The transition toward agentic AI marks a significant evolution from passive tools to active digital partners, though it risks oversimplifying the profound security implications of autonomous system control. It reflects a broader trend where the pursuit of personal optimization increasingly relies on delegating human agency to automated workflows.
Deep Dive
Prerequisite Knowledge
- No data available.
Where to go next
- No data available.
Deep Dive
Autonomous Intelligence - part 1
Added:[music] [music] [music] [music] >> Without further ado, just got to sit back, relax.
And we're going to go to the airport.
>> I don't know what I expected, but >> Okay.
>> cool.
>> Seems to come full circle. I remember very similar scene at an airport many, many eons ago.
Listen, James. Here's what's going to happen.
In a loud and clear voice, you're going to say, "Hey," that word that hasn't spoken yet, and then >> He who must not be named.
>> And then you're going to wait about 0.75 to 1.25 seconds, and then you're going to say, "Take me to," and then you're going to read that address.
>> Okay.
>> screen that says start self-driving.
Just click it.
>> Do I have to press the brake?
>> Nope, you don't.
You don't.
And now you just And now you just be present, and just experience, and just really just soak in all the [ __ ] intelligence.
>> Can we watch House?
>> You know what? We absolutely [ __ ] can. I'm glad [laughter] you asked.
I have this iPad right here, or Should I say this?
Whatever the [ __ ] tablet.
>> Bro, what the gigapad?
What the [ __ ] actually is that?
>> If you want to go to the gym and the plat like whip it real quick, that's [ __ ] fun. That's great. But when you want to visit like [ __ ] knows who and like somewhere in Florida, you have to drive like an hour and a half, 2 hours, you really want to sit there and like hold the wheel? Like just This does everything for you. It's crazy.
>> watch Cocomelon?
>> Cocomelon on the iPad?
>> Is that what you watched today >> Uh, no.
>> Don't say yes.
>> I don't know what the [ __ ] is. It sounds like a ShopRite aisle.
>> [laughter] >> I can tell your androgen quotient is quite high cuz you're you're immediately you just kind of trust it just whatever.
Just low test people >> [laughter] >> is it okay holding your life is [ __ ] fine but you just know.
>> An androgen quotient. All right, all right.
>> Margin Lane, right?
>> Oh.
>> Easy, easy, very easy. Intelligence.
>> It is.
>> Intelligence.
>> What about the tolls?
>> The tolls? It's literally it's not an issue. It was thunder [ __ ] rain on the way here. Like it was so bad like I looked at the cameras and I couldn't make out if there was like a car there or if it's just like just a thing on the road but it just knew. It could just take that feed and somehow intelligence it and it's just doing it right.
>> had any instances and be honest Elon's not going to be mad where you've been genuinely like oh [ __ ] that was almost really bad if I didn't take over?
>> Um back on FSD 13. Yeah, not since like 14. So any disengagement I make now it's never due to anything safety critical it's just due to the navigation being [ __ ] >> I see. So FSD so there was a significant jump with the new FSD update.
>> I think it's in like unbelievably better. So much [ __ ] better it's unbelievable.
>> And the hardware 4 is like essential if you want a fully immersive full self-driving experience. You need a newer Tesla.
>> I mean >> do you live from Kyle's? Like >> It's a 2-hour drive for me to get here but is it really a 2-hour drive?
Is it really though?
>> So you drove from like where where do you live? Egg Egg >> Yeah, if I were to be home I would be sitting on my couch talking to Telegram.
I did the same [ __ ] in the car.
[laughter] It's the same [ __ ] Like driving somewhere means nothing now. It doesn't have that tax to it.
>> you to bring a strap.
Must have missed that one.
I got to like hold on to your wrist.
Oh [ __ ] this [ __ ] >> [ __ ] >> [music] [music] [music] >> Tesla FSD expectations versus reality.
>> From the way you've described it, it's like one-to-one.
>> One-to-one or but I just I usually describe [ __ ] in very colorful words, yes.
But it lived up to that absurd expectation.
