Ep 124: 5 Ways Generative AI Shows Up in 2024

As we dive deeper into the era of artificial intelligence, the potential of generative AI keeps expanding. From automation to personalization, the applications of generative AI are boundless. Let's explore five key ways in which generative AI is expected to shape and transform the year 2024.

1. Advanced Data Utilization:

In 2024, the organization and utilization of data will become paramount. Companies and individuals will leverage the power of generative AI to harness the vast amounts of available data. OpenAI's GPT-4, for example, leverages the entire Internet as a language model, while Adobe's FireFly utilizes image data. By effectively using data, businesses can unlock valuable insights, drive innovation, and make informed decisions.

2. Automation in Workflows:

Automation is set to be a significant trend in 2024, allowing individuals to streamline their workflows and save both time and effort. With the integration of generative AI into various tools and applications, tasks can be automated, increasing productivity and enabling individuals to focus on value-added activities. From automating simple tasks to generating content, AI-powered automation will continue to evolve and transform industries.

3. Personalized Customer Experiences:

In a world where 71% of consumers expect personalized experiences, generative AI is set to revolutionize customer interactions. AI-powered automation in call centers will enable tailored and efficient customer support, while making interactions more meaningful and personalized. Through analyzing vast amounts of data, AI will understand customer preferences, enabling companies to deliver personalized services at scale.

4. Enhanced Accessibility:

Generative AI will play a pivotal role in making content and services accessible to all. By leveraging multimodal technologies, AI will enable individuals to interact with content in various formats, such as text, images, and videos. This inclusivity will democratize information and provide equal access to diverse audiences.

5. Prompt Engineering:

Prompt engineering refers to the art of effectively working with generative AI models by properly formatting prompts and maximizing their output. While prompt engineering may be deprioritized in the future, it's still important in guiding AI for specific learning objectives or desired outputs. Prompt engineering is like learning HTML, JavaScript, and CSS to create web pages, highlighting the role of tools and the effectiveness of their usage. Overall, prompt engineering plays a crucial role in optimizing the capabilities and efficiency of generative AI models for content creation and other applications.

Conclusion:

As we look ahead to 2024, generative AI shows immense potential in transforming various aspects of our lives. From personalized customer experiences to automating workflows and revolutionizing content creation, the capabilities of AI continue to expand. However, it is crucial to strike a balance between leveraging AI's capabilities and maintaining content quality. By harnessing the power of generative AI, individuals and organizations can unlock new possibilities and navigate the AI-powered landscape with confidence.

Topics Covered in This Episode

1. Data Usage and Organization in 2024
2. Automation Trends in 2024
3. Multimodal Workflows and Future Interactions
4. Prompt Engineering and its Future
5. Governance and Legal Aspects of AI

Podcast Transcript

Jordan Wilson [00:00:17]:

Generative AI has taken 2023 by storm. But what about 2024? Right. As we enter the Q4 of this year, we're actually gonna not reflect back in today's episode of what's happened so far, but we're actually gonna look forward into what might be coming when we talk about generative AI in the year 2024. I'm extremely excited for today's episode of Everyday AI, and welcome. If it's your first time here, thanks for joining us. So Everyday AI is a daily Livestream podcasts and free daily newsletter helping everyday people like me and you keep up with generative AI, but how we can not just keep up, but how we can Put it to use for us. So if you are joining us live, thank you. Get your comment in.

Jordan Wilson [00:01:04]:

What are you seeing in the generative AI landscape in 2024? If you're listening on the podcast, Make sure to check out the show notes. We always leave links back to sign up for the free daily newsletter and to even join the conversation on LinkedIn. So before we get to that And, yes, please please get in your takes right now. What are your questions?

Daily AI news

But before we get into it, let's take a look at what's going on in the world of AI news. So, first, China's, Baidu is coming for ChatGPT publicly. Right? So China's, new release To its, ErnieBot 4.0 from by, Baidu, which is obviously kind of the, China's equivalent of Google. They dominate, the search market, but, they are just releasing some new details on their ErnieBot 4 point o. Again, kind of its competitor To, GPT 4 from OpenAI.

Jordan Wilson [00:01:54]:

So, it should be interesting to keep an eye on what's happening with Baidu and earning bot 4 point o. You know, pretty pretty much they're saying out loud, yeah, we are competing with the GPT 4 technology. Alright. So New York City is trying to take the lead in responsible AI use. So New York City mayor Eric Adams released his plan for the New York City artificial intelligence action plan, Which is, I believe, the first of its kind for a major US city. So with this, they will also launch the My City business site, Which features a pilot for New York City's 1st citywide AI chatbot. Interesting. Right? Also in this plan, the New York City artificial intelligence action plan.

