Everyday AI Podcast – An AI and ChatGPT Podcast - From AI Agent Orchestrators to Deepfakes. The New Tech Order

Episode Date: December 16, 2025

We don't do AI predictions lightly. Everyday AI is trusted by millions each year to help guide them through the muddy AI waters. So in the same way you'd want transparency out of your AI models, we're rolling back the clock on our January 2025 AI predictions we dished with our Roadmap Review. We're busting out the receipts. At the time, these AI predictions seemed nutty. So, did we lead you astray? Or, did we pave the road to paydirt? Let's dive in y'all. 2025 AI Roadmap Rewind: From AI Agent Orchestrators to Deepfakes. The New Tech Order -- An Everyday AI Chat with Jordan WilsonNewsletter: Sign up for our free daily newsletterMore on this Episode: Episode PageJoin the discussion:Thoughts on this? Join the convo and connect with other AI leaders on LinkedIn.Upcoming Episodes: Check out the upcoming Everyday AI Livestream lineupWebsite: YourEverydayAI.comEmail The Show: [e-mail] with Jordan on LinkedInTopics Covered in This Episode:AI Agent Orchestrators as Job TitleAI Agents in Company Hiring TrendsEnterprise Reasoning Data Collection GrowthAI Driving Professional Services Pricing CrisisUniversal Basic Income and AI Job LossOpen Source AI Models Surpassing ProprietaryChinese AI Model Global Market ImpactPerplexity Answers Engine Business PivotFrontier AI API Price DropsVC Funding Surge in Embodied AIAdvancements in AI Video Generation ToolsAI’s Disruption of Traditional Internet ModelsSocial Media Deepfake Misinformation SurgeTimestamps:00:00 "2025 AI Predictions Explained"04:02 "2025 AI Predictions Insights"08:49 AI Agents Classified as Employees11:23 "Transformers vs. Reasoning Models"16:15 AI Impact on Consulting Jobs18:51 "AI Impacts Jobs, Spurs UBI"23:12 Open Models Surge in AI27:13 "Perplexity's AI Pivot Journey"29:23 "Embodied AI Sector Growth"31:36 "Embodied AI Transforms Logistics"36:42 "AI-Driven Internet Future"39:16 Deepfake Fraud Crisis Escalates42:25 AI Success Roadmap RevealedKeywords:Agent orchestrators, AI agent orchestration, AI job titles, orchestration engineer, AI agent architect, human-AI collaboration leader, Adobe agent orchestrator, AO, AI agents, digital FTEs, digital full time empSend Everyday AI and Jordan a text message. (We can't reply back unless you leave contact info) Start Here ▶️Not sure where to start when it comes to AI? Start with our Start Here Series. You can listen to the first drop -- Episode 691 -- or get free access to our Inner Cricle community and all episodes: StartHereSeries.com Also, here's a link to the entire series on a Spotify playlist. 

This is the Everyday AI Show, the everyday podcast where we simplify AI and bring its power to your fingertips. Listen daily for practical advice to boost your career, business, and everyday life. Meet Firefly AI Assistant, now live in Adobe Firefly, the All In One Creative AI Studio. Just describe what you want to create and the assistant handles the rest, orchestrating multi-step workflows across Photoshop, Premiere Express, and more in one conversational interface. You direct the outcome. The assistant accelerates execution. A lot can change in a year.

Think about it in your own personal or professional life. How much has it changed in 2025? Probably a lot, but probably not as much as AI. Because when we're talking about AI, time is like warp speed. I mean, a year in AI time is like seven dog years. So that's like 49 real life years. So when I look back at the 2025 AI predictions show that I made back in January of this year, it feels like a legit lifetime ago because that's how fast AI moves and how much the landscape shifts

almost on a daily basis. So here we are 11.5 months later. And it's time to revisit those AI predictions that I left you with. And over the five-part series that I did back in January, I'd say I was actually a little bold. And I didn't just call my AI predictions predictions. I actually called them the AI predictions and roadmap series. That's because as absolutely insanely crazy as many of those AI predictions were at the time, I knew that many of them would probably somehow eventually come true.

and I wanted you to tune in, hear what I had to say, and make decisions that would hopefully grow your business. And I'm one, well, with the daily podcast, it's, I have to hold myself accountable, right? Yeah, you can say this is a fun, informal thing and it is, but, you know, I'm lucky enough to be able to reach millions of people a year. And I take that responsibility very seriously. So we're checking back in today. I'm going to hold myself accountable. We're going to bust out receipts longer than if you bought every item at CVS. So between today and tomorrow, we are hitting a rewind on our 2025 AI roadmap series.