>> Especially when the bus started like being a little bit of a dick and the car automatically slowed down like my instinct was, "Oh, I should grab the wheel. Oh, never mind."
You know, it started drifting over. I'm like, "Oh, I should take over now. Oh, never mind." Cuz in my car, I have autopilot which just keeps me in the lane. But I don't have full self-driving.
>> You caught a real one, not a glitch, and it was my fault for playing it brainlessly.
There's a standing injunction in my constitution that voice notes should be a casual summary, under about 500 characters, roughly 30 seconds. It exists for good reason.
>> Since when?
>> Gemini's voice synthesis silently cuts off around 2 minutes. Telegram stops rendering clips >> [ __ ] flankers, man.
>> about a megabyte. And for routine back-and-forth, you don't want me reading you an essay. But I was applying that 30-second [laughter] ceiling >> There's like three different tiers of AI usage that I've like kind of like thought about, right? I guess this is like a loose conception. Still fresh, still pending.
There's like tier number one, which is like sort of like a hater, someone that basically like knows about AI, generally has like negative like feelings about it cuz maybe like consciously or subconsciously like, "Oh, it might like replace my job.
I'm not so thrilled about that." So, it's going to color like my like opinions about it as kind of a bad thing.
And slash they just kind of haven't really used it. Maybe they're a bit like stubborn, they don't like new [ __ ] like they like things the way they are, they're all like nostalgic and [ __ ] There's like that type of person.
Then you have like one tier further, which is a person that like is relatively neutral to like AI, is like well calibrated incorporating into the world, and they like use AI a little bit. Like they they they know, oh, there's a ChatGPT, like I'll maybe I'll download the ChatGPT app, maybe if I'm feeling a little spicy I'll get like Claude or something like that.
>> Maybe maybe maybe the $20 a month one.
>> you know, hey, like 20 bucks a month, like [ __ ] it. Like it says the word pro, it's got to be legit, right? Like I'll buy it. [laughter] Like it sounds [ __ ] lit. And they like use it, and they're like, oh, this is like super cool. Like they look friends, like oh, look, I can like figure this out.
But then they're like, oh, it kind of like it kind of hallucinated this, it's not good, it's not really accurate, so it's like not that great. It's like a cool little party trick, but like I kind of can't take it seriously cuz it just makes like kind of mistakes, and it's a bit like inaccurate, a bit silly at times, right?
So there's that person [ __ ] that.
>> [sighs and gasps] >> And then there's just then there's just >> A big gap.
>> There's probably a few more in between that I don't know them. I just like [ __ ] skipped them, and I just like went >> [laughter] >> straight to the finish line.
But well, I just want to touch upon this, right? So when people people that have like negative feelings towards AI because of its like underperformance during certain times. Like for example, like it'll hallucinate [ __ ] it'll like get like a fact wrong, like it doesn't know how many Rs are in strawberry, or it can't like it like fumbles like basic like math.
The reason for that is because they're using like a cheaper [ __ ] model.
Because for the AI to do its like work, like it's actually like unbelievably like computationally expensive. Like even like all like the frontier models like right now, even like the subscription tiers that cost like two $300 a month, right?
It's it's still If you have a $300 a month subscription and you use it on a regular basis, it costs more for them right now to do all of that compute than it is like than how much like you pay them. So, it's very computationally expensive. So, if they're going to give people free versions, they're going to give them the shittiest, smallest, like, horrible [ __ ] models. And those models, they prioritize speed over like depth. So, when you like are using an AI and then you ask it something and then it's [ __ ] typing its answer up before you even click send, it's just like so fast. Like, that is just like a garbage response cuz like I said, it's prioritizing like speed over depth.
So, I'll use like Chat GPT for example cuz you have it. When you're using Chat GPT free version, like, >> [snorts] >> you might as well like you're just speaking to Google on LSD. That's a compulsive liar. Like, like it's [laughter] just like it's just [ __ ] It doesn't give you But with Chat, you have three tiers, right? Like, you could actually toggle the setting. Like, you you only see two of them cuz you have the $20 a month. But you have the instant, right, which is free. Then you have thinking, then you have pro.