Jordan Wilson [00:02:35]:

That's a mouthful. It'll develop a framework for city agencies to carefully evaluate AI tools and associated risk, Help city government employees build AI knowledge and skills and also support the responsible implementation of these technologies. So Pretty interesting to see. You know, New York was also kind of ahead of a curve, ahead of the curve in terms of applying, you know, different AI laws into hiring for the city. So, also, it's it's it's gonna be interesting to watch what happens with this NYC, AI action plan. Last piece of news. AI is Powering the economy. I talked about it yesterday, but here we go.

Jordan Wilson [00:03:12]:

We got another study to back us up here. So AI is powering the economy to the tune of $2,400,000,000,000. So a new report from venture capital firm, Accel, showed how much AI-powered companies are actually leading the charge for the US economy. It showed that US tech leaders including, Apple, Microsoft, Alphabet, Amazon, you know, all the all the big players, including, NVIDIA, Saw their market caps grow by $2,400,000,000,000, in a year marked by the rise of gen by generative AI, and obviously, NVIDIA's, chips. So the economy, I think, Right now is largely being powered by generative AI and what that means for our future.

About Josh and JoshCavalier.AI

And speaking of what that means for our future, that's what today's show is all about. You know, talking about where we're going with Gen AI in 2024. So, to help talk about that, I'm excited to have on the show today.

Jordan Wilson [00:04:09]:

Please help me welcome Josh Cavalier, the founder of JoshCavalier.AI. Josh, thank you for joining us.

Josh Cavalier [00:04:14]:

Hey. Good morning, Jordan.

Jordan Wilson [00:04:16]:

Good to be here. Alright. So hey, Josh. Tell tell everyone a little bit about what you do, with Josh cavalier dot AI.

Josh Cavalier [00:04:24]:

Yeah. So currently, I'm offering online courses, workshops, coaching in and around Gen AI, specifically in the l and d area Under the HR umbrella. So that's my flywheel. But I have been doing coaching, with individuals outside of those boundaries, which has been extremely interesting because it's given me some ideas as far as what's happening in the industry. So that's where I'm currently at.

Jordan Wilson [00:04:50]:

Yeah. Love this. And and, again, I'm so so excited for today's episode because it seems like every day On everyday AI, we're talking about things that have already happened. And this is one of those episodes where, you know, talking talking with Josh and I saw this, and I'm like, oh, this is gonna be good. But let's Let's start, Josh. And and, you know, as well, maybe even before we get into the 5 different points. Yeah. But let me just ask you.

Jordan Wilson [00:05:14]:

What are you even personal, Like, personally excited about, you know, in the, Gen AI world, you know, moving forward.

Josh Cavalier [00:05:22]:

Yeah. I think the one thing that drives me is access and accessibility. And what I mean by that is being able to reach more people with less, input. In other words, like, you're gonna go ahead and create content and utilizing AI either through accessibility or captioning or Converting your voice to a different language, you're gonna be able to reach more people. I know that when I create my content, it's very much USA, Europe, Some of India, and that's about it. And in the future, I'll be able to reach the whole world without much lift. And I think that's amazing. And not only that.

Josh Cavalier [00:06:00]:

Think about all the individuals in other countries who, via language barrier, have wonderful ideas. All that's gonna be broken down. So I can't wait to see how that Shows up in 2024.

Jordan Wilson [00:06:12]:

Mhmm. Me as well. Alright. Well, hey. Before we get into the list, which I'm so so excited about, As a reminder, if you're joining us live, get your questions in for Josh. What do you wanna know about the future of Gen AI in 2024? And thank you everyone for joining us. Ben saying he's he's joining us from Dallas as well. Kevin saying good morning from Atlanta.

#1 Data Collection

Jordan Wilson [00:06:30]:

Monica saying the stream looking sharp. Yeah. We had some tech problems this week, so we had to, had How to try to make this a little better. We had to get ready for Josh. You know? Josh Josh is coming in with, the big the big hits Right now, but let's let's start at the top, Josh. What is Yeah. Maybe number 1 on your list of things that we're looking forward to, in Gen AI for 2024?