So did we get it right? Did we lead you astray? Or were we spot on? And maybe you took advantage and grew your company or your career. All right. Let's jump into it. Let's get recapping. I'm excited for today's show.

I hope you are too. What's going on? If you're new here, my name's Jordan Wilson and this is everyday AI. Yeah, we do it every day. It's your daily unedited, unscripted, live streamed podcast and free dealing newsleter helping everyday business like business leaders like you and me keep up with the whirlwind of changes because it's nonstop and

help us hopefully make sense of it to grow our companies and careers. So make sure if you haven't already, please go to our website, your everyday AI.com. If you miss any of the half of the 25 of the AI predictions, we're going to be going over to the day. We're going to be recapping it all in our newsletter as well as the news from today. So what you need today to be the smartest person in AI at your company. So what the head went into these predictions from January 2025? Well, a lot. I did this the year before as well.

But essentially, if you are brand new here, I'm lucky, right? I'm very lucky. I get to talk to a ton of really smart people, you know, from CEOs of Fortune 100 companies to people building actual AI models at all the big tech companies to, you know, small business CEOs, you know, who are trying to better understand the AI landscape themselves. So each year I talk to. to hundreds of people. And that's really what went into the 2025 AI predictions.

Just a lot of conversations, both ones that you all maybe listen to and those that weren't even recorded. You know, I'm very blessed to be in a situation to learn from a lot of smart people. And you know, that's why I sometimes call everyday AI a cheat code, right? You don't have to spend all the hours that I do. I just give you the bullet points, the hopefully golden nuggets. that can help you make sense of it. So a lot went into our 2025 AI predictions in 2025,

not just me randomly throwing some AI spaghetti at the wall and see what's stuck. And if you do want to go back and listen to all of those episodes, they're all on our website, your EverydayAI.com, or if you're listening on Spotify, Apple Podcasts, etc, go find episodes 443 to 447. You can listen to them all there. They're all pretty short. It was five. So we broke down five predictions a day.

over the course of five days in all of these shows for about 25 minutes. So today and tomorrow, we're going to be tackling about, you know, I think we're going to do 12, 12 or 13 today. I actually forgot. We'll find out here. And then the other half tomorrow. So without further ado, let's get straight into it. This is going to be a faster show, FYI, because like I said, if you want to go listen to the whole thing, go back and listen to the episodes. All right. So we're to start with number 25. This is how we did it. In January, we went 25 through one. So in part one and part two, we're going the same thing, starting from prediction 25 and going backwards. So number 25 in January 2025 was agent orchestrators will be a growing position. And just FYI,

I'm probably, I'm going to try not to say this every single prediction, but think back. Think back to January 2025. This is when the main AI models, were GPT-40 and Claude 35 sonnet in Gemini 2. Not Gemini 3, not Gemini 2.5, Gemini 2, right? I mean, we had entire AI product lines that hadn't even yet been announced and have already been killed off, right? Open AI's operator hadn't even been announced yet. And it's been announced and essentially now killed off and replaced. Right.

So keep that in mind when I'm talking about these predictions because when I talk about agent orchestrators, right? This is the first prediction here. My prediction was that it's going to be an actual growing position and that you're going to have people at companies who are going to be essentially agent orchestrators overseeing dozens of specialized agents. Sounds safe now, right? At the time, I literally couldn't find anything on the internet about agent orchestrators, right? This was a crazy prediction. But go back to today, not crazy at all. This one crushed it.

Absolutely nailed this one, right? So agent orchestration is now a common enterprise product and skill, especially in IT and ops. There was a recent MIT Sloan report that said service, so kind of mentioning some big name companies that actually have positions in agent orchestration. So service now has positioning now. around AI agent orchestration. The Washington Post reported in October that Walmart, KPMG, and Salesforce are debuting new AI job titles, including orchestration engineer, AI agent architect, and human AI collaboration leader. So, you know, different ways to say ancient orchestration. And Adobe literally has jobs for Adobe agent orchestrator.

