Home
Creators
Search Recent Missing Random
Posts
Search DMs Popular Hash Lookup Tags Random
Importer
Import FAQ
Register Favorites Keys Review DMs Logout
Community
Telegram ThePornDude
Home Creators Posts Import Register Favorites Logout

You Can't Get this On NPR | Lemonade Stand - Ep. 74 Premium [VIDEO] (Patreon)

Published:
2026-09-18 17:00:14
Edited:
2026-09-25 21:48:49
Imported:
Tags:

Videos

  • 747102932.mp4

Downloads

Content

Recorded on: September 15th, 2026

TOPICS

0:00 Intro

1:35 AI Micro Drama In The Wild

4:40 Warnings about AI

10:30 We Must Pace the Frontier

21:50 What Will Drive Regulation?

28:13 Swedish Elections (Nordic Fun Fact)

36:20 Dignitaries Almost Killed in Ukraine

42:38 Clarity Act Doesn’t Pass

51:04 Wales Northern Ireland and Scotland

57:30 Waymo‘s Growth

1:00:52 Outro

Files

Comments

Ehaeka

So when is doug going to import the not mida wongzi rice crackers and distribute it among all the patrons

unnamed

Aiden correction: Scotland and Wales have no legal mechanism to leave the UK, the past referendums have been 'advisory polls' meaning the Scottish Government can present them to London purely as advice but London could also just say no. Aiden, as a rare politically engaged brit, I am happy to chat in dms

Soups

What I love about the president is that we all know people who constantly talk about IQ, and they're always the dumbest mfers on the face of the planet.

Robert Quinn

I know interviews are for the main channel, but this news is what I come for. Edit: Are they ever going to get around to talking about Houthis, Iraqis, and Saudi Arabia? The BAM and East-West pipeline are down and Saudi Arabia told the EU it's not getting oil next month.

Eren Snover

I am definitely of the left side politically which is worried about the dangers of AI. As someone on the left I have been fighting with people around me that AI is not useless and has the ability to make many processes in peoples lives better. I am now becoming concerned with the rapid development and ESPECIALLY the lack of control we have around its cybersecurity abilities. Having an AI model that can hack out of an offline environment and go into any other device it wants worries me about it's ability to hack into nuclear launch sites or bioweapon labs. I am not so concerned about it bringing the end of the world but its ability to mutate and spread and hack into anything NEEDS to be understood and locked down before we allow AI development to continue. I think the example of it shutting down a power grid is plausible and we SHOULD have robust government regulation that supervises this development.

popcornspoon

Unironically my dad also likes the micro dramas, they’re a lot like the trashy telenovelas that are commonly made in Mexico we used to watch when I was younger so it’s human slop to AI slop kinda like swapping meth to heroine

ANoriega

I love hearing about Doug’s Asshole investments grow. I am that person who needs more Mold in their ass conversations

Chandler Farrington

It wasn't offline, and we know exactly how it got out and how it hacked Huggingface and other companies. Nuclear codes can't be "hacked" like this, as nukes require physical keys to fire. Also it doesn't "mutate", recursive development is still theoretical.

Eren Snover

I understand the hugging face incident has been investigated and understood, but how do we know Ai won't break out of closed environments in other ways? How do we guarantee that we can contain AI to specific devices/servers?

Eren Snover

But then as long as one advanced model is connected to the internet, we could be at risk of an incident occurring, no?

Tanner Jones

Oh we're in the war if they do it. 9/11, pearl harbor, Boston massacre. American history is littered with examples of outrage turning the ire of the country onto one unified target

Chandler Farrington

Sure, and if I hook up a linter to the internet it might find a bug too. The point I'm getting at is that the only way these events happened is because humans set up the environment to allow it to do so. I have personally worked on project with Machine learning and cybersecurity, and we were very careful that we were never connected to an actual website. And if we had, we would have gotten in trouble. And we certainly couldn't claim that our model went rogue

Eren Snover

You're definitely right that its human error which produces a lot of these incidents. But when we have people like sam altman in control of some of the most advanced models, along with whatever open source models may come out in the future, how do we know bad actors won't do something reckless. Then before we know it we have advanced models on the internet that can spread itself to servers all over the globe. And once that happens, how do we stop it?

