Loading...



So the use cases I want to let I want to trust Claude to make me content. The content that I want to have created is meant to go onto a service called blue sky. I need to run a short post for blue sky about the AI course that I'm teaching. The posts should include an image and a hyperlink to the Irish innovation skill net. See what Cloud can do. I hear it does really good stuff with natural language. I'll give it something. I'll upload a file which happens to be a summary. And you can see, can't upload my files greater than 30 megs. So I'll have a look at this just a minute. So I'll add information from a URL, and this is a markdown let's see if it can do that. Listen to clogged, analyzing what I want, a short post. I hope you do it. Let me gather information from that URL. I'll check it. It fetched it, and it says automated reminder. Use citations, looking for information. Now, let me search for more information, automated information. I'll craft a short post. Here's what it's telling me. All right, I can use this
even with hyperlinks. And do I have a doc? Yes, I do. Do to have an image I'll see. Kind of give me, give me a branded give me a branded image from IR, dg.ie, see if it does that. It did. It grabbed all that, and you'll see that in real life next time before we meet. So I'm going to take, I'm going to take the PDF, the first hour and the second hour of what we do tonight, two things together, and say, make a LinkedIn post about we're just starting this program, and I may have it focus on what people want to get from it, as opposed to what the PDF itself talked about. So it'd be interesting to see what the AI can make for I've taken notes about what people want to get from it, but the goal is to actually, while it's fresh in my mind, and why I still have notes to have an AI do a synopsis, and then see if it's appropriate for the content that someone would read on LinkedIn, and we'll see by next week whether I've been successful in doing that. Okay, there are programs that 
can do all this stuff, and their programs, or their AIS, or their gpts, or their their programs that run inside of web browsers that can can make this work. It's part of the 4p framework that I have. The question becomes, what kind of tools do we want to use? How do these tools integrate? And then there are some of these tools for personal digital transformation, for collaboration, for content creation. This is an idea here for like I have. If I have a workflow that starts with an idea that I want to export to WordPress and publish it and then share it out to different services. That's a workflow. And everything you see on this screen here could be gathered. Some of the stuff's coming in from Ai, some of it's going out to different services like blue sky or LinkedIn or Instagram, all these things we're doing, transformative process we're doing have been part of your life for a decade. I'd say so you if you've used an iPhone for 10 years, you've been talking to Siri. If you've allowed Amazon's products, their speaking products, Alexa in your life, well, then you probably talk to Amazon. And if you're like my house, maybe you've had a situation where you've let the Google speaker argue with the Amazon speaker, the Alexa talk to the Google Assistant back and forth is quite hilarious and interesting. So they're in your space. They may be in your car telling you where to turn, where to stop, where to if you're if you have an easy chances are you have an AI inside the car telling you where to recharge facility is so I know people that talk to the phone, setting their calendar alerts, checking by talking to the phone, what's happening next. Series been around since 2011 you can use your voice to search on Bing with Microsoft or with Google supermarket apps. I mean, Tesco tells me where I'd forgotten to order, if I order online, and all kinds of social networks, to include YouTube, my most common space, uses an AI. Your music selection is being generated from AIs. Think it they know the mood you're in and the music you should be listening to. Predictive text on your phone being powered by an AI, okay, okay, I have a personal large language model as my 
main program. So I save a lot of stuff into a service called obsidian, and then it uses an open source LLM to find information I've written to myself over the last 10 years. It's quite interesting, and you might be joining a business, or you might be in a research section, or you might discover your university has built their own large language model. They use an open source one, probably one that's no longer you no longer have to pay for, and they're taking that large language model and bolting it onto the back of a research organization or department, or they might the company you're joining or the facility you're in might use a virtual private network to tunnel into a large language model that's being sold by one of the major providers. So I use obsidian as mine. I have journal items written over since 1999 and those things are in Markdown language text format inside obsidian, and it's aided by a large language model. So the large language model lets me find the stuff that's in there. If I go out and they use a large language model on a server, I'm going to burn more energy. So if I use chat GPT to look and talk and see something, and I execute a chat GPT routine that's going to generate text that I type into audio that's produced, or video that's produced, or images, it's going to use a whole lot more energy. I mean, generally speaking, the stuff I'm about to have you do with chat with Microsoft copod is going to burn. It's going to burn more Well, realistically speaking, it's going to use about the amount of energy required to boil a kettle of water. That's actually what it comes down 