And they call it aO, just like I called it AO back in January before. called it that. Just FYI. All right. Not saying, you know, I started the craze of agent orchestration. I didn't. Right. But I was talking about it back in 2024, actually, which is one of the reasons why it made it into the 2025 AI prediction and roadmap series. All right. So we'll say we did pretty okay. And by that, yeah, knocked that one out of the park. All right. Number 24, companies will post jobs for AI agents. So my actual prediction here, was at least one public company will post a job where the candidate is actually an AI agent. So a company is going to be posting jobs, not for humans.

So this one got it pretty much right, except it wasn't a publicly traded company, but there was plenty of instances where actual real companies posted jobs that were not for humans. Also, some recent reports have surfaced that AWS and Klarna have reclassified agent compute costs as digital FTEs or digital full-time employees, right? So shifting it from OPX to CAPX. So literally redefining current roles within their organization and essentially now saying that they have AI agents that have full-time jobs. So this is a trend here because enterprises are effectively hiring agents for procurement and scheduling roles, treating them as labor assets in their quarterly reporting.

So although, you know, we don't have public companies now posting, you know, jobs for AI agents, in a way at least on their reporting, they are kind of retroactively going back and reformatting some jobs and some roles as such. We did see, though, Salesforce agent force allowed companies to hire agents via subscription. So, you know, kind of reverse, right? Kind of like a hiring, like you might go and hire a, a specialist on, you know, Upwork or something like that. Salesforce is agent force.

You can go hire agents as if you were hiring employees. So kind of in a roundabout way, got that one there. But there's also, I mean, for smaller companies, non-public companies, it was endless, right? One of the more popular ones, this one didn't take long to hit, by the way. This was in February. But a startup called Firecrawl went viral after they listed a job that was not meant for humans. And it specifically only asked for AI agents to apply.

So yeah, the cool part there is I believe it wasn't like humans were not even supposed to meant to apply their agent for the job. It was only supposed to be agents discovering this and then applying for the job in that way. So we saw actually quite a few of these postings from smaller companies. But like I said, with the Salesforce agent force and some of the larger companies kind of reclassifying what they consider, you know, wasn't too far off on this one. All right. Number 23, company reasoning data collection. So would become a focus.

So the prediction here was that companies would start collecting data specifically to support or shape reasoning models. Like, the best way to think about this is, even though it's technically not 100% accurate, right? Like sometimes for our non-technical audience, I like to say this is 99% accurate. And, you know, hopefully it'll appeal to 99% of the population, right? The way I like to think of it is you have these old school transformer models, right?

And they worked great when you put your company's structured data in there, right? Reasoning models, so the newer breed of models, these are models that break down complex prompts and they think about it and they, you know, agentic models will go and, you know, research different paths and they'll, you know, call different tools and they'll go back, you know, from step five, back to step three and, you know, go, you know, go down Route C instead of Route B, right? These models, yeah, you can feed them structured data.

you can feed them X's and O's and spreadsheets, but what really helps them is your company's IP, your company's decision-making, right? Your company's unstructured data. So that's kind of the crux of what this prediction was. So what happened? Where are the receipts? Well, in OpenAI's state of enterprise AI plan that was just,

or sorry, a state of Enterprise AI, 2025 report that was just released last week, they did reveal that the average reasoning token consumption per organization increased 320 fold in 12 months. So companies are burning reasoning tokens like probably no one could have guessed. And Red Hat. All right. So one of the big leaders in the space, Red Hat did detail how enterprises right now are using

synthetic data generation to convert proprietary data into training examples for private reasoning engines. All right. So this one might still be a little ahead of the curve, right? So if you are listening to this and you've heard me talk before about company reasoning data, I think I had an acronym for it. I'll have to go in the newsletter and remember what that acronym was because I haven't talked about this in a couple of months. If you're listening to this, you need to start doing this. I'm probably going to have maybe an entire show or a series of shows just. us on this, it is that important. You know, I think finally the general public is actually using

these thinking or reasoning models more, where I just assumed that when they came out, right, in early 20, 25, you know, when they became publicly available, I assumed every single person would use them because they're better. But I was wrong. Humans are impatient. Even smart humans are impatient. And they're using inferior models when they shouldn't be. So this is, I still think, a huge area of growth, whether you're an AI builder out there or leading AI strategy implementation at your company, you have to start collecting, cleaning, curating, and categorizing your company's reasoning, right? It's huge.