Josh Cavalier [00:06:50]:

Yeah. I think the big topic for 2024 is gonna be data, and the organization of data and how corporations and even individuals will go about, collecting their data and leveraging it. And I know that, you know, currently, you know, a Gartner projection says that Within corporations, 75% of companies are just sandboxing right now, but they are showing up in 2024. And the big story there Is there data? If we think about, like, OpenAI and how they position ourselves with GPT 4, well, they're leading the charge because they have the best data. They went ahead and siphoned the whole Internet and leveraging it a large language model, and it's showing up. Think about Adobe and FireFly. What what are they leveraging? They're leveraging all of the image data, you know, and all of the information that they have, with Sensei And taking in full advantage of the data. And so I know that with corporations, with consumer data or customer data like point of sale, I think one of the issues that may come up is customer privacy

Jordan Wilson [00:07:58]:

Mhmm.

Josh Cavalier [00:07:58]:

Around data and the balance there. And so we're gonna see as we move forward this aggressiveness of companies to collect the data to leverage AI, but then probably a push back from the consumer. Like, no. This is a little bit too invasive. And I think we're starting to see that right now.

Jordan Wilson [00:08:15]:

Yeah. Josh, the The dork in me and the digital strategist, is is so happy that you said data number 1 because it seems like companies think of this On the back end. Right? They they try to dive into Gen AI, and then they try to figure out data later. What's what's maybe in in your experience one and Common mistake that you see companies making, when it comes to their their data collection or how they even use data when it comes to AI.

Josh Cavalier [00:08:42]:

I think it's just dirty data practices. And what let me clear clarify what that means. I mean, let's say that you're a frontline sales and you're entering things in CRM. Right? You're gonna go ahead and check the box that you entered something in CRM, but everybody is entering the information in differently, or they're using abbreviations, or they're going in just Not even putting information in, but the action is occurring. And so I believe that practice isn't around clean data. Data that is actionable is gonna be the 1st step before even leveraging AI. Because why would you even build a large language model around and Inefficient or wrong data. Mhmm.

Josh Cavalier [00:09:18]:

That's gonna be a huge issue. And, hopefully, nobody spends the money to go ahead and and put forth that effort And then find later on that they can't even leverage the model because the data is just incorrect or outdated.

#2 Automation

Jordan Wilson [00:09:31]:

Yeah. Absolutely. And, you know, we talk so much. You know, when working with large language models and just generative AI in in general, like, oh, you know, garbage in, garbage out. Right? But it's not just Props. It's it's it's literally what you give your model access to. So, I'm so glad that was number 1. Josh, what's what's number 2 on our list?

Josh Cavalier [00:09:53]:

Number 2 for 2024 is automation. And we're beginning to see that right now where, you know, you take applications like Zapier or Make, and you can start seeing instances where individuals are not only leveraging prompt engineering, but they're leveraging additional applications and putting them in a workflow. We're seeing application developers implement generative AI Inside their applications and collapsing down workflows. And so automation of your task, whatever you do day in and day out, It's gonna show up strong in 2024. I bet for myself personally, every day, I'm taking a look at the landscape of Tools that are out there and my my workflows, my personal workflows, and it's impacted every day where I could go in And now leverage tools where I record a video, and now the video is a transcript. And I can take the transcript and put into an AI workflow and get 5 different outputs and put those into my channels. I would take hours and hours to do that in the in the past, and now it just takes minutes.

Jordan Wilson [00:10:59]:

Yeah. And, you know, Josh, I think that, you you know, especially in the marketing and the digital world for so long, automation has been a a you know, I wouldn't even say a hot topic, but it's It's been a requirement, right, for for any any business really to survive. If you're not automating, you you know, you're really falling behind. But when we Talk about, you know, kind of quote, unquote stacking. Right? Because hey. Now we're not just, in encouraging everyone out there to use all these and AI tools and and processes and systems, but we're also saying, hey. In addition to AI, you also gotta keep your automation in there. You can't take it out.

Jordan Wilson [00:11:37]:

What's what's your best, you know, some of your best advice for companies to balance that to, you know, because they might feel a little overwhelmed trying to dive into AI headfirst, but saying, hey. You gotta keep And maybe, you know, improve your your automation as well.

Josh Cavalier [00:11:51]:

The first thing is just documenting what those workflows are, Either at the individual contributor level or if it's external to the customer, what what does that look like? And Once you analyze those specific tasks, you then have an easier go at it when you look across the landscape of tools and AI, accessories and apps that are out there as far as what can we try. And it has to be I mean, it has to be low risk because the The landscape is changing daily in regards to AI tools. And so I would look for automations that are gonna have the biggest impact on top line growth or the bottom line and your own personal productivity. You know, again, you know, you have a lot of individuals out there that are trying to leverage AI. Document your workflows. Be pragmatic. Like, really look at tools out there and try a few different tasks with AI and go, yes. This is gonna save me hours here.

