend, welcome back to School Business Insider. I'm glad to have you back on.
- [01:29]
B
Thank you. Thank you for having me.
- [01:31]
A
All right, so let's jump right into it. It's so nice to have you back. What? Your episode is one of the most highly regarded I've had in recent times, and I think not only does that speak to your professionalism and prowess in the industry, but just the need for school business officials to get a little bit more in tune with how to effectively use AI. So I'm excited to have you back on and kind of do a little deeper of a dive. So what are you seeing right now in terms of how school business officials are embracing AI or experimenting with it? You're traveling all over the country, you're doing a lot of professional development. What are you seeing out in the field?
- [02:09]
B
What I see currently is that during the trainings, everyone comes in with a bias saying that AI is cheating. I cannot use AI because I want to use my own material. And after the training, people realize that it is not the cheating material, it's just an assistant and it can help you to do secretarial tasks a lot faster. And it helps you, for example, take notes for you and you don't have to take notes and then read your notes and convert them to minutes, for example. When people realize the usage, when people realize how it can create efficiency for their professional life, then everything Changes.
- [02:57]
A
So how are you getting them to have that light bulb moment where they, they realize exactly what you just said? I mean, they're seeing AI as an assistant, as a tool doing clerical repetitive tasks. What does it take to get someone who may come in with their bias to that point?
- [03:13]
B
I think the most important approach that I've been doing is to put it in perspective. Because if you don't have use cases, people do not know what it will help them with. For example, I'll give you a note taker example and this is most of the time I talk about this. When you say there is this AI tool that takes notes for you, people's reply will be, I can take my own notes. But when you say there is this AI can take notes for you and during your meeting you can have it take notes. Then, then copy your transcript, put it in ChatGPT or Gemini and ask it to create a follow up email to send to all meeting attendees with all the tasks, with the summary of the meeting with the important action items. And that changes the perspective. I give another example for teachers. If a teacher records their class during their teaching and then put that transcript in ChatGPT and say, act as an experienced math teacher, tell me what I can improve in my teaching. It's not just a note taker, but it's important how you use it. The use case changes everything.
- [04:33]
A
Right? And I mean I use the note taking function all the time. And another thing that I love about it is I can actually be 100% full focused and engaged in the meeting, not trying to make sure that I'm copying down important things that people are saying. Are you finding that as well? Are you able to show that, listen, you don't have to really worry about taking notes anymore. You can actually be a participant, not a note taker.
- [04:58]
B
That's another very strong thing that note taker brings to the market is giving your pure attention to who you're speaking with. And yes, people do tell me after they start using the note taker, they say that my pure attention now is in the meeting. I don't have to take notes. I don't have to be in two places. I can do eye contact a lot more because in my meetings when people take notes, it's really hard to focus because they're not looking at you, they're looking at the agenda that they're taking notes to. Yeah, right.
- [05:34]
A
And you know, we talked about biases and we were implying that people come in with a negative bias. Have you had any of your participants come in that are just way over the top with AI and it's just the solution to all of their problems. Are you seeing school business officials maybe using it too much?
- [05:51]
B
I haven't seen that yet. Everyone that uses it first, school business officials are new to the field. I didn't see many that are extremely tech savvy. I saw a few. And those people who embraced AI since the beginning and they're using it mostly they're using as an assistant. It's not to create a material, but it's reshape the material that you already have so that you don't have to go through that many, many times and you don't have to ask your colleagues or friends to look through that. But there are a few business officials that use it extensively and they created a lot of efficiency in their districts.
- [06:43]
A
Has there been anything that has really surprised you relating to the adoption curve of school business officials and AI? You just mentioned there's a few individuals you've met that are really kind of utilizing it to its full capability. Are there any other surprises that have really crept up throughout all of your trainings and consulting?