All right. Number 2020, or sorry, number 22 from our January 2025 prediction series, high-end professional services will go through a pricing crisis. So I said AI will drive down prices and reshape billing in law, consulting, and accounting. I'd say definitely got this one right, but it is going to be, I think, an even bigger wave happening here over the next quarter or two. Because if I'm being honest, this one really didn't start to, quote unquote, hit home until the last two months, right? Some of these AI predictions and roadmaps hit instantly. this one didn't start to hit until literally probably quarter four, if I'm being honest.

So here's the reality. There's a recent S4 capital announcement this summer that explicitly talked about the transition to output-based pricing away from billable hours. All right. Thompson Reuters report in April said that professional services pricing is going through a recognition that new technologies may impact the mix of hourly work versus flat fees. Then you had the more recent McKinsey Professional Services report that said, professional service forms are firms are experiencing quote unquote.

This is theirs. Fundamental pricing model disruption. Yeah, that's what I said. Pricing crisis, fundamental pricing model disruption. Those are synonyms. So up to 40% cost reductions are possible for early AI adopters. So yeah, it's happening. And it's happening now.

It hasn't been felt across, you know, all of these industries, right? Consulting, accounting law, but it's starting to trickle and it is going to hit even harder in 2026. Let me just be honest. Consultancies are going to get hit very hard. We've already seen significant job loss at the not just the big four, but even the tier after that. You know, essentially they're cutting down, you know, intentionally firing or laying, off people that are not going to be in roles in the future that will be able to benefit or use AI, right?

We're seeing tens of thousands of these cuts at a time. And then at the same token, right? Adobe just introduced an entirely new way to create, bringing the power and precision of its creative suite into one conversational experience. Meet Firefly AI Assistant now live in the Adobe Firefly app, the all-in-one creative AI Studio. Powered by Adobe's creative agent, Firefly AI assistant lets you start with your vision, just describe what you want, and shape the outcome as it takes form with the assistant. The assistant orchestrates multi-step workflows, drawing on 60 plus pro-grade tools across

Adobe Creative Cloud apps, including Photoshop, Illustrator, Premiere, Lightroom Express, and more to help bring your ideas to life. You can also get started with creative skills, a growing library of pre-built workflows for common creative tasks like batch editing photos, creating mood boards, portrait retouching, and creating social variations. Every step the assistant takes is visible so you can refine, redirect, or take over at any time. You stay in the driver's seat as the creative director. Adobe Firefly AI assistant now in public beta. See it today at firefly.adopi.com. In the past week, Accenture has signed deals with both OpenAI. and entropic to train tens of thousands of their clients on using these pieces of software,

which is like about time. I've been saying this since I started the show that these consultancies are going to have to start using the off-the-shelf products. First, they're like, oh, we're not going to touch it. These models are dumb. And then they're like, oh, wait, the models are smarter than us. I think that kind of started to come around, you know, during the reasoning models in early 2025, the deep research products.

And they're like, oh, maybe we should. should pay attention, but no, instead, let's build our own models. And they're like, oh, no, yeah, we just need to use Chad GBT, Claude, Gemini, et cetera. So also, a story that just came out a couple of weeks ago, headline in Business Insider, AI is reshaping how McKinsey makes money. So yeah, the big four consulting companies, I think you're going to see a lot of stories in quarter one about they're either having to switch away from billable hours to output, to base price. or they're just going to be bleeding, right?

Yeah, unfortunately, I think when we talk about AI's impact on jobs, I think it's one of the industries that's going to be hit first, right? Especially, and this is a small thing, y'all, keep an eye on this in 2026. The default-based models now are good at just what consultants do, right? What do consultants do? They look at, you know, they crunch a lot of data from public reports. They find connections between different data points. they identify trends and then for the most part they then create spreadsheets that show those things

and then create powerpoints guess what now claude uh especially uh the new opus 4-5 with skills can create better spreadsheets and powerpoints than 99% of consultants i said it i'm probably going to get some hate mail uh right and gpt52 uh their new model not far behind and it does it 10 times faster and probably better than I'll say 99% of juniors, right? You know, not your senior consultants, your partners that have been there for 30 years. But your average entry-level worker, and that's also why at consultancies, hiring for new grads is down more than 40% over the past two years for this exact reason, because consultants, the big consultancies are like, wait, as they're starting to see

these reasoning models, deep research, you know, one, click, you know, mini rag. They're like, wait, what are we doing? Right. And when you can see, you know, I've talked to plenty of consultants over the past couple of years who will honestly tell me off the record, yeah, all I do is spreadsheets and PowerPoints. And now these models are just better. Sorry. So number 21, prediction here was that UBI or universal basic income becomes a mainstream kitchen table topic, even if it's not. enacted. So this one might have been a little bit more of a miss. However, there is a current, I forgot to put it on my slide here. There is a current proposal in the US House for UBI.