Josh Cavalier [00:12:48]:

Let me go and implement that in my workflow.

Jordan Wilson [00:12:51]:

Yeah. Literally, that that step that you said there, Josh, Seems so simple, but it it's it's also that so many so few people, I believe, are doing it, like documenting their actual workflow. It's something I tell people all the time. People are like, hey, Jordan. Where do I get started with with AI? It's like, alright. Well, where do you spend your most time manually? Right? Like, what's your processes, and sometimes that gets people, scratching scratching their head a little bit. Right? And, hey, just just just as a reminder here, you you know, Thank you, Brian. Brian says, hey.

#3 Personalization

Jordan Wilson [00:13:24]:

For for automation, look at task magic. What's what's all what's your guys' best recommendations for, You know, the 2 things that Josh has talked about so far with data, automation, and just what are what are you all looking forward to in in 2024? So, hey. As a reminder, we have, Josh Cavalier here, the founder of Josh cavalier.com. We're hey. We're or or sorry. Josh cavalier dot ai. So what's what's our number 3 way that Gen AI is gonna show up, in 2024, Josh?

Josh Cavalier [00:13:51]:

Personalization. That is gonna be huge. And I know that a McKinsey report came out and said that 71% of consumers expect a personalized experience. AI is gonna drive that. And I know that I have seen examples in the last 2 months that have just Blown my mind, honestly, in regards to call center activity and how AI is gonna completely automate, call centers to the to the point where somebody on the other line really doesn't even know that it's AI. And I I know you probably have seen some of these demonstrations too, but, you know, it really comes down to having the consumer or having your audience, have the type of experience where it's meaningful to them. An AI is gonna allow us to go ahead and personalize all these consumer based interactions or, you know, what you know, whether you're delivering any form of media, that personalized touch. And and it can can get down to, like, accessibility.

Josh Cavalier [00:14:51]:

Even something that's a barrier, like creating really good closed captions or open captions burned into a video. It's just kinda time consuming, really. But AI is expediting that, and allowing that personalization in there whether you wanna have your audio on or off or anything like that. So, you know, it's personalization of experiences out in the marketplace. It's personalization of media, and I think we're gonna see it show up big in 2024.

Jordan Wilson [00:15:16]:

You know what? I could talk about this one for hours, Josh, because I feel, there's there's 2 sides to the Gen AI coin. One, I think it's gonna help people, in in companies who truly care about the message that they put out there. It's gonna help them do an even better job. But for those companies that maybe didn't care as much or maybe they don't have the resources or maybe if they're just looking for a shortcut, There's another side to that Gen AI coin, which is putting out a lot of mediocre content at scale or a lot of Just kind of, you know, not hot garbage, but medium garbage. That's right. Content at scale. But personalization, I think, is so important because As consumers, I just think that we're gonna be bombarded with more content than ever before because of how easy it is to create, Which makes then, Josh, the the the personalization part key. Right?

Josh Cavalier [00:16:12]:

Yeah.

Jordan Wilson [00:16:13]:

How how would you recommend?

Josh Cavalier [00:16:14]:

Yes. So

Jordan Wilson [00:16:16]:

A a tool or process or what?

Josh Cavalier [00:16:17]:

Yeah. Noise and the con you know, content creation that's garbage. There are gonna be amplifications of content, AI generated content, because people understand how to be Storytellers. They understand their craft, and they're gonna go ahead and separate themselves from individuals that don't understand how a channel works or how to get a a certain piece of content out. And so I I think it really comes down to individuals that already have these skills of storytelling and putting out great forms of media, leveraging AI, they are going to stand out, in this, You know, this arena where there's just tons and tons of content. And I, you know, I think, really, the crowd decides, hey. Who's gonna be up on top either through likes Or, you know, pushing or driving a video forward in a in a channel. Yeah.

Josh Cavalier [00:17:11]:

So even though AI is gonna generate a lot of noise, I still think you're gonna have and Exemplar individuals out there creating amazing content that's still gonna resonate.

Jordan Wilson [00:17:20]:

Yeah. I think being able to resonate and understand your client, I think, is gonna become more important than ever before. You know, I think in years past, it was more of, you know, getting, like, categories or buckets of, you you know, your, ICP or, you you know, like, whatever, acronym you wanna throw out there, but you say, oh, here's our 5, you know, customer types. Let's go after them. I think now you're gonna have to say, here's our 50 customer types because you have to be that much more specific, and and more personalized. Right? Like like Kevin here saying personalization, totally agree. How about personalization of medicine? I I don't know about that. Like so here.

