AI in 2025: From Agents to Factories – Ep. 282

0:00:16 Hello, and welcome to the NVIDIA AI podcast. I’m your host, Noah Kravitz. Today, we’re
0:00:22 looking back on the year in AI 2025. But before we begin, if you’re enjoying the AI podcast,
0:00:27 please take a moment to follow us on Apple, Spotify, or wherever you’re listening. Thanks.
0:00:36 Our year began with NVIDIA’s Mingyu Liu talking about the importance of world foundation models
0:00:43 to advancing physical AI in episode 240. Forty conversations later, Jacob Lieberman introduced
0:00:49 us to the future of enterprise storage, AI data platforms, in episode 281. Along the way were
0:00:54 advances in AI models and the infrastructure that they run on, like the rise of agentic AI and the
0:00:59 AI factory. We heard firsthand from pioneers in healthcare, higher education, life sciences,
0:01:05 marketing, and other industries about how they’re using AI to advance their fields and
0:01:10 make work better for the people doing it. And we talked to everyone from researchers to roboticists
0:01:15 about the dawn of physical AI, where intelligence moves from our screens into the robots building
0:01:22 our cars, assisting our surgeons, and walking among us. 2025 was quite the ride. Let’s dive in.
0:01:32 This year in AI began as last year ended, with lots of talk about agents and agentic AI. So what exactly
0:01:39 is an AI agent? An evolution in the way people use generative AI, agentic AI is a move away from simple
0:01:43 call-and-response-style chatbots towards systems that have true agency.
0:01:52 Chris Covert from in-world AI breaks this evolution down into phases in episode 243,
0:01:58 moving from simple conversation to an adaptive partner, and finally, full autonomy.
0:02:03 We have this, you know, first, again, is that conversational AI phase. And I’ll use a gaming
0:02:08 analogy, right? The conversational AI phase gives avatars, gives agents, I’ll use them interchangeably
0:02:13 today, extremely little agency in doing anything other than speaking, right? It may be able to respond
0:02:18 to my input if I ask it to do something, but it’s not going to physically change the state of something
0:02:24 other than the dialogue it’s going to tell me back. It is an adaptive partner phase, where the AI is
0:02:30 observing and responding to changes on its own. It’s not micromanaging every decision, but it feels
0:02:35 like you’re collaborating with an agent or a unit that has just enough context to make smart decisions
0:02:41 on its own. Like an evolution of a recommendation engine being driven by, you know, a cognition
0:02:46 engine here. So it’s not just learning, but it feels like it’s learning what we need even before
0:02:50 we ask it. Again, I think that’s phase three. I think there’s still a phase four. And I think that’s
0:02:56 a fully autonomous agent. And that stage, you know, again, continuing our analogy, is a player two,
0:03:02 right? Where player three is it’s adapting to us. Stage four is, hey, this thing is an agent on its own.
0:03:07 It feels like I’m playing against another human. It is making decisions that feel optimal to its own
0:03:13 objectives, alike to mine or not. The immediate payoff of this capability is freeing human workers from
0:03:19 repetitive, error-prone, non-creative tasks, what we often refer to as toil. But here’s the key.
0:03:25 We don’t need the agents to be perfect to be valuable. In fact, we don’t even need them to do all of the
0:03:30 work for us. As NVIDIA’s Bartley Richardson points out in episode 258.
0:03:33 If it gets you 75, 80% of the way there, that’s fantastic.
0:03:34 That’s great.
0:03:39 Because what’s, you know, I’m sure, you know, you do your fair share of writing, right? Like,
0:03:41 the hardest part for me about writing is that blank page.
0:03:42 That blank page, totally.
0:03:46 Right? And if I can get something that’s 80% of the way there, it’s great.
0:03:52 AI runs on data, and agentic AI is no different, which is good considering the sheer velocity of
0:03:57 information creation in today’s enterprise. Data growth is creating a widening gap between the data
0:04:04 we have and the insights we can actually extract. Cytoreasons Shy Shen Orr describes this challenge in
0:04:10 The Life Sciences in episode 276, comparing the struggle to keep up with data growth to the Red
0:04:12 Queen effect from Alice in Wonderland.