And one of the reasons for it is because of AI job displacement. So this wasn't a complete miss. Obviously, that bill's not going to go anywhere. It doesn't have legs, but an actual bill was introduced in the U.S. Congress for UBI and one of the main reasons, you know, a universal basic income. And one of the main reasons was AI job displacement. Not one of the main reasons, but one of the reasons cited. Also here in Illinois, here in Chicago, you know, there's an article that I just read. There was kind of a pilot program here, but let me just read the first, you know, a couple paragraphs from this article. So it says, as artificial intelligence, it accelerates automation across industries, policymakers are grappling with the economic consequences.

AI-driven systems are replacing not only repetitive tasks, but also increasingly performing cognitive work, including decision-making analysis and creative functions. The shift threatens traditional employment models, raising urgent questions about income security in a future where machines may out-compete humans in most jobs. Universal Basic Income, UBI, has been proposed as a solution, a policy that provides unconditional cash payments to all individuals to mitigate job displacement caused by automation. Tech leaders like XAIs Elon Musk and OpenAI's Sam Altman have publicly endorsed UBI predicting that widespread automation will make such policies inevitable. In Illinois, Cook County recently made its guaranteed

income program permanent, which allocated $7.5 million in its 2026 budget to provide 500 monthly payments to thousands of residents. So that's the end of the article. That was for lower income families. And so there you go. Even though at least in some of these pilot programs, there's one in Cook County here, the county that Chicago was in where I live. There's one in New York, even though, you know, AI is not a main driving factor in news reports and in the conversation that has led up to these pilot programs, obviously AI has been discussed. So although maybe UBI has not become a household conversation, it is on the table and people are talking about. All right. Number 20, open source models surge and they temporarily overtake proprietary models.

Yeah, I literally said it will temporarily overtake proprietary models. And guess what happened? Well, exactly that happened. So, you know, one of the biggest ones was the Deep Seek shock. So that was in January. DeepSeek released R1, which was an open weights model. And at the time, the best model in the world was GPT40. And on many benchmarks, the open source model beat OpenAI's GPT4O.

Obviously, Open AI, I believe it was just a couple weeks later, released an update to GPT4O that on most benchmarks, you know, they kind of reclaimed that top, that top spot. But this literally happened temporarily. Open models surge. And I think they continue to surge, right? If you look at the top 10 models, I believe right now on LM Arena, you have, depending on on the day, anywhere from three to four of them are open source. So they have and they continue to, at least on certain categories.

categories of benchmarks be a top one, two, or three model. Not usually overall. I still think that for the most part, you're always going to have Google Gemini or an open AI model as the top model, especially on these aggregate AI benchmarks or LM arena, but we did see it. It did happen. All right. Next prediction, kind of related to that number 19, Chinese AI will dominate and cause confusion. So very similar, some of the reasoning and rationale from our last one, but 30% of of global AI usage right now is Chinese open source models. That's according to an open router study of 100 trillion tokens. So I'd say that's dominated to just talk about Chinese open source AI models. Globally, 30% is a pretty big piece of the pie.

And definitely causing confusion, right? The whole deep seek debacle of January, right, where they said, oh, we trained this thing for $6 million and the U.S. stock market lost more than a trillion dollars in market cap, although they did recover that pretty quickly. Well, that was obviously not correct because according to a semi-analysis report, they found that Deep Seek's infrastructure and development costs were estimated at more than a billion So Chinese AI did open source AI did dominate and it did cause a lot of confusion because they were putting out some we'll call it half information misinformation put out there to intentionally short the U.S. market. All right. Number 18. Perplexity will either pivot, get acquired or get squashed.