#4 Multimodal

Jordan Wilson [00:18:04]:

We've we almost every day in the newsletter, there's something, some new news in the, medical space, about, yes, personalized medicine going to the, even to the DNA level. So it's extremely extremely usenating, fascinating use case For sure. Alright. Let's let's get to, next on the list, Josh. What is our number 4 thing to look for, with how generative AI is gonna show up in 2024?

Josh Cavalier [00:18:31]:

Number 4 is multimodal. And so what I mean by that is we just saw GPT 4 vision come online and and accessible hope to the masses. What does that look like? We're gonna see different kinds of workflows In all aspects of work with multimodal. I know that for myself. The first prompt that I used in image was a slips and falls image. It was a picture of a mop and a mop bucket and but just back behind a door, There was 2 feet sticking out. Obviously, somebody just fell. Right? I upload this image, and it was shocking.

Josh Cavalier [00:19:10]:

It, like, went ahead and it knew exactly what was happening, and it knew that somebody fell. And then I used that to go ahead and create, some, you know, multiple choice questions and a video script and all kinds of different outputs around that specific event. Think about Individuals walking around with a phone, they see something happening in their environment, their work environment, they take a picture of it, and they're like, am I really seeing this? Like, should I be about this, and AI is gonna be right there to assist them and give them some background as far as a situation, a piece of equipment, anything like that. We haven't even thought about those workflows yet and how that shows up and how, you know, companies can leverage that technology. And so I think with multimodal, just not images. Right? You know? So we're talking about being able to talk to ChatGPT. We're we're talking about uploading videos or Taking text and having it output to a video. So you have all of these various multimodal workflows that are gonna show up, that are going to reveal new methods of work, new methods that we can show up for our customers, and it's gonna be just and Eye opening because you're gonna have you're this is gonna be this massive period of experimentation that's gonna happen that's gonna reveal all these new ways that we can communicate And that we can interact with each other.

Jordan Wilson [00:20:29]:

Yeah. You know, doctor doctor Javier Castro is saying he's been calm in 2024 multimodal. Yeah. I think a lot of people Even since November, right, when chat GPT was first introduced, you know, to the masses. I think people are like, okay. This is cool. You know? Inputting text and getting text on on the back end is great, but what about inputting photo, video, sound? What about Yeah. On the output, Photo, video, sound.

Jordan Wilson [00:20:55]:

You know, even, you know, Josh, I'd love your personal take because, you know, you talked about, OpenAI and chat and just released their kind of, you know, quote, unquote vision, functionality over the last couple of weeks. But then you've had other companies, you you know, Microsoft, BingChat Has had certain functionalities. Google BARD has had certain, you know, kind of, multimodal functionalities. How how are we gonna See it all come together, or or what is that magic recipe of getting the multimodel right? Because, you know, it's been out there now in the wild for many months, but it still doesn't seem like it's just, hitting just yet.

Josh Cavalier [00:21:33]:

Right. I I think the one issue is hardware. It's scaling. And so, you know, we have this period of time where everybody wants the NVIDIA cards, and people are trying to create tensor chips, and there's all this activity happening in the background. You We still won't realize the full potential of multimodal at scale until we have hardware at scale that can support all these various types outputs, and I think that's the reason why you start you you see, like, the appetizers of multimodal out there. It's not like the full course dinner. It's so like, it's it's a tease, and I can kind of premeditate how it's gonna show up, but I still yet, Like, for instance, just in GPT 4, I can't use a plug in with an image upload and, you know, and other multimodal type functionality. It's just kind of like cherry picking.

Josh Cavalier [00:22:25]:

Like, what what kind of option am I gonna use on this prompt?

Jordan Wilson [00:22:28]:

And

Josh Cavalier [00:22:28]:

so I think we're gonna see a maturation occur Once the hardware and the server farms are set up in such a way that can actually handle the true output types that we want.

Jordan Wilson [00:22:40]:

Yeah. You know, you bring up a good point. And even if if if we're just talking about chat gbt, like, one thing I would personally love to see, Josh is like, yeah. We have these great, kind of multimodal options now within ChatGPT. You know? You have your your DALL E 3, insight. Yeah. And then you have your ability to, you know, kind of use, you know, your voice and and to upload, you know, photos, But they're all in separate modes. Right? So I keep thinking like, oh, is this right now as we have it truly multimodal? It's like, I'd say probably not because I you know, in in my opinion, one of the best use cases of chat gpt, and I talk about it all the and All the time.