- [07:04]
B
I'll give you one example would be the custom GPTs. Everyone talks about custom GPTs, but I have one of the business official in Chicago, Ron Anderson of Bellwood, what he did, he created these kind of custom GPTs per document. For example, he has a custom chatbot, let's call it a chatbot for his contracts. He has a chatbot for, for his ERP software. He has a chatbot for employee handbook. And he created dashboard off of those chatbots. And now employees can easily go and interact with it. These are the things that create efficiencies and these are the things that it's not cheating, it's just having that available for all of your staff at their fingertips. Instead of them calling you, they can easily go and ask questions and you're.
- [08:04]
A
Kind of freeing up your own brain capacity because you're not having to remember every little nuanced piece of maybe like a collective bargaining agreement, you can put it through a GPT and probably get the answer much quicker than you having to thumb through a hundred page document to find the answer. So I mean that, that, that's a great example.
- [08:23]
B
Yeah, exactly. And I actually showed both examples. Some districts, I ask a question about collective bargaining agreement, they pull out a binder that is 100 pages, they go through the binder and they find my answer. With this district, when I ask, he goes to chatbot, he asks the question and right away we get the answer and answer is 90, 95% accurate most of the time because it reads from the document before it answers the question.
- [08:56]
A
Yeah, incredibly impressive. You know, one thing that stood out to me the last time I had you on is we talked about the importance of prompting when it comes to LLMs. And that has really been a game changer for me. And I use artificial intelligence pretty regularly in my everyday work. So let's talk a little bit more about how to effectively prompt. Can you tell me in our listeners what makes a prompt great instead of just kind of functional?
- [09:23]
B
I would compare it to a reporter. What makes the reporter great or not would be their questions. If they ask right questions, they usually get right answers or great answers. I think prompting is crucial to our AI interaction because if we don't have a good prompt, we will not have a good response. And structuring the prompt correctly, treating it as an assistant or when I say in my trainings, I say five year old. So instructions like a five year old. Take this from this table, walk five steps, put it on that table, but on the right left corner or right center corner of that table.
- [10:11]
A
That kind of sounds like my everyday life with my toddler right now.
- [10:14]
B
Exactly right. So when you treat it as a five year old, that changes everything because now you have to tell everything, you have to specify the format, the output, the tone, all of that you need to specify in the prompt. And when you have that prompting, you know, proper prompting, then you get a better response.
- [10:37]
A
I also noticed too, in terms of prompting, trying to fit too much in one single prompt can lead to a lot of unexpected behavior. So to your point, kind of chunking it out and maybe putting like themes together in a prompt seems to be more effective than maybe a paragraph of text. Because what you can do and what I've noticed is when you prompt and it responds to you, you may catch an error that's specific to that one chunk of prompt that you can correct. Rather than putting a whole block of text and trying to work backwards from there. Are you finding that similarly with how to effectively prompt in AI?
- [11:17]
B
Exactly. I think the topics should be separated. If you're talking to an AI about few topics, try to separate them so that it's not getting confused used. That's one. And the second thing, to your point, putting everything in one paragraph, I compare it to a narrative and a checklist. When we have a narrative to do what we need to do in a paragraph, we can miss a lot of things. But when we have a checklist listed by numbers. Then we check off and make sure that we did that. And that's why the checklist exists. So if we are asking multiple questions or needing multiple responses, then we can list those as numbered questions and say answer each question individually and make sure do not miss any question kind of approach rather than putting everything in one paragraph. Then it gets confused. Yes.
- [12:16]
A
And a couple of the best practices that I've learned from you is you had mentioned this briefly before, but telling the instance what it is and what the expectation is like you are an expert note taker or you, you're an expert in collective bargaining agreements and negotiations, setting the stage that way and then also at the end of the prompt telling it to ask any clarifying questions before proceeding. Can you talk to me a little bit more about why those two prompts are really important to really create effective output?