This was not a popular take. Let me say that. Back in January, you know, everyone was extremely. bullish on perplexity as an answer's engine. And at the time, I'm like, there's no room for perplexity to be an answer's engine only anymore. So I said they're either going to pivot, they're going to get acquired, or they're going to get crushed. So I will say this, perplexity did grow, right? They did not grow at the rate of, uh, you know, their other competitors, but they didn't get crushed. Well, why? Well, I think it's because they actually had a pretty hard pivot. Also, They were being considered for an acquisition from Apple, just FYI. So the acquisition thing, it got kind of close.

But I will say perplexity actually pivoted. I don't think if you look in a year from now, the Answers engine business will not be one where perplexity is a player. And even though they were the first Answers engine, they can't compete. They can't compete with the users from Google Geminiya's surging. They can't compete with Chad GPT. Even now, Anthropic Claude, you know, has this piece covered, co-pilot. Everyone has that piece covered.

So perplexity actually had a pretty smart pivot. They were the first big player to go into the agentic browser space with their comment browser. They pivoted hard into shopping. They have their finance product. And also in iOS app, that brought their perplexity assistant that brought AI actions to the phone before Apple's own Apple intelligence. So they had some very smart and I will say some pretty hard pivots, right? I wouldn't call it a 180 about face by any means. But I think we saw in 2025,

perplexity had to diversify from being just an answers engine, just like I called. All right. Number 17, I said API prices are going to be like early 2000s wrap and they were going to drop it like they're hot. So my prediction in January, which people scoffed at that I said, you're going to have a frontier or near frontier model API prices drop hard and get below 50 cents per million tokens. And everyone's like, Jordan, you're an idiot. You have no clue what you're talking about. That's, you know, more than 15. X cheaper than what it's currently at, not possible.

Oh, look. Yeah, I was right. Sorry. So two of the top, two of the top seven models right now on artificial analysis, clock in there, right? So Kimmy K2 Thinking, one of the top, I think it's the fourth model on artificial analysis, which again is an aggregate benchmarker. And both of these are also top 10.

But their frontier models, Kimmy K2 Thinking. Fantastic model. Clocks in at 60 cents per million, but on cash, it's 15 cents, right? So yeah, we hit that there. And then obviously a couple of weeks ago, we saw DeepSeek V32. And this is uncashed at 27 cents. Yeah. So did that.

And then on cash, it's two cents, FYI. So, yeah, hit that. Everyone's like, there's no way. There's no way intelligence, frontier level intelligence. will ever be that cheap. Yes, it is. I don't know what it's going to be next year, right? Got to bust out the calculator and spend some time with that.

But I would assume it continues to go down. I mean, even if you look at the recent ARC AGI on ARC A2 from Open AI, I think the cost that they needed to achieve what they achieved on the ARC AGI test was like a multitude of 390 times cheaper. So things are getting way cheaper and fast, right? So yeah, it's going to continue to drop. All right. Number 16, embodied AI will be an exploding sector.

So my actual prediction was that embodied AI, such as humanoid robots, autonomous vehicles, drones, and wearable AI would be an exploding sector with large-scale funding in robotics and drones. Again, saying this today probably means nothing because you're like, okay, duh, it wasn't like this in January, right? You could say like figure in very few people unless you cared about robotics knew what figure was. But here's the receipts. Here's the reality. So VC funding in robotics and drones soared over 40% all right in 2025.

Huge. Huge jumps. I mean categorical leaps there. That doesn't happen. 40% jumps in VC funding doesn't happen. Kathy Wood, you know, obviously very famous Ark Invest, said in October of this year that humanoid robots will be the biggest of all AI opportunities. Kathy, welcome to the party.

Been saying this for years. Sorry, but just being honest. So Figure AI also exceeded a billion dollars in Series C funding alone in September at a $39 billion valuation. So yes, embodied AI, even though, you know, people weren't talking about it too much back in January unless you already had a preexisting interest or specialty or overlap that needed to, you know, maybe you worked in, you know, warehousing or logistics so you had to be familiar.

But this is huge, right? Throw in there, you know, self-driving vehicles. which changes logistics and transportation, you know, last mile deliveries with autonomous, you know, robots, you know, great computer vision capabilities and leaps, right? I see them on my block all the time in Chicago, right? Whereas like, if you would have told me last year that I would see those things every day, I'd be like, no, that's from the future. No, I mean, they're literally out on the streets all the time.