0:04:18 You can think about it like data is exponential, insight is linear. Everyday percent data utilized to give
0:04:27 insight is lower. The analytical side of this and the AI solutions for this have been missing. The field is
0:04:34 still largely a manual field where you give people some data, they sit in front of their computer, you know,
0:04:39 they try to figure out, they make some value and insight for this. And I figured that’s not a sustainable
0:04:47 solution. This field needs to move to ultimately build much larger integrative solutions that bring in many
0:04:54 different angles of machine learning, AI statistics, and so forth to ultimately bridge this. The data
0:04:59 insight gap keeps growing. So you basically are constantly in a game in which you need to make it
0:05:04 faster. Just, you know, it’s actually what’s called, you know, the, in evolution. And then you remember
0:05:11 Alice in Wonderland, the Red Queen, right? Where she said to Alice, you have to run just to stay in
0:05:17 place, but the Red Queen effect. So this, this need for us to continuously run is a huge driver for
0:05:24 automation, acceleration. And I would even say the cognitive meta-analysis that we as humans need to do
0:05:30 to somehow describe to a machine, how we make decisions so that we can automate them.
0:05:36 Right. To handle this exponential growth, we need massive compute resources. AI factories provide the
0:05:42 infrastructure to support enterprise-scale systems. But traditional ways of approaching storage had a
0:05:47 major flaw, data gravity. The data was heavy, and moving it created security risks. Here’s NVIDIA’s
0:05:55 Jacob Lieberman in episode 281. So far, in order to do AI, you’ve had to send your data out to some
0:06:01 kind of AI factory with a GPU, do all your processing and copy it back. Right. So the data has gravity,
0:06:07 and it turns out that instead of, instead of sending all your data to the GPU, you can actually send your
0:06:13 GPU to the data. And what that looks like is actually putting a GPU into your traditional storage
0:06:20 system on that same storage network and letting it operate on the data in place where it lives
0:06:26 without copying it out. And the advantage of generating these AI representations with the
0:06:31 source of truth data is that if the source of truth changes, you can immediately propagate those changes
0:06:37 to the representations. The new version reverses this. Instead of shipping data out, we bring the GPU
0:06:43 compute to the data. This evolves the AI factory from a distant processing plant into a unified,
0:06:49 efficient pipeline. Sarah Laszlo from Visa explains what this modern factory approach looks like in
0:06:58 episode 256. What that means to me is a single pipeline that goes from a data scientist with an
0:07:04 idea about a model they want to build all the way to the model running in production. It’s interesting
0:07:10 because I hadn’t really thought much about this AI factory terminology like I had heard it, but I
0:07:15 hadn’t really thought it was what I was doing until I came here to GTC and I started hearing other people
0:07:22 talking about it. And then I realized, oh, that’s what my platform does. So my platform, recently we’ve
0:07:28 adopted what we call the Ray Everywhere strategy. So we use AnyScale’s Ray ecosystem to do the whole
0:07:33 thing, the whole shebang. So data conditioning, model training, and model serving, we do all in Ray.
0:07:39 And it is, and it is intentionally trying to be more of this factory concept where it’s just, it’s,
0:07:45 there’s not a whole bunch of distinct parts or distinct tools that are living in different places
0:07:51 that work differently. It’s just one unified, consistent pipeline from start to finish.
0:07:55 This shift isn’t just about efficiency. It is critical for data sovereignty.
0:08:00 Countries and companies need to ensure their sensitive intelligence stays in their buildings
0:08:06 and on their own soil. In episode 247, Karen Hilson from Norwegian telecom operator Telenor
0:08:10 explains why they built a sovereign AI factory in Oslo.