- [12:50]
B
Yes. And that's usually what I leave everyone with during the training. I say if you don't take anything, just take this about prompting. If you structure your prompt with act as, which is assigning role, then you put all of your context, all of your questions. And then third, you ask for the constraints, don'ts the formats, the output, how you want it. And then you tell it to ask you clarifying questions before it responds. First act as, let's talk about that. If you say to AI, I am sick, what should I do? It will say, relax, drink more water, what happened? Call 911, see the doctor, and so on. But if you said act as an HR director at my company, I am sick, what should I do? Then it will say you need to check how many sick days you have. You need to make sure that you inform your employer before you, before you take the sick day. So the perspective changes. And that's why we have roles now talking about the questions, clarifying questions, every single thing that we miss to tell in our prompt. AI will assume those. But if we allow questions, it will ask us questions. And with those questions we will answer and then give more context, the response will become a lot better. And that's why the questions are important. Because if it doesn't ask questions, then basically you're saying you can assume anything you want, but that doesn't work. And you can realize that with deep research, because deep research is more expensive tool. Before deep research starts, it usually asks questions. So they made it mandatory for deep research to ask questions, clarifying questions before it starts the deep research.
- [14:59]
A
So you're really eliminating any room for error or assumption there. Because to your point, in our minds we may have our own assumptions that ChatGPT is going to take the next step that we would logically take, but you really need to approach it as if it's never done this before, in a sense, because you need to be incredibly explicit with what you want it to do. And to your point, with a narrative, you know, when we're doing a block of text, we're human, we forget things, we may not have thought of, certain things that we have to explicitly ask Chat GPT to do. I know in my own usage that that prompt at the end, ask any clarifying questions has been a game changer because as thorough as I thought I was in my prompt, it has asked really specific clarifying questions. I'm like, oh, I didn't even think of that. And that totally makes sense. It's just been really, like I said, a game changer for me in terms of being able to effectively use ChatGPT, rather than, you know, your example of I'm sick. What do I do? Put some context around it, Right?
- [16:00]
B
Exactly. And I compare it to your assistant. You have an assistant, you don't take a manila folder and drop it on their desk and leave what you do, you explain what you want with that folder. You say what kind of report you would like, what the format will need to be. And then I asked this question last week, do you. During my training, I said, what else do you say before you leave? They said, one said, do it. The other one said, do it fast. And other one said, when can you finish it? I said, no. As a good leader, you say, do you have any questions regarding this task, which is a natural thing to do? And that's what it is. Treat it as your assistant. Let it do everything but explain what you need. At the end you say, do you have any questions?
- [16:59]
A
Yeah, I think that's kind of the larger gaps that I've seen is it's just, I don't think it's natural for people just to talk to what they would say, a computer like they would a human being. So this advent of large language models allows you to do that. I just think this is a very unnatural thing to do. We, we don't necessarily talk to Siri like they're our assistant, although that's kind of how it's modeled. Other pieces of computer AI functionality just doesn't function like LLM. So I just think maybe that's part of the problem is that you're solving with your. Your PDS is just changing the mindset that it's okay to talk to this as if it was someone sitting next to you.
- [17:39]
B
Right, exactly. And that I experienced that firsthand. I bring people for a demo to the stage and. And I say, okay, I want you to create a prompt for deep research. I will open the dictation and I want you to dictate instead of typing. And after I open it, I say, look at me. Don't look at the computer. Because when you look at the computer, then you start structuring your sentences, making sure that you didn't miss anything, making sure that all of the sentences are in the correct order. But AI has improved a lot. We don't need to do that. When I dictate, I'm like talking to you. I explain everything. I explain the format, the output, everything in a natural tone, and then I just send it in. Because it understands what you're trying to say.
- [18:32]
A
Right? Is AI pretty good at that? I mean, I speak and can be very fragmented sometimes, and as thoughts enter my head, I may go down a different path. Has his AI evolved to a point where it can kind of parse through all those human characteristics of not being perfect in terms of sentence and thought structure?
- [18:52]
B
It is. I mean, in my experience, it has been pretty good. What I do is I explain myself and then I say, for example, if I'm explaining something, something in Excel, I say, column A is my school, column B is first name. Oh, no, wait, column B is actually last name. So I fix myself. Now, when I fix that, the response, I see that it fixed it. So it takes your natural language and it understands the natural language and it responds based on that, I would say to everyone who is listening, try to naturally speak. When you do a prompting, when you, when you put that text in a box, just open dictation, naturally speak your mind and see the result. Because when we naturally speak, we cover a lot more than we structurally explain.