Thinking is like a bit more solid, right? It's a more It's a more powerful model. It's like a larger, more capable model that like thinks for a bit longer.
Cuz how long it thinks also like is highly correlated with the quality of answer. But if you get the Chat GPT pro tier, for example, like, >> $100 a month?
>> It's like at a minimum $100 a month.
Like, it's on a different [ __ ] astral plane. Like, depending on like what kind of question you ask, like, it could literally sit there and think. I've had it think for like 35 minutes on a single question. Think about that. 35 [ __ ] minutes. You're used to just like getting [ __ ] back like this. 35 [ __ ] minutes. Are you kidding me? Can you just imagine just like the quality of answer that that would derive?
>> But now, instead of Let's talk about the agentic AI.
>> Okay.
>> Cuz this is this like This has changed my life.
>> So, basically, when you use Chat GPT, right?
It like it exists on a server somewhere in like, uh, Sam [cough and clears throat] Altman land, right? Like, some kind of [ __ ] tunnel where he like lives. Like, it all like exists there.
And you're using You're using that like intelligent brain at all on the server like through like um, through like chat GPT like formal software. Like the chat GPT app or the chat GPT website. So, you're accessing that like first party like portal to use that [ __ ] brain, right? But, there's a way that you could use that brain's that like AI brain in something outside of the formal chat GPT system. Cuz when it's in the chat GPT system, it's great, it's convenient, it's nice, but it's like highly limited in what it's able to do. So, what people recently figured out is that they basically figured out a way [snorts] to have any AI of their choosing to leverage that AI's intelligence, like borrow its brain, and have it run your entire operating system on your computer. So, think about it. When you're using chat GPT or Claude, right?
You open up the website, you submit a question, and then it gives you an answer. You can't have it send an email for you. You could just have it like write the email that you have to then [ __ ] copy paste and [ __ ] send it like do all this. So, it's a little it's it's like a Q&A little like back and forth thing. So, what these agentic things do is basically instead of using the AI through a limited portal of like a chatbot, it uses that same AI's intelligence and has it be able to have full control over your entire computer.
So, anything that you're capable of doing on a computer sitting down, the AI can now do.
Right?
So, for example, that can involve creating apps, like writing code, uh manipulating files on your computer.
>> Sending emails for you?
>> Sending emails for you. So, one like example, you could be like let's say you set it up. But, I don't have it set up in this way cuz there's certain like security like reasons why it's suboptimal, but theoretically, you could easily do things like be like, "Okay, um go through my emails, find every single flight I have ever booked since the inception of my inbox, and I want you to go in there and find all the necessary information to be able to retroactively track all my airline miles. You have all my logins to my airline, and if I ever use some funky weird third-party airline that I don't have an account for, make an account, you know my password preferences, you know my username preferences. Just shoot, go off and do that, right? It would like do It would do [ __ ] like that. So, you could like really create these like like tasks to do.
So, basically, you have your computer, and what people do is they wipe their computer fresh or they buy a new computer, right? And then they install this like software in there that gives any AI if you're choosing to get permission to use your whole computer, and you could actually put multiple AIs into it. So, example for me, I have ChatGPT, I have Claude, I have Gemini, and I have like Rock, right? Like the main four that Frontier models. So, they're all connected together, so I could leverage the best attributes of all them, cuz each of them is slightly better at something than the rest, right? Which it changes month to month, but it's always like ongoing. So, basically, you choose you choose you choose the smartest one to be like your quarterback brain, your general contractor, and then you have like other ones do like subtasks.
Something called reverse prompting, right? Cuz you're hearing all this, and you're probably thinking, "How could this even apply to me? Like, what would I really use this for?"
And Dave told me, "You should just ask it. You should just literally ask it that question." So, I did.
I said, "I I my friend told you or told me about you. How can you actually help me?" I'm talking like talking to the AI agent, and it said, "How about you take me through your day minute by minute, everything you've done today, even the boring shit?"