0:08:17 So like Hive Autonomy, for example, I mean, they work with logistics, robotics. So they are actually
0:08:25 innovating, I say, a lot of industries, whether it’s ports, as I said, sort of factories or this sort
0:08:31 of in their operations and have efficiency cases. So they have very specific customer needs that they
0:08:37 are trying to solve. But they are, the reason sort of why they were very interested in coming to the
0:08:42 AI factory is that they’re sitting with sensitive data. So it was very keen. They wanted it to be
0:08:44 really on Norwegian soil, right?
0:08:49 The Telenor brand sort of represents security, you know, so there’s sort of a gain that really helps
0:08:56 them. And then the sustainability part is super key. And so that was sort of the combination of these
0:09:03 three. Capgemini is also a customer of ours. They are developing a product of doing voice-to-voice
0:09:09 translation. And we can say, yes, that can be done. But these are for sensitive dialogues.
0:09:15 Not all dialogues can go out in the cloud somewhere. And so very sort of sensitive dialogues, if you
0:09:21 think, you know, within the health sector, within the police. So not so much on program, but again,
0:09:28 it’s sort of a safe, secure environment. And that’s sort of really key. And another customer is working
0:09:35 a lot with the municipalities in Norway. And again, with sort of sensitive cases that they
0:09:38 sort of really would like their data to be secured.
0:09:44 To build trust in these factories, openness is essential. Jonathan Cohen from NVIDIA explains
0:09:50 in episode 278 how open models, like NVIDIA’s Nemotron family, allow for the customization
0:09:52 required by sovereign projects.
0:09:58 If you say, you know, NVIDIA trains a model, a Nemotron model, and it’s great. But since you’ve
0:10:01 disclosed all your training data and look at your training data, for whatever reason, we have some
0:10:07 policies for this data we can’t use. And we can say, that’s fine. Everything you need to reproduce what
0:10:11 we did is there. You can train your own model, excluding that data. Or you say, well, I like the
0:10:16 data, but the mix is wrong. I don’t know. I’m a sovereign project. And it really needs to be
0:10:21 very good at speaking this language and understanding this culture. And that data wasn’t as represented in
0:10:26 your training set as I want it to be. Everything that we did is transparent. And so you can make
0:10:27 these modifications yourself.
0:10:33 Since our first episode back in 2016, the AI podcast has told stories about the real-world
0:10:39 impact of artificial intelligence. 2025 was no different. A big area of impact this year was,
0:10:45 once again, in healthcare, where AI is helping with everything from drug discovery to reducing
0:10:51 physician burnout. Here’s Anne Oztuot of Moon Surgical from episode 272, explaining how their
0:10:53 maestro system supports surgeons.
0:11:01 Physician fatigue is absolutely real. We did our first inhuman study in Brussels in Belgium with
0:11:07 a surgeon, and he used the system over 50 cases. And he told us after a few weeks, “Hey,
0:11:15 when I get back home in the evening, my wife tells me that I’m, you know, a lot nicer than before.” So,
0:11:21 like, what’s going on? And, you know, I mean, he attributed that just to his own fatigue.
0:11:28 He’s like, you know, I end my day in a way that is a lot more relaxed. It’s about both the physical
0:11:29 and the mental loads.
0:11:35 With AI and healthcare, safety is the number one priority. Hippocratic AI has tackled this
0:11:40 by building a constellation architecture, using multiple AI models that constantly double-check
0:11:45 each other. CEO Munjal Shah describes how it works in episode 262.
0:11:49 We literally have multiple models double-checking each other. Right.
0:11:54 And what people don’t realize is that a lot of the models now, they say you can give a lot of
0:12:00 input tokens to them now. Just put it all in there. It’ll figure it out. And which Gemini is like,
0:12:04 what, a million? I think it is now, a million tokens. So it’s like, oh, okay, no problem.
0:12:07 Right. But it can’t reason across it all. Yeah.
0:12:10 They’ll show you examples of what we’ll call needle in haystacks, where it’ll be like, okay,
0:12:15 it’ll find that one thing. Yeah. I mean, grepping for a word is not that hard in computer science.