- [19:49]
A
Right. That'll be my next thing once we sign off. I'm going to try the dictation piece and see how that goes. So let's talk a little bit more about custom GPTs and prompt assistants. You've been working on these throughout your work with building custom custom GPTs for prompt assistants, prompt engineers. Can you explain to me what those are?
- [20:10]
B
It's actually great transition after this. Right now, what I do, usually I create custom GPTs, and those custom GPTs would be prompt Engineer, for example, when I have important things I do. For example, if I do deep research, if I'll be using ChatGPT agent to create a presentation out of my expense Data. What I usually do, I go to my prompt engineer, I speak my mind, and then that prompt engineer restructures it to a proper prompt. It says, your role will be this, your objectives will be this. This is what I want out of your response. And here is the format. And then it says, ask me clarifying questions. And it's very simple to create those custom GPTs. I recommend a white paper by Google. It was written regarding prompt engineering. What I usually do, I take that white paper, it's about 60 to 70 pages, I attach it to that custom GPT and then in that custom GPT I trained it, I gave it instructions. I said, you are a prompt engineer, you cannot answer my questions. Your only job is to take my prompt and make it a structured prompt. I will tell you if it's for deep research, I'll tell you if it's for custom GPT instruction, I'll tell you if it's for video. Whatever it is, you do not respond. I am not interested in your input. I only want you to structure my prompt. And what that custom GPT does, it takes that prompt, it processes it and then it puts it in a great format. Most of the time when I do a deep research and I go through that practice, I don't get questions because it's so clear, right?
- [22:13]
A
So what's the advantage of creating a GPT like this Prompt assistant rather than just opening a new chat and starting from there?
- [22:20]
B
Because you need to repeat yourself every time you open a new chat. You need to explain the structure, give the instructions every single time. With custom GPT you give instructions only once. You can refine instructions later, you can go edit those instructions, you can add remove from those instructions. But that only has one job. To give you the proper response in a format that you like. I'll give you another use case for custom GPT is board meeting minutes. Imagine you take notes with note taker during your board meetings. And then you have a custom GPT that you uploaded your board minute template. You can upload past 10 different board meeting minutes for tone, style and how much description you want to put in there. And the template will be there. And then your instructions will be act as a board secretary, take these notes and create a board template. Now to do this in a new chat, every single time, you have to upload five to 10 files. You have to do those instructions every single time. Then you need to paste your transcript instead. You do it once and you reuse it all the time.
- [23:41]
A
So really you're unlocking the potential of GPTs. When you're identifying a repetitive task that requires you to that that's very input heavy. You're basically once you get that set, it already knows what you're looking for before you even start the prompt. Is that correct?
- [23:57]
B
Yes, exactly.
- [24:00]
A
Stretching your school's budget is tough, but smarter purchasing can help. That's why schools across the US turn to KPN and peppm. These programs, created by an educational service agency in Pennsylvania, give you pre bid contracts with nationwide vendor access and ready to use discounts. No extra bids, no extra work, just real savings for schools from schools. Learn more at www.thekpn.org and www.peppm.org.
- [24:32]
B
So.
- [24:33]
A
How can school business officials create GPTs like we just talked about? That really turns their ideas and goals into optimized prompts automatically. What, what first steps should they take? I mean, I know there's different variants. There's Gemini, there's ChatGPT, there's a whole bunch of them. What would you recommend a school business official do to get started to create their first GPT?