I'm like, "Okay, well, I didn't really do much." But then I I as I was typing it out, I realized I actually did a lot.
It was just mostly on my phone. Like, I checked this, I checked that, I responded to this person, and then I conceived of ideas for this company, conceived of ideas for this company, had to talk to this person about this company, blah blah blah blah blah blah.
I'm like, "Okay, well, it's like a lot of like menial tasks that I've done today."
And it came up with truly like really good ideas as to how it can help me by automating a lot of these tasks. One of them, for example, it's done a bunch of things for me, but one of them is like I want to scroll Instagram to see what my peers are posting, but I don't want to get caught into a Reels loop where I'm just start I just start rotting because the algorithm is really good at taking me from like doing something constructive, which is like seeing what's going on in the fitness world or just the influencer world, and turning it into just like I'm just I'm watching like dog and cat videos.
So, I had the agent scan through like 40 [clears throat] accounts that I look up to, David being one of them, accounts where I want to in some way, shape, or form like resemble their content. I want to be part of that niche.
And I want you to scan every 24 hours everything they posted in the past day and see how that content is doing and show me the outliers. What is doing better than average and explain the post and then also link me the URL so I can click directly on that post, watch just that post, and not scroll anything else.
I just want to see the cream of the crop posts that have been posted by these 40 accounts that I look up to like, you know, like Trend Twins are on there, Dave's on there, people like that.
And if any of these posts have stuck out in the past 24 hours, send them to me and explain them. And then, on top of that, it decided it would not only do that, but it would kind of look at everything that's done well and give me ideas relating to my life cuz I had just explained my whole life. I'd explained everything about myself, everything I do in a day.
How can How can you How can you, James, take some of the stuff that you that's part of your life and make content about it? And I'll base it on this other content that's done well in the past day.
And it sends me an update every morning of like, "Here's Here's the posts that have done really well."
>> Hey, Revan. Revan.
>> Oh, let me find the right soundtrack.
Wait, no, that was you. That was That's the car that I saw.
>> yeah, yeah, but that wasn't like what >> Yeah, yeah, yeah.
>> This thing has no self-driving, but for some reason I really like it. I don't know.
>> I might need one.
>> So, this workout you're about to see, it's uh it's just like a casual uh pull day with me, James, and Kyle.
And this is not like the normal workout that I've been running lately cuz lately, for the past like many, many months, I've been doing like very specialized training to kind of think about it as like functional training, just really making sure that my body is just extremely healthy, balanced, and optimized cuz before, I would basically just do push-pull-legs all the time, no matter what, regardless, you know, if something hurt or something fell off, I would just keep pushing through. The pain would accumulate to where like I kind of just couldn't really do my workouts.
And then I would just basically stop.
So, now, um my entire training is just completely restructured. And honestly, I feel I feel phenomenal. Like I feel better than I felt in like many, many years just plainly in terms of just my day-to-day experience living in my own body, just the lack of pain, the mobility, the fluidity. I'm able to like do movements in the gym, not have any pain. So, basically, I guess recap on like, you know, the key areas of my injuries, shoulders are doing spectacular, back is doing phenomenal. Like everything just feels super, super good. So, this workout is obviously not part of my training, just more of like a casual, more like bodybuilding type of like hypertrophy-esque workout. Don't worry, I didn't go anything above RPE 7. We kept it nice and conservative, lots of fuel reserves, so nothing shady happening in that realm.
But, you could see, probably somewhere in this video, we're cooking up a pre-workout. That's my Transparent Labs pre-workout that is going to actually going to be launching very, very soon and this is like this pre-workout is genuinely elite. Like if you guys like like my last Euphoria, this is actually like better in every single way, but I'll talk more about that as we're coming closer to releasing that and um one more main thing I want to say is Gymshark right now is having their big sale, so whole site 50% off and um if you use code DAVID10, you get another 10% off on top of that. So I timed it so this video goes live when the sale goes live. So if you guys want to pause, check out the site, get your favorite items, and then you can come back and enjoy this workout.