0:12:21 It’s like, we can find a word. But what you’re really trying to do is reason across it. So I’ll
0:12:26 give an example. If you ask your care manager, can I have ibuprofen? And they say, sure, you can have
0:12:31 ibuprofen, but don’t take too much. That’s fine, right? Because it’s an over-the-counter medication.
0:12:35 Unless you have chronic kidney disease stage three or four, then it’ll kill you. Right.
0:12:42 Well, if you put the rules for ibuprofen and CKD into GPT-4 and then ask it, it’ll do great.
0:12:47 If you put in all the rules for all condition-specific over-the-counter medications and ask,
0:12:52 it’ll still do pretty good. It’ll start missing some sometimes, which is still not okay,
0:12:56 because you could kill people, but fine. If you put in the patient’s medical history,
0:13:02 the patient’s last 10 conversations with you, all of those rules for over-the-counter medication
0:13:07 disallowance, and the current checklist for what you’re supposed to follow with that patient,
0:13:12 and maybe a few other things and then ask it, good luck. And what it is, is we have
0:13:16 an attention span problem. But if you have multiple models, we have these other models
0:13:21 only focused on checking one thing at a time. So there’s an overdose engine, and it listens to
0:13:24 every turn of the conversation. It’s like, are we talking about drugs? Are we talking about drugs?
0:13:29 Yes, we’re talking about drugs. Okay. And then it’s like, well, okay, did somebody just say a number
0:13:35 that’s an overdose relative to their prescription or relative to max toxicity of what you can have of
0:13:40 that drug? Okay, it did. And it may not seem that hard for pills versus two pills, but when you’re
0:13:43 talking about creams and injectables, it gets quite hard. Sure. I took a whole bunch of my
0:13:48 testosterone cream and I rubbed it on my hand. Was that an overdose? Right. I don’t know. How much
0:13:52 cream was in your hand? Right. What’s a little bit? What’s a little bit? Was it a pea size? Was it a
0:13:57 cherry tomato size? Was it an apple size? Right. RLM knows how to ask all these questions. Yeah. And
0:14:03 knows how to navigate assessing whether it’s actually an overdose. And you cannot have, if a patient shares an
0:14:08 overdose information with a care manager in a clinical setting, you need to do something. AI is also changing
0:14:13 healthcare from a totally different angle by transforming agriculture. Paul Mikesell of
0:14:19 Carbon Robotics explains in episode 270 why his company’s approach to weed control swaps chemical
0:14:28 herbicides for AI-guided lasers. I’ve also learned a lot about the quality of our food system. And I know
0:14:34 that there’s lots of discussion about this now. We are becoming more aware of it, that different herbicides
0:14:41 are being banned in Europe, United States, etc. We are learning about more of the long-term negative
0:14:46 health effects. Again, the ones who really suffer from it over their lifetime is the farmers who get
0:14:51 exposed to this stuff in much higher doses than the consumer. But even the consumer, you know, even you
0:14:58 right now are participating in some form of a multi-decade, maybe multi-generational science
0:15:04 studies experiment. We all have glyphosate in our system at any, and so if you were, if you take
0:15:10 everybody listening to this podcast right now, if we all went and did a urine sample, you would find
0:15:15 about 90% of us would have glyphosate in our system right now. What’s glyphosate? It’s the active ingredient
0:15:21 in Roundup. Right. We know that it’s carcinogenic. Like any carcinogenic, it’s only a question of exposure
0:15:27 over time. So, we should be able to, with the kinds of technology that are available today,
0:15:33 with the things that AI can do, we should be able to take a step back and say, do we really need to be
0:15:40 spraying this stuff on our food in order to grow it and survive as a population? Yeah. My answer to that
0:15:45 question, I think, is no, we don’t need to do that. And we should be able to do things like laser reader.