to to make the images we're about to make. When Facebook or meta figured out how to train llama two. It burned through 539, metric tons of CO two, and the larger language models that the companies want to build are using a whole lot more energy. So I'm going to burn some energy. I burn some energy by asking what an AI knows about me, the use case why I'd want to do that? Because maybe the about page for the run EU course called Creating proactive workflows with AI. Maybe it needs a little bit of information about me, a little bit about Philippe banks. So use some chatgpt to look at it. I can use perplexity as well to look at the two of us and look at the course that we have. So there are two different companies open. AI is one in terms of chat GPT. Perplex is another company. They're using the same GPT four large language model model, but they tweak it in a certain way. I have found out that chat gpts specialized apps are better than what perplexity has. But perplexity does stuff with web search. I think that's stronger than 
chat GPT. You can restrict what chat GPT is doing by buying into an enterprise version of chat GPT, which means that none of your stuff actually goes outside of the enterprise service that may be running in it, in the your local country or inside the EU for the service you're getting. But the enterprise version of chat GPD cost you 200 euro a month to use. So this is asking chat GPT about me, and see what you know I was asking, you know, show sure that I know what I'm talking about with it when it comes to like digital transformation. So chat GPD went out across the internet, answered the question, and at the circling thing here responded. It'll go to a to Vimeo to where this video clip is saved, and not going to have much patience to wait for it to spin up. But I do know, when you play this slide deck on its own, this, this should work better. I don't know why it's not spinning. Why it is spinning? Well, here it comes. Here we go. I asked it. I asked write an about page for Bernie Goldbach. Make it sound like he knows about digital transformation. And one out chatgpt went out,
it produced all this flowery prose about me, you know, having deep understanding and research capability, and you know, I'm not sure that you know, you need to know all that. Okay, what we need to know is this, I want to see if, in the remaining time of patients that you have for our session tonight, where they're going to whether you're going to be able to create an avatar selfie. Now, a few of you by email said, look, I can generate that in another program. Okay, off you go. You can do that. But if you haven't done this before, ask an AI to produce something that looks like you. Well, then you need to be able to type in something about you. So you know, I want to make sure you record the prompt and pop it into the Zoom chat. And if you had a prompt notebook, a handwritten notebook you could keep, that's important, this would be the first prompt in a prompts notebook that you keep. But you're asking, you'll be using the Edge browser with Microsoft Copilot to create an avatar selfie. And then when you get your result, please pop it into the Zoom chat along with the words you use to do that. And then I'm going to show you, when I stop sharing the screen, that I want you to pop it across this shared Google Drive that all of you should have been invited to. But when you do that, when you put it into the Google Drive, make sure you put something about your name in the file name of the image you upload. If you don't remember to do that, well, there your email address that you're using will attach itself to it, and that'll be fine, but that's what I want you to be able to do. And I'll show you a sample of this. That's the first part of the breakout. The second part is to use Google Geminis nano banana to modify your selfie record the prompt that you use. And the modification I want you to want to do is I want you to include a piece type of food you enjoy eating, pop that prompt and your image up to the Zoom chat and then upload it to the Google Drive. Okay, let me show you some stuff that I've done now, first, I'll show you go to the chat to make sure. Okay, I'll show you where I am, when, when I was doing this. So first, first thing to do, I will show you this place. I'm inside of Google Drive right now. And if you have not found this this place, then what I will do is pop it into zoom and see into the Zoom chat and see, can you actually, can you see it? Can you see in the Zoom chat? Can the Zoom chat link? Can that link take you to where a shared place is? If it's not shareable, then I'll go back in and add it to but if you cannot see where I was in the Google Drive, run EU folder, it would be a folder I would have shared by an email a couple weeks ago. Well, then I need to make sure you're able to do that, because I need you to put it into the 02 images folder, the images folder into a secondary folder that's called selfie avatars. The thing you're going to produce. So I'm going to try, I'm trying to open the oh two images and