Jordan Wilson [00:23:21]:

One of the most untapped things right now in the business world is using ChatGPT with plug ins. But at least right now, you Can't use DALL E 3 in the in the plug ins mode, and you can't use the kind of the new vision feature in the plug in mode, either. What's your take on that? Is that, Like, can you see that as a big, as a big miss or, you know, maybe it's just, more of a feasibility and and software engineering feat that would require more?

Josh Cavalier [00:23:46]:

Yeah. So, I I mean, I think I think that aspect of it needs to be worked out for sure. Like, you know, as far as, like, from an engineering standpoint to get all that and word away so that you can expedite the content creation process. But I think in the in the meantime, what we're gonna see, and I was talking about this where you before we got on the show, is A good example of how multimodal shows up is in the tools that we use day in and day out, like PowerPoint or, I was, You know, the example I was telling you about was Adobe Captivate, and I saw an amazing demonstration at a conference a few weeks ago where, they have a tool called Adobe Captivate where you can go ahead and just simply prompt in a topic and some learning objectives, and it builds the entire course out for you, perfectly laid out. But here's the thing. They're also leveraging Adobe Firefly to generate The perfect image for that content. That's true multimodal. Like, we can go in and we can orchestrate, you know, individual bits of content from a multimodal standpoint, but it's when we have applications that leverage true multimodal experiences And allow us to go ahead and create content at scale that's really good content, man, that is a massive game changer, because we're just like we're playing in the minor leagues right now.

Josh Cavalier [00:25:03]:

What I saw was Major League, and I think that we are on the cusp of a period where an individual is fully empowered To create the content that they envision, but they never had the skills to do it. And so I think that is the that's the true, look of what multimodal is going to be and how

Jordan Wilson [00:25:22]:

it shows up for us. Such such a good point. Yeah. Josh Josh has some great insights, You know, with Adobe, he had some some things going on with them. And, hey, as a reminder, we've hit this list very hard. And if if if you maybe join in the middle Or if you can't keep up, maybe you're you're also, you know, drinking your morning coffee, doing the dishes, whatever. Don't worry. Go to your everyday ai.com.

Jordan Wilson [00:25:43]:

Sign up for that free daily newsletter. We're gonna be Literally hitting these things point by point in the newsletter. So don't worry if you didn't catch up because we are now at our final one. And just to quickly recap, Josh, so we had number 1, data. So these these are the 5 ways Gen AI is gonna show up in 2024. So number 1 is data. Number 2 is automation. Number 3 is personalization.

#5 Prompt Engineering

Jordan Wilson [00:26:04]:

Number 4, which I just took us on a side rant on, is multimodal. And then what is our last? What is that number 5, Josh?

Josh Cavalier [00:26:10]:

Number 5 is prompt engineering. And let me just say, prompt engineering is dead. Long live prompt engineering. So it's been fascinating because I know that with my peer group, I have a a lot of peers that say you won't need to understand how to prompt as you move forward. And I think there's some truth to that. You know, the more I get into it and the more I see what's happening with applications, there is a balance that's going to occur As we work with AI, you're gonna get guidance in regards to prompts, and I see it right now in Adobe Firefly. I see it happening now when you do a DALL E image, you might enter in 5 words, but behind it there is a prompt that has 20 words. It's gonna guide how the image looks.

Josh Cavalier [00:26:52]:

I get it. But going back to point number 2 of automation, if you truly want to have agency, if you truly want to have autonomy, Prompt engineering is still gonna be in play in 2024 because these models change. And if you want to have, Again, empowerment of your outputs, empowerment of your automation, you still will need to understand how to prompt engineer. Like, you'll need to understand, of the model itself and then based upon the words that you give it, the output types that are coming back. And if you don't understand that, well, you lose agency. You you are now the tail's wagging the dog at that point. You are asking the vendor, Whoever's creating this tool to basically do the prompt for you. And I see this with prompt as a service.

Josh Cavalier [00:27:43]:

Right? There's a lot of tools out there that are prompt as a service, And you're leaning on them for the prompt. Now some tools I use have prompt as a service, but they allow you to use your own prompts. So when I do a video, it gets transcribed. I can then leverage my own prompt to go ahead and create the educational output types that I specifically need, and Very, very detailed. Saving me tons of time. It's these other tools that are out there where you're leveraging on their prompt, and it's just Okay. Right? So I think prompt engineering is gonna be important in 2024, but it's just gonna show up different for everybody. It depends upon, you know, your, how you're comfortable with technology and whether you wanna go down that road or not.