- [24:56]
B
I personally use three different options. One is two ChatGPT, the other one is Gemini and then the other platform, it's like PCAX Chatbase. I'll talk about those in a little bit, but for ChatGPT and I'll explain the differences. Chat GPT. You can only create custom GPTs if you are paid subscriber. It doesn't matter if you're $20 subscriber or the you know, the pro subscriber, you have that option to create custom GPT. After you create custom GPT you can share with your colleagues, your friends, the link and they can use that created GPT. If it's available for anyone with Gemini Gems, it is available for free users. You can create Gemini Gem which is custom GPT for Gemini. However, to the point that I looked, I think I looked at last week, you are not able to share it. This is for only your use. You cannot share it with people. And that's the difference. If it's only for your use, Gemini is free and education subscription for Google would cover Gemini and Gemini Gems. Another difference between Gemini Gem and Google Custom GPT is that with ChatGPT's custom GPT you can actually chat with the custom GPT and create instructions by chatting, which means that your input will be converted to proper instructions with AI assistance. With Gemini you cannot do that. You have to write your instructions on your own. And another trick for Gemini would be going to another chat in Gemini and say create me a proper instructions for this custom GPT and it will automatically create proper instructions and then you can copy and paste it in the instructions page.
- [27:06]
A
So it sounds like ChatGPT just kind of took that extra step where you can use natural language to build out instructions and use cases for GPT.
- [27:17]
B
Yes, exactly. It makes it easier. You can upload the file to as. As you. As you use ChatGPT regularly, it has that kind of platform. You can upload your file, you can put your instructions, you can actually dictate. I usually dictate, I open the microphone, I explain what I'm trying to do with this ChatGPT, and then it automatically creates the instructions, creates the thumbnail, creates the name. All of that is done in that chat interaction.
- [27:46]
A
Great, you've already mentioned one with the collective bargaining agreements and human resources. But what are some other practical uses for custom GPTs and school business?
- [27:57]
B
I would say most of the documents that we usually look into would be a great custom GPT. Examples, staff manuals would be great examples. What if we do staff manual GPT and have it available for our internal website for all of our employees? Another custom GPT I created for one of the districts is the Transportation Helper, the student transportation Helper, I would say what it is they have on their website, they have PDF data of all of their student numbers and with the intersections where that student gets dropped off or picked up, hours and everything, what I did with this custom GPT, we uploaded those PDFs, two PDF files and now when you say where's my bus? It asks for student id. When you provide the student id, it automatically lists when is that student getting dropped off, what is that intersection, what time the bus is in the morning, what time bus is in the evening, in the afternoon. All of that is done in seconds. Instead of calling the transportation department, them looking it up and giving you the time. That's all time consuming and time is expensive.
- [29:23]
A
So this GPT is a publicly facing GPT that any parent can go to the website and type that prompt in. Is that right?
- [29:31]
B
And that brings us to the thing that I didn't cover, which is the PCAX and Chatbase. These platforms are. They allow you to create custom GPTs and then they give you an embed link where you can actually embed it on your website. And that will be facing anyone you want. So basically a fast GPT creator, if you would you put in knowledge, you choose a model, you choose. If you want to use OpenAI's 4.1, 4.0303 model, all of that you choose, and then it creates the GPT, gives you the link. You go to your website, put that link anywhere, and it becomes a chatbot.
- [30:19]
A
Oh, that's great. So you can essentially turn your own custom GPT into what you just said, a chatbot on your website that's really specific to whatever you need. And you, as the school business official or whoever in the district, has ultimate control in terms of fine tuning it. You're not relying on a third party company to go back and forth. You have all the control, is that right?
- [30:45]
B
Exactly. You can always go and edit. Let's say you're doing staff manual. The staff manual changes. You go to that custom GPT, remove the previous file, add another file, and it's updated.
- [30:59]
A
And having used it, I think it's pretty safe to assume you don't need to be a tech whiz to do any of this. Right. It's just becoming familiar with how these GPTs are created. You don't need any kind of coding knowledge or anything like that. It really is just nice, natural spoken language.