inside the oh two images folder, you'll see a thing called Avatar selfies. And what I want to do here is I'm going to upload a couple of images that I did. And sure enough, I'll bet I didn't put, you know, I got something here. I asked, I asked the Nana banana to do an image of me with us, with sushi. And copilot produced this image of me. So I got to put the word Bernie in the file name, because nobody will know it's me, so I'm going to rename it. Okay? So that's one place. The second thing I want you to put, the first thing that I want you to do is actually produce that image. So I'll give you a sample here inside of Microsoft, coppod, and what you see happening here is I'm in the Edge browser. You can see the copilot is running in the ED browser. And I went to copilot.microsoft.com and I asked it for this, an image of a sec senior university lecture, resembles Robin. And they produced Robin, not me, but they got an image, and then I uploaded that image. Then I took that image. I downloaded that image first, and then I took I when I downloaded the image, I then went, then went into I downloaded the image to my local and then I went into Google. Gemini, gemini.google.com and I asked it modify the attached image by adding a plate of sushi in front of the man, and it produced an image. And I'll put the images and the prompts in the Zoom chat. Okay, incrementally. It's important to know this. It's really easy. I would like you to use copilot. If you say, no, look, Canva is my buddy. I'll do that. Okay, fair enough. Or if you have another way of producing an image, and that's in another service, that's fine. But the goal is, across this training program, that you're able to say, I have certain tools I'm using, and they have some strengths or and the strength might be they're free in my university setting. So I'm using a tool which I know works, and the tool is called chatgpt, and I know it can produce an image, or it's copilot, I know produce really good text, or it's clawed and it can do text better. I would like to know that you can actually make copilot produce an image. The easiest way to do that is to use the Edge browser, bang in your image prompt, get the image, put the prompt and the image in the Zoom chat, then move it across to the Google Drive and put it there, into the runnyu ai folder and into images folder, into the sub folder called Avatar selfies. Done. Please make sure that you're able to put a file name with your name on it when you pop it to the place, Hiram. You have a question,
right? It's asking me, I could apply this. Do you want me to say, I'm just gonna say yes to all the stuff it said, so it went to top goal.ie and said, I'm gonna grab stuff and I'm gonna infuse when I infuse. Into the academic scene it created. So theoretically, it's generating something else. So I could have, you know, I haven't in that website, and I may actually find an image of me, I don't know. So this is all new. I've got. So now, to make all this work, I should have, okay,
what I should have done is put this thing into my slide deck, but I didn't, so we'll do it. So that's this place called it's the digital transformation group. So one person on this call probably said, Can you let me in? So I will Maddie, you're in. What's what's going to happen is next, last session that we're together, you're going to see how people are able to put content up into this space, and to get academic credit for the content has to have a title, and then should have three or four paragraphs relating to what the image is. So specifically, I really want to have, like, one piece of content going up in the first three weeks of the course related to like transformative processes that might be coming through the creative workflow you discovered, or the new tools you're using. And then in the month of November, a second piece which either reimagines the first piece or brings into the group a big reveal, you know, like, wow, this is important. People ought to know this. They ought to know I know this, and I'm glad I'm able to share it. So as the classes evolve, as the Tuesday evenings evolve, you'll see it's natural to produce your campfire moment on the back of something you shared inside the space called digital transformation, that there's another workflow, which is actually producing a booklet of about 40 pages based on our elements of sharing that are be inside that Flickr group. So there's an API, an application programming interface that lets us produce a book or based on pages that are inside of the Flickr digital transformation group. And I need to be able to show you that so that when we're together in Portugal, you can actually have this hard copy thing that's really lovely, and it's all done by the process that's outlined in the course, and then by the technology that's enabled by the protocols that are working together to make that product. Okay. So the point is, if you've done the AI prompt of yourself and uploaded it, great. If you've modded it with a piece of food, brilliant. If you've uploaded that to the Google Drive, perfect. If you haven't, but you put the images in the Zoom chat, okay, I'll move those things around, but I need to make sure we can populate the same shared space. And if it doesn't happen by next week, well, then we'll have probably two different spaces, a Google Drive space and a team space, just so we have the ability of people to pop stuff up, and you can put stuff in one place or the other. In this case, you put the stuff in zoom chat, which I can harvest at low resolution, and into the Google Drive at higher resolution, probably higher res. And I'm happy with that. Okay, any questions about the stuff we've done here, because I want to finish up the deck. There's a couple other slides, but we're at a point now where people probably have some observations or questions, so any anybody want to raise any issues, so pop the hand up so you don't talk over each other. But any questions, okay, all right, I'm going to Go back into my slide deck. 