Jordan Wilson [00:28:28]:

Yeah. It's, and let me let me even hit rewind because I think some of our audience, Josh, is is, you know, following us. Some at this one might be scratching their heads, so I'll maybe oversimplify it here. But, you know, when people ask what is prompt engineering, You know, there's there's very technical answers to that, or you can just say it's learning how to properly work With any generative AI model. Right? So it's it's knowing how to go back and forth, how you should be formatting, your prompts, and and I would say, you know, maybe getting Getting more out of the same or getting more out of less, I would say, is the most oversimplified way to talk about prompt engineering. But, Josh, I actually have a hot take. And and and and it goes along with part of of your, you know, prompt engineering is dead. Long live prompt engineering, number 5.

Jordan Wilson [00:29:17]:

But my hot take is, I think prompt engineering is is going to become deprioritized, actually, which which might not make sense. But In in some recent, experiences I've had, you know, even as an example, working with, you know, MidJourney Back when it was, you know, the version 4 point something versus now when you're working with mid journey 5 point whatever, much fewer words give you fantastic results where it didn't used to be like, it it it didn't used to be that way. And then if anyone has used, DALL E 3 inside of chat g p t, When you give it a prompt so, you you know, DALL E is the new AI image generating model within, OpenAI's chat gbt. When you give it a short text prompt, It actually expands it in 4 different ways. Right. So our our our model is gonna be taking like like this, are the models gonna be taking care of, You know, quote, unquote, traditional prompt engineering for us?

Josh Cavalier [00:30:17]:

Well, eventually. But if I'm trying to Go ahead and create learning objectives or multiple choice questions or very, very specific outputs, not generic, not like common outputs like an image. But if I'm if I'm driving towards very, very specific type outputs, you know, like a report, Like a 20 page report. Right? I don't think a a a simple 5 word prompt is gonna go ahead and guide an AI for a 20 page report. So I think, you know, getting back to the point of, you know, it depends upon how you wanna use the tool. I think well, here's an analogy. So when the Internet came out, you had a choice. Were you gonna learn HTML, JavaScript, and CSS? Some people said no, and that's fine because you can go ahead and use tools that will automatically and create web pages for you.

Josh Cavalier [00:31:03]:

But for those individuals that learn those languages, now they have agency. Now they can go ahead and leverage, You know, being able to to create modifications of their website or whatever the case may be. Or for my in my area, modifying or putting Special elearning content together because they know how to how to program. So it just depends upon your level of comfort. Again, there are gonna be individuals at no prompt engineering that would be able to Take work and look at it differently and do different problem solving. But, you know, for those of you that don't wanna do prompt engineering, you're gonna be fine. I mean, you're still gonna be able to increase your productivity and show up using AI. It just depends upon what level you wanna take it.

Audience Questions

Jordan Wilson [00:31:44]:

Yeah. Josh, great point there. Yeah. I guess the more complex or, you know, the more, outputs that you may be wanting from the generated AI model, You probably still will need a certain level. You you know? Alright. So we do have a couple questions here. We got to our top five. So a question coming in is is how can AI return on investment of of learning programs.

Jordan Wilson [00:32:06]:

Or maybe I think this is from earlier when we were talking about, you know, using, you know, generative AI to create, you know, courses kind of like, you you know, with, the Adobe platform. But we're or or maybe in general, how can organizations Know the return on investment of whatever they're putting into Generative AI.

Josh Cavalier [00:32:24]:

Yeah. They gotta they gotta keep track of, like, Task, you know, doing task analysis and figuring out how long does it take to actually produce some certain type of output. I mean, if you're not tracking, you're just guessing at that point. And so it's gonna take work. This isn't gonna be easy. You know? You can't just do anecdotal, like, oh, we're working faster. Like, well, what does that mean? And so, you know, I think there's gonna be a period of time where when you analyze your work when you actually map your workflows out, how much time does that one past take in Canva or, you know, PowerPoint or whatever the case may be. Document it.

Jordan Wilson [00:32:59]:

Mhmm.

Josh Cavalier [00:32:59]:

And then now when you use your AI tool, How how much faster is that task? And, obviously, you know, based upon productivity and what an individual contributor is putting out Over the course of the week or on whole entire Salesforce, like, is that showing up different each week because we're using AI automation to contact 20 additional customers a certain type of way, You gotta track, and you and you have to be really intentional as far as what those data points are because there's, There can you can be tracking all kinds of different things, but what are the inflection points? Like, what's gonna truly impact the KPI? And then are you tracking against that as you are Going in and implementing AI.