- [31:17]
B
Absolutely, yes. And that's my message to everyone who listens to this podcast. The best way to start is to start. There is no training. You can learn AI from AI. So if you don't know how to create custom GPTs, go to chat GPT and say, I have never done custom GPT before. Walk me step by step through this process and hold my hand. It will do step by step, explicitly, everything, line by line. It will tell you. And if you don't have understanding, if you don't understand the step, you always go back and say, I didn't understand the step. Elaborate on this and it will open it up and it will talk more about it. So I, and I say this all the time, AI is not like technology that you have been experiencing. It's not like Excel, it's not like Word, it's not like PowerPoint or Photoshop Illustrator. It is a technology where you can learn that technology from the technology. Yes. You had that documentation, Excel, documentation that you could go and read all the time, but we never do because the documentation is hard to read with this. It changes, it's not a tool, it's shift that to the world. It will shift the world to a different direction and you can ask all of your questions and it will be answered hopefully.
- [32:52]
A
And I'm assuming you can have it, you can tell it to talk to you like you're a five year old too, if you have a hard time understanding it, right?
- [32:59]
B
Yes. Actually, one of the custom GPTs I recommend is the newsletter GPT. And in the instructions I put write the newsletter as a sixth grade language, sixth grade English, because, and I tell this to my audience, usually no matter what education you have, no matter what degrees you have, and no matter how great you write, if a person who reads it do not understand it, there is no point of that newsletter. So when you write your newsletters, when you write your emails, communications to parents or stakeholders, make sure to put that sixth grade level, seventh grade level, so that it's plain English, it doesn't use complicated words.
- [33:49]
A
Right. So you often refer to the six phase model of AI adoption. I'd like to kind of unpack that a little bit more. Can you walk me through each phase and really what it looks like in practice? We talked, I believe, a little bit about this on our last episode, but I'd really like to revisit it because I think it really captures the essence of how you can see that arc of becoming a pro at LLMs and AI.
- [34:17]
B
Yeah, I always talk about this in my presentations, that it starts with surprise. When you hear what it can do, then you become curious. You look over the shoulders of other people and see what they do. Then you start using it, and then you get confused, you don't understand what's happening. And then the lengthiest of all I say is the frustration phase. And everyone will get frustrated before they can become productive on it. That frustration phase is long. I still experience it and I use it on a daily basis. I train this. But every single day I experience it. Because sometimes you ask a question, it doesn't get you a correct response. The most important thing is not to give up and say, oh, if it's taking this much time, I can do it better. No, you cannot. You just need to stick with it. Because if you stick with it, then you become natural. And then you start that joy phase. When you start enjoying that interaction, when you start enjoying saving time and having more time with your family, then you become productive. And that's how I basically transition people. Because what happens, what I experience is that some people start using AI few times, 10, 15 times, they don't get what they want. And once they don't get it, they say, oh, it didn't help me, it's not good. And I, one of my friends said that I tried it so many times, it's not responding correctly. And I said, maybe you're not asking a correct question. And then he realized, yes, he didn't ask. It was a very Vague open question that it cannot respond.
- [36:06]
A
Right. So in your trainings and your travels, what would you say, most school business officials, what phase are they in? Primarily.
- [36:16]
B
I would say at least 50% is at that surprise and curiosity level. Because I get questions like, what is this? Is it a website? Is it an application? Which tells me that they're not familiar with this technology at all. I do have only very handful, maybe people that are at that productivity level, but most of the people are at that frustration level because they're trying to do it and they are not able to do it. And some people think that they're productive, but they're only using it to rewrite texts, to, you know, to just send emails or to come up with text, which AI is not that.
- [37:07]
A
Right. And what would you say is the key to progressing to the next phase with confidence? Is it just perseverance and sticking with it, or is there something else that school business officials should be doing to kind of progress throughout those six phases?
- [37:20]
B
I think, yes, sticking with it and trying over and over and not giving up is the key. But also there are a lot of trainings. They can get that hand holding at first, get that confidence, and then it will become naturally. Even those people who use AI, when I bring them and I open that dictation, I see that they cannot speak to it.
- [37:47]
A
Right.
- [37:47]
B
They cannot explain their thoughts clearly. And that is just an experience. You do it once, you do it twice, 10th time, it will come naturally. I think sticking with it, just experiencing with it. And I usually say that you need to train your subconscious, subconscious mind. That AI is easy. It's not like other technologies. It is extremely easy to use. Doesn't matter your tech level, you can just start typing or speaking. It will respond.