0:15:50 Yeah. Beyond the healthcare benefits, carbon robotics robots help farmers operate more
0:15:55 efficiently and sustainably. And they look really cool, too. Speaking of cool, in the world of
0:16:00 marketing and media, agents are fundamentally changing the relationship between brands and
0:16:07 consumers. Firsthand’s John Heller joined episode 242 to describe a shift where AI agents curate the web
0:16:12 specifically for the user’s intent. I had been working in the gaming world and some of the
0:16:18 generative AI abilities for gaming assets when language models really came out. And something
0:16:26 struck us, something very powerful, which is, and this is a metaphor for the math inside, but AI now
0:16:34 understands the ideas and intents, needs you may have from what you’re reading, what you’re watching,
0:16:40 what you might ask it outright. And it can go find the right response or take the right responding
0:16:47 action. And everything is presented to you in a very natural human way. And if you back up a step and think of
0:16:52 that happening all the way through a consumer’s use of the digital world, from when they’re searching and
0:16:58 becoming aware of things they might need, when they do some investigation and read up on products or services,
0:17:04 when they go to browse or shop, when they buy, all of those modes change pretty fundamentally. They don’t replace.
0:17:11 We think they get enhanced because instead of the world of the past, where I maybe did a search,
0:17:17 got some directions and a link, went to a place, read up on something, browsed for something,
0:17:22 went to, saw an ad maybe, went to another place to try to find the version I want. Those are all sort of
0:17:30 separate shops, the internet where it’s, you know, the same content everybody sees. AI instead is going
0:17:37 to understand and learn at each moment what it is you need. And as with most things, AI data is the
0:17:43 core. The people who have the most and best data about a product or service are the brands. They are
0:17:49 the retailers, the people who sell it. So they can create brand agents, which means your experience
0:17:55 on the internet at all of those moments in the journey from first learning about it to figuring out what the right
0:18:03 configuration is and comparing and browsing and buying is going to adapt on the fly through these agents. So it
0:18:10 doesn’t replace the web, but it changes things from you looking at stuff someone wrote to something that’s partially
0:18:16 adapting to what you actually need, understanding your needs. But the agents that are doing that for you are from the
0:18:23 retailers and brands themselves because it’s their data that is what you need. And that sort of changes the internet into
0:18:25 kind of your internet for both parties.
0:18:31 While software agents are transforming the digital world, a massive shift is happening in the physical world as well.
0:18:38 This is the dawn of physical AI, where AI models don’t just generate text or images, they control things that move,
0:18:44 like the aforementioned farm machinery. According to Sonia Fidler, VP of AI Research at NVIDIA,
0:18:49 the scale of this opportunity is staggering. Here’s Sonia in episode 249.
0:18:56 At the end of the day, robots need to operate in the physical world, in our world. And this world is
0:19:02 three-dimensional and conforms to the laws of physics. And there’s humans inside, right, that we need to
0:19:11 interact with. You know, we typically hear the term “such AI” that operates in the real physical world as “physical AI”. So I’ll
0:19:17 maybe use that term quite a lot, right? So physical AI is really kind of the upcoming big industry,
0:19:25 very likely larger than generative and agentic AI. You know, Jensen typically says everything that moves,
0:19:29 all devices that move will be autonomous, right? So that’s kind of the vision.
0:19:36 So a robot operating in the real world obviously needs to understand the world. What am I seeing? What is
0:19:44 everything I’m seeing, doing? How is it going to react to my action, right? So understanding, it needs to act.
0:19:51 But there is a catch. You can’t train a physical robot the same way you train a chatbot. If a chatbot makes
0:19:56 a mistake, you might get a typo. If a robot makes a mistake, it’ll probably break something. To solve
0:20:02 this, researchers like Mingyu Liu are building World Foundation models, AI that understands physics and
0:20:08 space-time, allowing robots to simulate thousands of futures before they take a single step in reality,
0:20:16 As Mingyu explains in episode 240. So I think World Foundation models is important to physical AI
0:20:23 developers. You know, physical AI are systems with AI deployed in the real world, right? And different to digital AI,
0:20:32 these physical AI systems that interact with the environment can create damage, right? So this could be real harm, right?
0:20:39 Right, right. So a physical AI system might be controlling a robotic arm or some other piece
0:20:46 of equipment changing the physical world. Yeah, I think there are three major use cases for physical AI.