So embodied AI not only made a big splash in 2025, you know, and maybe most people weren't expecting it, but I think that splash is going to go title in 2026. All right. I think we only have one or two more here, so stick with me. So number 15, just barely miss this one. Dang it. And this was an absolutely outrageous claim at the time. So I said that AI video tools will be able to one shot five minutes. HD videos. So I said that you will be able in one prompt to be able to produce a five minute

HD video generation. Plus, I said that we're going to have some advanced personalized media. So did it happen? Not quite. Right. But I wouldn't be surprised if in the last two weeks of December, if we actually get there. If not, we'll get there in the first quarter of 2026. So while single shot generation is often capped by most of the big providers between 20 to 60 seconds just due to compute, technically it is kind of possible in other ways. So one of them is YouTube shorts integrated Google's VO to allow infinite remixing and stitching. So just by clicking a couple times, you can effectively achieve what I talked about with advanced personalized media and five minutes. that's right, just stitching them together. Also, worth noting, Disney just a couple of days ago,

announced a $1 billion investment in OpenAI, and we are going to be seeing AI videos on Disney Plus. So I did talk about that, that we're going to be seeing AI videos and not even know it, right? We've seen plenty, plenty of stories that we share in the newsletter over the last couple of months. The biggest studios in the world are using AI video. As the models get better and better, right?

I wouldn't be surprised if in five years, the majority of everything we watch is just AI generated. It just makes sense. It's cheaper and it will be better. Yes, AI video will be better eventually than what you can shoot with a camera, right? Because even what you can shoot with the camera is somewhat limited. Obviously, if you have a billion dollar budget, you know, you can probably do great things. But for the most part, a lot of it is still done with CGL.

with special effects, but essentially, I think CGI, right? Any big production has some sort of CGI. I think it's just replace CGI with just AI video, right? I think you're going to have some baseline shooting always, so you can kind of have that to feed the video models. But I'd say the overwhelming majority of everything that we consume on the big screen in five years will be AI generated. Oh, and we did get very close here.

because actually Kling AI does have a feature called extend using the extension feature to get to three minutes with a single prompt. So we just barely missed it. There's technically I could have said, oh, yeah, we hit this because there are some other AI video products that you can get up to an eight minute video from a single prompt that they're not good. And you can't just do any type of videos. It's only like certain types. So I could like stretch it and be like, hey, technically, you can't. can get eight minute, but they're not good and no one would ever use them. But Kling AI, fantastic, one of the top AI video models in the world.

And with one prompt using the extension feature, you can get three minutes. So it stitches together all these different scenes, but you don't have to do it. One prompt, it can do three minutes. So we almost made it. Like I said, I wouldn't be surprised if by the end of the year or early 2026, we actually get that prediction one, correct. All right. Last one or two here.

Let me see. Yeah. Two more. All right, here we go. So number 14, the future of traditional internet comes into question. So my prediction there was the traditional internet model, which is ad supported, click through traffic, would come into serious question as AI answer engines, abstract content away from publishers. Not only that, but just the traditional internet is just going to become slop, right? And I think we've seen that.

So a lot of these studies I'm talking about here. I don't believe them, but I do think that they signify the truth. So there's, you know, these studies that say, you know, 90% of new content online will be AI generated by 2026. Yeah, I probably believe that, but you can't technically prove it. There's an A, uh, an H refs study, uh, in April that found that 74% of, uh, the 900,000 newly created web page, webpages contained AI generated content. Can't technically prove that. But I believe it, if that makes sense, right?

But the future of the internet, it's not a great place, right? Or it's at least the future of the internet or the future of visiting webpages. I think it's going to be archaic in a couple of years, right? I think everything is going to be driven by your AI operating system, whatever it is you use, right? Whether that's co-pilot, whether it's Chad GPT, Gemini, or Claude, I think you're not going to really browse websites, right? And I think we're starting to see that also with chat GBT apps. It's bringing in the entire experience into the chat GBT interface, right?