- [38:24]
A
So, you know, once people progress to that joy and productivity phase, I'd assume they're using it all the time. Maybe they're just making some assumptions in terms of the responses they're getting from AI. How can school business officials evaluate AI generated content for accuracy and bias control?
- [38:44]
B
That is very important field of this AI because AI hallucinates and I tell everyone that you cannot use anything AI gives you without reading and confirming it. I would say most of the things are common sense. So, for example, if you're creating policy, it would be common sense because you know your domain, you know what policy needs to look like. However, when you're asking for facts or when you create policy based on the state guidelines, you can always ask AI to give you sources, and you can then click that source, go Verify the information. Make sure you verify. And sometimes I've been hearing, I haven't experienced it, but I've been hearing that AI actually can create its own sources and those sources will be fake sources. Make sure that the sources are legitimate. It's going to a research institute or it goes to a journal or some, some somewhere that you are familiar with when you read that information. Don't rely on the source too. But I would say if you're asking for a fact or if it's a guideline regulation, make sure to verify it with that source and do not rely on it. It does hallucinate a lot.
- [40:05]
A
Yeah, creating its own sources, that's very Orwellian. I'd be a little nervous about that. So, yeah, make sure that you're checking those sources. Okay, so winding down here, I'm sure everyone listening would agree with me when I say that free time, or time in general is at a premium. We all have so much going on, so many plates in the air. Aziz, if someone only has about 15 minutes a week to focus on AI growth and maybe really get better at it, what's the best way to spend that time?
- [40:36]
B
To do projects? To start doing a project a week, let's say they can say, okay, this week I will just create my presentation of my revenues for my board with ChatGPT agent. You already have that presentation, that's perfect. Let AI create it and then you can just compare it just for fun to see if it does what it needs to do. I think the, the more we start using it on our real life projects, the better the result will be because then we can actually use the response most of the time instead of just trying it for fun and seeing if it'll do good job with, you know, this fake or imaginary tasks. Just do your own tasks. Even if it's 15 minutes a week, doesn't matter your own tasks. Just pick one and get on it. Try AI and see what the result will be.
- [41:42]
A
Great. So what's the next year look like for School of Business officials and AI? What is, what is your prediction? You know, if we're, we're sitting here In August of 2026, what are you hoping to see in terms of progress?
- [41:54]
B
I personally hope to see our leadership to embrace it and start using it on a daily basis. What is happening? All of our students use AI to its potential and our leadership and teachers, if they don't use it, there will be one time, maybe in a year, maybe in two years, two parties will speak two different languages. Students will have different level, different knowledge teachers and our leadership will be different knowledge on AI. So I think our importance is to take it, start using it, start interacting with it so that we are better than our students as our leaders. And the leadership is, I think the leadership's role is to kind of equip their team with the proper technology and not hesitate, but embrace it and start enforcing it so that everyone uses it. Because using AI can reduce tasks from eight hours to half an hour, which is good, eight hours or half an hour.
- [43:12]
A
Right. And just think of how much free time you'll have to either do something you want to do or get to other tasks that are on your to do list just by saving seven and a half hours worth of time. Right.
- [43:23]
B
Exactly. And I usually ask that how many of you, I say to the audience, how many of you have a policy that you wanted to revise but you never had a chance to do that? And this technology allows us to do that, allows us to focus on a bigger picture and not just day to day little tasks that occupies all of our time and we don't have time to spend sphere to the big bigger picture.
- [43:49]
A
Right. Well, Aziz, thanks for joining me today. Every time we speak, I have a list of other things I need to try with AI. So that list has grown after this conversation. But I appreciate your time and really enlightening our listeners on all the capabilities and potential with artificial intelligence. And I'm sure you'll be back on in the near future.
- [44:07]
B
Thank you. Honored to be here.
- [44:11]
A
Thank you for tuning in to School Business Insider. Make sure to check back each week for your favorite topics on school business.