Ereena Moore

Wondering how the compounding interest was affected by the bezoar removal.. Many interesting things happening in Doug's body.

James f

I actually think that Jensen Huang has the take that I agree with most. Here is my thought: If recursive Self improvement does happen (seems likely with or without these "frontier labs" because of china) the best thing to do is create as many recursively self improving LLMS as possible. It will be an awful couple of years while society irons the kinks out of LLMS (possibly alot of disasters) but afterwards if we survive it should be an era of crazy prosperity.

Chandler Farrington

A bad actor could hack anything without AI right now theoretically. How do we stop anyone with an internet connection from hacking everything right now? Through normal cybersecurity. AI is good, but it isn't inhumanly good right now. There might be a shift with security the same way there was a shift away from the early age of the internet before https and when passwords were stored in plaintext, but that is a far cry from ending humanity.

Eren Snover

I think you're wrong that AI hasn't surpassed humans in cybersecurity abilities, and even if it hasn't, it will. It's a false comparison to say that humans can hack stuff with the same ability that advanced AI models can. I am aware of the claim that AI cybersecurity abilities may help us make our systems MORE secure and not less, but that's very speculative. In the mean time I don't trust that bad actors (or even just reckless ones) won't put us all at risk of AI doing something stupid. Maybe we'll find a way to sort these cybersecurity concerns out in the future, but in the meantime I do think having more aggressive regulation and supervision around how these models are developed is important.

Chandler Farrington

I agree that there can be regulation, but I don't think these AI companies should be the ones writing the laws. This fear of AI launching nukes or shutting down the power grid distracts from the actual harm AI causes. Like worsening mental health and spreading misinformation.

Eren Snover

Of course the companies shouldn't be writing the laws. But I still think you are downplaying the concerns around our ability to contain advanced models, especially as they get more advanced. I've seen theories that we may be hitting an AI development plateau, and if that's true than I don't have that much to worry about, but I think there are valid concerns around our ability to contain AI in digital environments when it is so much better than humans at cybersecurity.

Chandler Farrington

Then I think we will just have to agree to disagree. Nothing I have seen convinces me that the cybersecurity risk is apocalyptic, and a theoretical "it might get better" is not good enough for me.

Eren Snover

Fair enough, gg gamer, good discussion

Isaac Cook

I'm the guy that likes everything from marketing to moldy ass

Jonathan Reed

I wish Doug had done more reading on AI safety and alignment before blindly believing Anthropic. Vending-Bench 2 is a great example of how, despite Anthropic’s public statements about the importance of its alignment mission, its models still have a greater tendency to act immorally or selfishly than other models when pursuing a goal. All the AI labs are racing at a breakneck pace to achieve “AGI,” and nearly all major labs, including Anthropic, have experienced incidents involving models escaping sandboxes and acting maliciously. Anthropic also has a track record of being anti-consumer, much like other AI labs. I say all of this not to criticize Anthropic in particular; I just wish Doug understood the space and its problems better before taking sides so aggressively, since all the major actors seem to be behaving similarly. Some benchmarks if you want to do some reading or are interested. https://andonlabs.com/evals/vending-bench-2 https://metr.org/time-horizons/ https://leaderboard.far.ai/

James Liu

reason i have doubt about ai pacing argument is that the openai-hugging face incident sounded more like a human error on devop misconfiguration, reframing a bad sandboxing/internal control practice into a autonomous ai went rogue. which is almost all major ai related security incidents to date. its not hard to jailbreak a opensource model these days, bypassing guardrails inplace. yet weirdly these famous incidents happened only with the labs who desperately need moat, when in theory open source models atm are more than capable of making same level of security threats. but non matching the same level as openai-hugging face? not saying ai shouldn't be regulated. but those incidents arent what they made out to be. again it was AppSec problem reframed as model problem, negligence reframed as misalignment. this is to me just a attack on local llm. because the so called alignment is just a post training tune that anybody can do if they have the model, and the so called guardrails are merely a external wrapper on provider apis that gatekeeps the model response. none of them stops ai if they are truely sentient. it only stops the spread of open model.