0:20:52 Okay. It’s all around simulation. The first one is, you know, when you train a physical AI system,
0:20:57 you train a deep learning model, you have a thousand checkpoints. Do you know which one you want to deploy?
0:21:03 Right. Right. And if you deploy individually, it’s going to be very time-consuming. And so then it’s bad,
0:21:09 it’s going to damage your kitchen, right? So with the world model, you can do verification in the simulation.
0:21:17 Right. So you have, you can quickly test out this policy in many, many different kitchens. And before,
0:21:23 you know, you deploy in the real kitchen. And after this verification step, you may be narrowed down to
0:21:30 three chip points. And then you do the real deployments. So, you know, you can have an easier
0:21:35 life to deploy your physical AI. Once these brains are trained safely in digital worlds,
0:21:40 they need bodies. And while we may see many form factors in factories, there is a massive surge in
0:21:47 humanoid robotics going on outside those factory walls. Yashraj Narang from NVIDIA’s Seattle Robotics Lab
0:21:53 explains in episode 274 how this isn’t just an aesthetic choice. It’s a practical requirement
0:21:58 for robots that need to work alongside us. You know, there’s a group of people, you know,
0:22:03 forward-thinking people, Jensen very much included, this is near and dear to his heart, that felt that
0:22:09 time is right for this stream of humanoid robotics to finally be realized, right? You know, let’s,
0:22:14 let’s actually go for it. And, you know, this begs the question of why, why humanoids at all? You know,
0:22:19 why have people been so interested in humanoids? Why do people believe in humanoids? And I think that
0:22:24 the most common answer you’ll get to this, which I believe makes a lot of sense is that the world has
0:22:32 been designed for humans. You know, we have built everything for us, for our form factors, for our hands.
0:22:40 And if we want robots to operate alongside us in places that we go to every day, you know, in our home,
0:22:47 in the office, and so on, we want these robots to have our form. And in doing so, they can do a lot of
0:22:52 things, ideally, that we can. You know, we can go up and down stairs that were really built for the
0:22:58 dimensions of our legs. We can open and close doors that are located at a certain height and have a
0:23:05 certain geometry because they’re easy for us to grab. Humanoids could, you know, manipulate tools
0:23:11 like hammers and scissors and screwdrivers and pipettes if you’re in a lab, these sorts of things,
0:23:16 which were built for our hands. As AI moves from the screen to the physical world, it is also
0:23:23 fundamentally changing our creative and professional lives. In episode 265, Canva’s Danny Wu talks about
0:23:25 AI and creative superpowers.
0:23:31 You kind of see like the magic of Canva is integrating all the different steps and different
0:23:38 parts of design into a simple page, as I like to call it. And so we really invested in our content
0:23:45 library, in millions of templates that make it easier to start. And what we saw and got really excited
0:23:51 about AI was that firstly, we can offer, you know, all the amazing high quality content for people to use.
0:23:58 That when the user might want something, they might want to have an idea that didn’t necessarily exist. Maybe it has
0:24:04 actually never been created in the world. Like AI just gives us this superpower and ability to actually
0:24:10 create things on demand specifically for what someone has in mind or in mission and just kind of turn that
0:24:17 idea, turn that like, turn that search term or prompt into something they can use to express themselves.
0:24:22 But as these systems become more widespread, we must focus on inclusivity. We need to ensure that
0:24:28 the data feeding these models represents everyone. Angel Bush, founder of Black Women and AI, reminds
0:24:31 us of the goal of true equity in episode 250.
0:24:38 One of the things that I’ve always said to people is, I want Black women and artificial intelligence to be so
0:24:44 successful that it no longer has to exist. We’re really not looking for members. We’re looking for
0:24:50 people to be a part of a movement and really understand and trust that vision of the movement that
0:24:56 we’re going to make sure that you have all the tools you need in order to be a part of the AI economy,
0:25:01 in order to pivot into your career. And in education, leaders like Dr.