>> [music] [music] [music] [music] >> I think of you and I pray.
Memories on knuckles of you.
My lipstick still on your face.
[singing] Can't you see my point of view?
I dream about you and I hate.
Falling in my sheets of Remember you [music] said "I'll never change my ways."
I close my eyes and see you there.
[music] Deja vu.
Falling in [music] my sheets of blue.
>> [music] [music] >> So, what is what is the expectations versus reality objective take?
>> It exceeded my expectations.
>> What about the driver? How How'd you feel driving?
>> Oh, I loved it. I want it.
>> Woo!
>> So, what do you know? She wants it.
>> Yeah, it's Mark II, F 1.8.
>> What's the zoom to 50, right?
>> Zoom to 50, correct.
>> Is it?
>> Yeah, this is a Mark II. You think I don't know my own damn camera?
>> Okay, this is the ultra wide. Okay.
Yeah.
So.
But you say this is better.
>> My [ __ ] phone. So, what can I just throw this like right in that garbage right there? What's the What Is there any utility to this at all?
Shut I mean you can argue.
>> It's probably a little better in low light. Sensor's a bit bigger.
>> Right.
>> like like no, like optically and everything this is like superior.
But not by that much. Because it's not it's not just about that. You have Here's how you have to think about it, right?
>> Convenience factor.
>> When you're vlog just one it's so convenient. Everything that's shot on iPhone has like a signature iPhone look, which is like sub communicates like authenticity and casualness, right?
>> No, but I don't want that for YouTube. I want to like I want niceness.
>> But maybe No, but here's where you do here's you contrast it big time. When you're like, "Ah, I just woke up, guys.
Meal prep. I weigh this much." Like that like stuff super casual. And then when you go to the gym, then you whip out your big cinematic [ __ ] So, you just whip out your >> no need for this in between [ __ ] >> You do little time lapse in the corner or something.
>> [laughter] >> You can do something like that.
But it's that that's like a philosophical thing cuz for me I'm like, "Okay, I want like my vlogs to feel like casual cuz it's like it's actually easier and more casual to film. But then if I want to do something cinematic, there's like a big tonal contrast shift cuz you could like see it goes from [ __ ] quality to something more professional and it kind of makes it a bigger flaw in my opinion. I just think about like >> This kind of interrupts the flow cuz it's like in between. Yeah, it's a little better than a phone camera, but like not as good as something like that.
>> you think about how easy it is to just come out like this? Literally pop a button and you're just like filming, right? I'm sure I'm sure you know about this feature, right?
>> You ever film in HDR?
>> Oh, aerial Alexa log.
>> Yeah, it's aerial Alexa.
>> The Lumix GH5 GH7 is aerial Alexa tech.
>> It's uh they have So, aerial Alexa designed like their own color profile just for this camera. So, obviously it's not an aerial cuz the sensor is inferior, but it has the same color [ __ ] Like I'll send you this video and you're just going to When you look at back you're like, "Why does it just look so [ __ ] good?" And this is a 30 >> Yeah, but I'm not putting this beautiful video in the middle of my dookie ass [ __ ] [ __ ] >> [laughter] >> 30 minute float, best IBIS. This is basically a 24-70 equivalent. It's 12 to 35, but because of micro 4/3 sensor it's basically 24-70.
Great battery, 30 minute float, mic, no clipping like once it can pop little clip.
>> Wait, wait, what?
>> No.
>> No, I don't >> It has to be a hypothesis. It can't be wrong.
>> How long would it take you to redo the trick?
>> I just want you to speed just speed run.
>> No, but it's literally you looked at a whole bunch of [ __ ] >> There's no 7 or 8 cards to pick from. He said, "Think of the card, one of the cards you see think of it and remember it." I said, "Okay."
>> And I didn't rush you or let you take your time.
>> I don't think he'll get it again, that's what I >> [gasps and laughter] [cough] >> He can't [clears throat] get it again.
>> Why can't I get it again?
>> Because you forced the card.