Jordan Wilson [00:33:39]:

Yeah. Alright. I think I think we have time for 1 or 2 more quick ones here. So Craig joining us. Craig, thank you for the question. Saying, do you think there's gonna be some sort of, governance put on AI tools via the government or industry leaders at some point? Just a great question. So so maybe, Josh, You know, you and I are both based in the US, so maybe let's answer this, you know, out of the scope of the US because I think it'll be different. But, yeah, will there be governance, from from the US on these different, generated AI systems?

Josh Cavalier [00:34:06]:

There will be, but I'm trying to think of how to say this. You you There's a lot of influence in government.

Jordan Wilson [00:34:16]:

Mhmm.

Josh Cavalier [00:34:16]:

And so we we have a lot of active players that are currently in government that are influencing government. And so I I think there will be governance, but it's gonna be somewhat loose or open to where there's gonna be growth with AI, and we don't know what that means on the other side. Right. And so we still have we still have things that are going through the courts in regards to copyright. We still have, issues of nefarious characters using AI in really, really bad ways that need to get clamped down on immediately. And so I think there will be additional guardrails out there in the marketplace, but I think in the US, it's gonna be somewhat loose. At least that's my take on it. From what I understand what's happening in regards, you know, like, to the federal government and who is, you know, Working there, I I think it's gonna show up, but I think it's gonna be somewhat loose.

Jordan Wilson [00:35:15]:

Yeah. Yeah. I agree, Josh. And to answer your question, Craig, I'll be a little more bold. I was, You know, previously, political journalist. I don't think anything's gonna happen. I don't think anything's gonna happen right now. You know, I talked about at the top of the show.

Jordan Wilson [00:35:28]:

Right now, generative AI is is too important to the American economy to thwart it at all. So I I think there should be Way more guardrails. There should be some sort of, legislation regulation, but, unfortunately, I think, some of the big some of the decision makers Don't fully understand, and I think right now, it's too important, to the economy that lawmakers are gonna do it. Alright, Josh. We've been all over the place. We took some questions. We went over The 5 different ways generative AI tools are gonna show up in 2024. But maybe if you, peak someone's interest and and and someone out there now is like, okay.

Jordan Wilson [00:36:04]:

Yeah. Now I care about this. What is what is your one message or your one takeaway, that that you could advise companies? Maybe they weren't paying much attention to Gen a now. They've listened, and now they are. And they're like, alright. I gotta get on this. What is your one piece of advice or one takeaway for those companies on how they can actually Heading into 2024, make sense and make, generative AI work for them.

Josh Cavalier [00:36:28]:

Yeah. 1st and foremost, if you don't have an AI policy, Please implement 1 because that is gonna dictate how your company and how your associates show up. So, I mean, if you're if you're still here in October And you don't have an AI policy, get that in place. Once you have that in place, then go ahead and allow your leaders To go ahead and drive those individual conversations in the lines of business as far as how AI shows up. They know their craft. Right. And they understand what their job is as far as productivity or whatever they're, you know, driving towards, And they're gonna go ahead and give you guidance as far as, like, hey. AI is gonna show up this way in, accounting.

Josh Cavalier [00:37:09]:

AI is gonna show up this way in IT, whatever the case may be. But it's a discussion. It's a dialogue, and it's ongoing. So make sure that you're doing the hackathons and all these things to, you know, to test AI and do the work as far as, like, the cost. Also, you know, whether you're implementing Microsoft Copilot Or you're trying to you think that you may have enough great data to build your own large language model, what what what's the cost of that? The maintenance of it. Yeah. Right? What if we build whole functions around it? What does that look like? So, I think the policy is first, empowering your individual contributors to start using it and then have that converse have that conversation, the ongoing conversation.

Jordan Wilson [00:37:57]:

Such great advice. So many, great insights, in in tips and tricks, from Josh. So, again, Josh Cavalier, founder of Josh Cavalier dotai. Josh, thank you so much for joining us on the Everyday AI Show.

Josh Cavalier [00:38:11]:

Thanks, Jordan. I really appreciate it.

Jordan Wilson [00:38:13]:

Alright. And, hey, everyone. As a reminder, we did cover A lot. Don't worry if you didn't miss it. Just go to your everyday AI .com. Sign up for the free daily newsletter. We're gonna have be sharing more about, and different things that Josh is is working on, with Josh cavalier dot AI and a whole lot more. So thanks for joining us, and we hope to see you back for another edition of everyday AI.

Jordan Wilson [00:38:33]:

Thanks y'all.