right, people maybe should be at the top of this stack, but quite frankly, most people are on this course to learn how you could use AI and what kind of programs help. But people are important. I mean, artificial intelligence is just that you are the actual intelligence. All the many researchers say it's more than logical thinking. It's more than cognition. It it is something that connects you as a person to the ability that an AI might have to create text numbers and images along parameters. There, these services are very fast or very efficient. They can think quickly, but most of them aren't autonomous yet. But if you're like I am, you know that it's just a matter of time, and some of these automated features are going to be here all around us. Hopefully we'll be in a position where we'll whatever is unleashed in the next five years will still need a pause point or can be told to stop by someone's voice or action. People have to be kept at the front. Gerald Kane says the answer of doing all this is probably less technological. It's he outlines in his book called The technology fallacy, like how you are really the important key in making these transformative processes happening. He writes about it. MIT and Deloitte Digital found that if you're mature, if your digital technology and processes are mature, it's more about how you are doing it than how the technology is doing it. Digital maturity is more about people than it is about technology. If you're in a space where the culture that work you're working with embraces change, maybe a leader that is a change leader takes measured risks and invite and critically treasures the innovation that you bring to the space that's vital for the organization's survival, for the organization's growth. It means you and the people you're working with are dedicated to continuous learning, adapting to what the technology is doing, and structuring the technology around what the workforce is and workplace is good. Organizations encourage this kind of experimentation, but the problem that a lot of people encounter is, My God, where do you start, and can you trust it? And so if you're if you are part of an organization similar to the ones that I've been part of, there's not a lot of risks taking at a level or two above you. Definitely people are wondering, where should we go, and how can we trust that we're going to get there, right? These new digital skills that you're going to develop could actually disrupt the organization that's around you. And so just be aware that you might not be a change agent. You might be, you know, a disruptor, a person that's going to threaten an existence of an entity could be a subdivision, a staff member or an organizational structure, but it's important, because established business models today are going to have to change 

and flex to what AI is bringing to the mix. And if the flexibility isn't there and the individual skills aren't there, well then you know the organization is not going to be able to mature, grow and survive. The jobs market says that. So I mean, if you're starting at the very bottom as a solicitor, managing property, states of transfer of deeds and all that, lot of these things are automating. Now. Google's cut 1000s of staff. So as Microsoft and they rehired a lot of them, PayPal is saying we've got to cut some jobs and let the automation take over. Goldman Sachs says 300 million jobs lost or degraded by AI. So that's not to say there's won't be 300 million jobs replacing it says, Look, some specific entities, job skills, job requirements are going to change. Meta says, Well, we have to cut down because we're going to build our data centers, and they're targeting people at the bottom of the workforce as low as performers. Some of this stuff was used by Doge under Elon Musk to trim back staff members and part of the US budget. All these kind of things are happening
to prevent this from occurring. Part of this module teaches that your staff skills, your traits, are needed for digital transformation, how to cope with change is important. Working with people who you can suddenly think faster than and work faster than is important, because if you don't want to be identified as the wrong kind of change agent who's going to take somebody out of the mix, people have to be involved in the process, selecting the process, selecting the tool, assessing whether it's going to work. Learning and Development is important. Leadership. People in position of senior leadership, understanding what is happening is important, how their decision making can be accelerated by these tools. All these things are documented by people who earn more than the HR degree that I have. The EU says, on top of all this, follow the law. On top of all this, be ethical, there are principles, and I'm going to go deeper into what they are about, using an AI to profile somebody in a biometric way alive, there have to be a robust analysis of the social environment in which the AI tool is used. Seven key requirements spelled out by 
the hyperlink that's attached to this document. There are ethical guidelines. They should be empowerment. The ethical guidelines should be allowing you to make informed decisions, allowing you to respect and to enhance fundamental rights with oversight. Human in the loop, human on the loop, human in command, which is different than the way that open AI is pushing stuff out of the chat GPT. So just know that if you're on the top of and always getting the latest updates because you've got a paid account of it, you might be crossing over to where you the human are really important to make sure the AI is not running off and making conclusions and implementing strategies that take away from these ethical fundamental guidelines The EU has so the Act says there are specific risks practices that are prohibited, high risk applications that have to be identified, clear requirements for high risk and then specific obligations with conformity assessment before any AI systems put into place. I'll cover these, again, more in a more granular way, in follow on sessions. But this is the this is the rule, and it's the AI act is being implemented at national level 