>> How did I force the card?
>> It's very easy how you force the >> card. You think you're that much of an NPC where you would let me force a card off of you?
>> I do.
I [laughter] do.
I DO.
>> HE'S MANIPULATING YOUR BRAIN. He's bullying you.
>> He forced the card. Do it again.
>> What do you mean force the card?
>> does it again, I'll say cold.
>> You have to James.
>> I'm going to get >> I'll hold the camera.
>> Okay. Okay.
>> James, James, shuffle these.
>> He's not going to figure it out. This [ __ ] guy is not going to figure it out again.
>> [laughter] >> Don't say anything even if you do.
>> I don't even know how to shuffle, but >> I'm not going to ruin it. I'm not going to ruin it. I think it's awesome.
>> I don't think it matters how to shuffle.
>> James, James, James, say stop.
>> Huh?
>> Say stop.
>> Stop.
>> Okay. Okay.
Not the bottom card of the deck, right?
Not this one, but any one of them that you see. Just remember one of them.
>> Okay.
>> Just keep in mind, I don't know what the [ __ ] they are, right? So, it's not like you that you just like shuffled that one till you said stop. And like you can go for an easy you could double, triple, reverse psychology as much as you [ __ ] want.
>> [laughter] >> That's exactly what I was doing.
Yeah. Okay. I I got one picked.
>> And here, even uh you know, honestly, remember two of them. That way like you could change your mind in a little bit.
Just remember two of them and between them you could change your mind.
>> Okay.
>> You have one?
>> Uh-huh.
>> Yeah?
>> I have two, yeah.
>> Okay. Is everything fair so far?
>> Yeah.
>> I'm scared.
>> Don't react.
>> Mhm.
>> Wait, did you you you saw the same back of cards as James?
>> Mhm.
>> James, what's your card?
>> 10 of clubs.
>> Have I been holding this in my hand the whole time?
>> I believe so.
>> The 10 of clubs is gone from the deck.
>> I'm not going through all this. Let me see what that [ __ ] [ __ ] is.
>> How the [ __ ] did you do it?
>> Dude. [laughter] No [ __ ] way.
>> It was just a random spot. I like double, triple, quadruple [ __ ] checked like I I I was like, "Oh, no, he's going to think I picked this one, so I'll pick this one, but then he's going to think I picked >> Where was it?
>> And look, and I gave him options. I'm like, "Remember a few things and start changing it like last [ __ ] second."
>> But but where was it? No.
>> My brain hurts.
>> The the trick must have been done when I announced what my card was, obviously.
>> But, if you check the video back, look, the deck is held in my hands. I'm like, "James, say your say say your card." And then I'm holding as you check the video back. I'm like, "Say your card." You say it. I'm like, "It's not here in the deck anymore." Play back the [ __ ] part.
>> No, here he took the card out. So, the trick is done before that.
He knows what card you pick. I don't know how. It's very close. It's good, man.
>> See, my brain >> How many cloud API tokens do you want to give me?
>> [laughter] [music] [music]
Related Videos
AI Agent Mastery Certification Course: Lab 4 – Tools & MCP
arizeai
350 views•2026-06-16
Real-time Voice cloning, Kimi K2.7 CODE, GLM 5.2 and 3D reconstruction | AI News
kaiexplainsYT
111 views•2026-06-16
He Believes AI Could Replace Humanity Faster Than Anyone Expects
LondonRealTV
815 views•2026-06-15
General Session by Rami Rahim-The next generation of networking: From vision to self-driving reality
HPE
108 views•2026-06-17
AI Doesn’t Think Like We Do
Mindset_Operators
954 views•2026-06-13
[PLDI 2026] Flatirons 3 - LCTES (Jun 16th)
acmsigplan
191 views•2026-06-16
How the Pauli-X Gate Works ? Quantum Computing Basics
QubitVentures
106 views•2026-06-13
Google DeepMind’s AI Halves UK Housing Planning Time
60secondsignals
467 views•2026-06-17