0:25:06 Cynthia Teniente-Matson at San Jose State University are teaching students that no matter how powerful
0:25:13 the tool, the human element remains essential. Here’s Dr. Teniente-Matson in episode 275.
0:25:19 There are some students who are using the tools that I’ve talked to for study guides. There are some
0:25:26 students that are using the tools for first drafts. I think, however we use the tools, it’s important,
0:25:31 if we’re going to be writing about things or communicating, that we’re citing references and
0:25:38 saying, you know, this was co-developed based on whatever sort of information they might have
0:25:46 retrieved from the instrument. And also to validate it, because no, these hallucinations exist. But as time
0:25:52 goes on, the hallucinations are diminishing, especially if you’re building your own custom
0:25:57 GPTs. That doesn’t mean mistakes aren’t going to happen. Sure. But that’s, as I say to students
0:26:04 regularly, Noah and faculty and staff, you are still the human in the loop. We’re not trying to replace
0:26:13 the human in the loop. Be, you know, have the tool be your co-pilot or your assistant that you’re
0:26:21 directing. So, looking back on 2025, what’s our best piece of guest-given advice for the year to come?
0:26:27 It’s simple. Start now. As Derek Slager of Imperity puts it in episode 271, if you’re still on the
0:26:33 sidelines when it comes to artificial intelligence, it’s high time to get in the game. I would say the one
0:26:38 piece of advice, and I give this advice a lot, is start now. It’s so important. It’s so important
0:26:43 because, like, it’s early, right? We’re still figuring out the patterns and the practices,
0:26:50 you know, like, as an industry, we’re learning a lot about kind of how to, you know, put these
0:26:56 incredible new technologies together in ways that really, you know, move the needle. And,
0:27:01 you know, right now, you just have a choice, right? You can be a doer who’s in that learning loop,
0:27:05 or you can be an observer and kind of, you know, wait and see. And I think, you know,
0:27:09 we talk a lot about this here, like, you know, speed’s the only thing that matters. And so,
0:27:13 I don’t think it’s viable in the current market to be outside that learning loop.
0:27:14 Right.
0:27:19 And the good news is it’s early, right? And so, you’re not too late, but it’s getting to
0:27:24 the point where pretty soon you’re late. And so, I think we’re certainly past the point. And again,
0:27:27 this is something that’s changed in the last six months. We’re past the point where people are like,
0:27:34 well, we’ll see if this AI thing plays out or not. Like, it’s overwhelmingly obvious where things are
0:27:40 going. And so, yeah, get off the sidelines, get in there, try stuff, learn. It’s easier than ever,
0:27:45 you know, to do that. There’s more information out there. And of course, you know, AI feeds itself,
0:27:48 right? AI can also help people figure out where to start and how to get through. And so,
0:27:53 yeah, start now and go really fast. That’s the path to success.
0:27:59 We are moving toward a future of collaboration, where human creativity is amplified by silicon
0:28:04 capability. NVIDIA’s Jacob Lieberman leaves us with this final thought on the partnership between
0:28:07 people and agents in episode 249.
0:28:16 There will be teams composed of carbon people and silicon agents, and they’re collaborating on tasks.
0:28:22 And at various times, the humans will be conducting the orchestra, and at other times,
0:28:27 the orchestra will be conducting itself. And that might be the most efficient way to get the work
0:28:32 done. Human judgment is critical. Human strategizing is critical. And there’s always room for that.
0:28:40 So it’s a way to complement the things that we’re very good at with some of the things where we could
0:28:42 use some help. Yeah.
0:28:51 2025 was an incredible year for AI, and all signs point to 2026 being full of more breakthroughs and
0:28:57 transformations in artificial intelligence and how we use it to change the ways we live and work.
0:29:02 Follow the NVIDIA AI podcast wherever you get your podcasts to stay up with the latest in the industry as
0:29:10 told by the people creating it. And browse the complete archive of episodes at ai-podcast.nvidia.com.
0:29:24 Thanks for listening.
0:29:24 you