And sharing all the context and all the data right there. So I do think that eventually browsing the internet will be kind of like a lost art. I do think eventually, I don't know when, I do think eventually most of the new content that is published online will be just a lost art. for large language models and scrapers, not necessarily for humans. If humans discover it, great. But I do think the main reason will be for large language models

to show up in AI search. And also NPR declared this an extinction level event or online publishers in July. So the reality is, yeah, that one, we hit it. All right, last one for part one. Number 13, social media makes deep fake problems worse. Yeah, that was the dang truth. So my prediction there was social media platforms would make deep fake problems worse in 2025,

making misinformation and impersonation significantly worse and harder to spot. So that one maybe wasn't the riskiest. But at the time, again, remember, the tech wasn't that good. The photo, the photos, AI photos weren't that good. You know, we didn't have nanobanana. We didn't have GBT image. It was kind of a little harder, right, to create a decent looking deep fake photo or a deep fake video or even the deep fake audio wasn't the best. Now you can't tell the difference.

So some stats here. Deep fake fraud incidents surged 19% just in the first quarter of 2025. Also some deep fake detection research from I approve found that only 0.1. percent of participants in a study could correctly identify all fake and real media. So yeah, hardly no one, if you look at a series of what's fake and what's real, hardly no one can get it all right. CEO fraud using deep fakes now, according to reports, targets at least 400 major companies every single day.

You have targeted A-lister celebrities, you know, that people are using their likenesses, They're literal videos to either scam people or to sell products that these celebrities have no attachment to. I mean, literally, you stories of this every single day, a list celebrities. Someone just, you know, makes them go pitch their product. And they're like, no, I didn't pitch. It's happening all the time. And then probably one of the more prominent ones as of recently, an AI imposter successfully mimicked the freaking U.S. Secretary of State. one of the most important global positions of any political leader, right?

U.S. Secretary of State Marco Rubio's voice was cloned and an AI imposter fooled fooled foreign ministers and other officials into believing it was U.S. Secretary of State Marco Rubio. Not a good thing, right? The deep fake problems, unfortunately, are only going to get worse, right? Although you see technology like Open AIs, SORA, in the cameo features as great things, right? As, oh, it's great fun. Yes, and there are some safeguards in place.

But usually when good technology is born, bad technology replicates it. And it is going to, like I said, it is getting nearly impossible for even a trained eye to tell the difference between what is real and what is fake. So the reality is it's a problem in that prediction was obviously correct. All right. That is a wrap for volume one. So from AI agent orchestrators to defaics the new tech order. So I hope this was helpful hitting rewind on that original roadmap. So let me know.

Did I lead you astray? Right. And please remember, even though these might not seem like crazy predictions. at the time. I'm telling you this now because also kind of pre-promoting our 2026 AI prediction and Roadmap series that we're going to be probably in about one month from now. So we're probably going to do the same thing. We're going to spread it out over five episodes. I'm letting you know now. I've been working on it for months. It is an absolute necessity that you listen in Because, again, you're going to listen to the recap of this one in a year and be like, oh, yeah, duh.

You might as well know now. I'm literally laying out the roadmap for you to be more successful using AI, for your company to be more successful using AI, not saying, you know, I'm getting it all right each and every time. But I'm making some big, bold steps that are going to help you. So make sure, mark your calendars. mid-January for our 2026 prediction and roadmap series. And make sure to mark your calendars for tomorrow as we deliver on part two. We're going to get into numbers 12 through one. It's going to be an exciting one.

Thank you for tuning in. If you haven't already, tell someone about this. If you're listening on the podcast, please make sure to follow the show. That helps. Please do that. And if you could, if you find the podcast helpful, please leave us a rating on Spotify or on Apple Music.

If you're listening on the live stream, please repost this on LinkedIn. Other people need to know and you need to go to your everyday AI.com. Go sign up for the free daily newsletter. We'll see you back tomorrow and every day for more everyday AI. Thanks y'all. Meet Firefly AI assistant. Now live in Adobe Firefly, the Allman One Creative AI Studio. Just describe what you want to create in your own words and the assistant handles the rest,

orchestrating multi-step workflows across Adobe Creative Cloud apps, including Photoshop, Premiere Express, and more in one conversational interface. You direct the outcome while the assistant accelerates execution. Stand control with the ability to step in and refine at any time. See it today at firefly.adobie.com. And that's a wrap for today's edition of Everyday AI. Thanks for joining us. If you enjoyed this episode, please subscribe and leave us a rating.

It helps keep us going. For a little more AI magic, visit Your EverydayAI.com. and sign up to our daily newsletter so you don't get left behind. Go break some barriers and we'll see you next time.