across the run EU universities. And as it does, it collides with the US saying, Hey, we can count on the AI to lay off people, reduce on site quality control. AI is being used to replace what used to be funded scientific research. Chat bots are replacing call centers algorithms replacing personnel with layoffs that are happening across the US Federal workforce. So the poor four part framework ends with protocols, which might be just platforms that tie things together. Protocols enhance mapping. Routines that I use protocols, or sophisticated ways of connecting applications together through application programming interfaces, when your mechanisms connect the protocols, data you establish in one place ends up transferring to another place. That could be a problem, and you need to know when you've done that workflow you built some with make.com or n8 and or with a variety of other services, I will show you that data can start to leak, and that can be a problem if it's sensitive data, but all this stuff's happening fast, and that's why you're on this course, to see these kind of leakages And know that future proofing, what you're doing will happen by looking at specific implementations that we'll look at. So answer this form if you're still on the call, I need to know from top to bottom, move these things around to say, hey, look what I really want to be able to do is, you know, create and revise stuff you're looking at here would have been given to a business group as stuff, you know, like, it could be, like the number five things important to you to say, You know what, I need to figure out my personal AI that I'm going to be able to use. That's important. It might be no no. Biggest job I have is document management. And if it's something that's not mentioned here, that's okay. You have mentioned a lot of that already, and we will try to cover that stuff in a follow on lesson, because to keep you interested in the program we're doing, and to keep and to ensure that the in residence portion in Portugal is exceptionally relevant, we need to know that we're going to help you get to a point to where you're able To use these kind of tools and and to get get actions faster and information quicker, and collaboration at a higher level faster. So I do appreciate the stuff that you've you've that folks, you've answered here and and, like I said, we will definitely try to make sure that we're on message and we are in focus for what you want to get done and what you are find as being most important. So interesting to see how people I thought I was the only person going along meetings, and I thought that I didn't. I did know I wasn't the only person having to revise documents. So it's good to see we're in the same thought space here. I use different protocols to make this deck, so I ran it through a couple different AIS find out if I really know what I'm talking about. 

this that lots of stuff that we might have thrown at you, that I threw at you here is sort of innovative out of that you feel it's kind of chaotic. That's okay, because Steve Jobs would say that's the way it is. Innovation often happens at the edge of chaos. It does fool a lot of people in an AI tool can fool the music industry and AI tool can write essays. Can fool the standardized testing mechanism America has in universities across the states. Bots now are outnumbering humans online. You've probably encountered plenty of them on websites saying you want to chat with a robot, interact with a person. Hard to get a human sometimes. Perspective a lot of people have, like the person that runs signal is that they're what we're doing right now is definitely above all. There's more VC money in it than money being paid by subscriptions. The cost of energy is massive, and it just can't be sustained. So Meredith Whitaker is saying, Hey, I tell you what, at least be learning a workflow, because the tool that may become your favorite tool might cost a whole lot more in the near 

future, there are biases. As we know every AI is going to generate a result which may not be exactly what you want. It has an inherent bias based on what it learns. Know that you have oversight. It's your supervision, advisory capacity that makes this all happen. So what you did, you saw copilot generate outlines from a web page. You saw some tools, saw Claude, write some content, use Copilot to create an avatar and then revise it with Gemini. Question whether that AI can be used for productivity. You saw sway running, which you didn't use, but you saw the Edge browser, and you might have used chatgpt Four Omni, but chatgpt Five is probably what you use, as well as cloud copilot in the Edge browser, Gemini, Flexity and behind the scenes, the thing that powered some of the stuff that I that that is in my website, called topical dot A is powered by a service called obsidian. Slido is your tool. So don't worry about answering this poem, because I know ask you too much stuff, primary sources are in this deck. If you want to follow the links, it'll take it to them.
These are the links directly to the videos. My final thoughts are, you have to experiment. You have to be aware that the goal in the near future is to automate some of the stuff you're doing. Get insights from each other, enhance your own creativity, keep yourself at the central this process, so it's authentic. Be flexible, be curious, be skeptical, and please be around next week so we can talk to you about it, any questions or comments, because I've reached the end of my limits. Yeah. Okay, Maddie, I have to figure this out. So pop it on signal, if you wouldn't mind with your your email on signal, I'll help figure Sara, do you have a question or just just clapping? All right, I will stay online for a few minutes. Well, until the last I'll stay online for another 29 minutes, how about that? And then go, I will send you a calendar of invitation to the next meeting on Thursday for next Tuesday, you'll see a couple of things coming to you. So auditors should probably send someone said, Hey, here's the summary here times two, and you'll get a link, two links, well, probably one link to a folder on Vimeo. So you can actually rewind this clip. It's about a half a dozen people that weren't on this call. Well, about four people don't call that say they said that they wanted to be and you can always rewind this stuff. Filipe, how are you?