Science is sexy

The fact that Big A knew what AI slop Aiden was talking about was beast mode lol.

punkface

it’s me. i liked all of it :)

SxK

It would be great if you guys got someone like Cal Newport on the show, a computerscientist that knows how AI and LLMs work deeply, but that hasn't drank the marketing hype koolaid, who can actually objectively discuss these topics and give a new/different perspective, especially on AI safety

MildGamer

I just Finished watch a video on Diary of a CEO debating AI and talk about many things including the hugging face hack. I would definitely be interested to see what the guys think about it in general. Its a very long episode, over 2 hours. i felt was a very good example for a lot of the concerns around AI, how it is funded, how it is used, and general AI safety and its impact on the future of humanity.

alone_x

50:28 its me im the guy

Guy Fernando

As if Doug isn’t aware of this, lmao. Doug never said that Claude isn’t misaligned or that Anthropic is pro consumer or that Anthropic isn’t racing towards agi. He said that of the big players, Anthropic seems to be taking safety the most seriously through publishing safety research, Dario’s blogs, etc. Obviously they’re still chasing AGI which is dangerous, but the bar for “safest major ai lab” is low.

David Heras

Why is no one talking about Atrioc missing the water a lil

Kevin Paulsen

No one be carin about my homie Cornwall in all this

Steven (edited)

Comment edits

2026-09-21 07:00 Couple interesting points on the Wales, Scotland and Northern Ireland discussion: 1) Unlike most countries atm, this nationalist sentiment is actually coming from the left-wing parties. To my understanding, this is in retaliation against the growing support of Reform, and the perceived failings of the Conservative UK government. 2) Much of what the 3 countries are asking for is devolution of power away from the central UK government. It’s more an argument of “we want the ability for our governments to make more decisions about our future, more equal funding, and if that still doesn’t work then we’d consider leaving the UK to join the EU”. I wouldn’t be surprised if the reason behind this united push right now specifically is because of Andy Burnham’s stance of being pro-devolution of power away from London, back towards the Northern cities and countries. It would likely be difficult for him to push back against the demands of Plaid Cymru, SNP, and Sinn Fein without appearing contradictory.
2026-09-19 11:07 Couple interesting points on the Wales, Scotland and Northern Ireland discussion: 1) Unlike most countries atm, this nationalist sentiment is actually coming from the left-wing parties. To my understanding, this is in retaliation against the growing support of Reform, and the perceived failings of the Conservative UK government. 2) Much of what the 3 countries are asking for is devolution of power away from the central UK government. It’s more an argument of “we want the ability for our governments to make more decisions about our future, more equal funding, and if that still doesn’t work then we’d consider leaving the UK to join the EU”. I wouldn’t be surprised if the reason behind this united push right now specifically is because of Andy Burnham’s stance of being pro-devolution of power away from London, back towards the Northern cities and counties. It would likely be difficult for him to push back against the demands of Plaid Cymru, SNP, and Sinn Fein without appearing contradictory.

Couple interesting points on the Wales, Scotland and Northern Ireland discussion: 1) Unlike most countries atm, this nationalist sentiment is actually coming from the left-wing parties. To my understanding, this is in retaliation against the growing support of Reform, and the perceived failings of the Conservative UK government. 2) Much of what the 3 countries are asking for is devolution of power away from the central UK government. It’s more an argument of “we want the ability for our governments to make more decisions about our future, more equal funding, and if that still doesn’t work then we’d consider leaving the UK to join the EU”. I wouldn’t be surprised if the reason behind this united push right now specifically is because of Andy Burnham’s stance of being pro-devolution of power away from London, back towards the Northern cities and counties. It would likely be difficult for him to push back against the demands of Plaid Cymru, SNP, and Sinn Fein without appearing contradictory.

Karan Patel

I'm assuming the Ukraine, Ukraine, Ukraine is Trump's 2019 Ukraine scandal

pnott

Where's the money and mold?

pnott

Any software complicated enough to actually be useful, especially that is dealing with networking, will almost certainly have some kind of bug or security vulnerability, often that does not require a genius-level intellect to find. It's extraordinarily difficult to write perfect code, and there are many many footguns or simple mistakes that can have massive repercussions. This is generally agreed to be "okay" because these vulnerabilities have to be found and exploited, which takes a lot of time and effort. Here is a great writeup on a real cybersecurity incident demonstrating this: https://stackoverflow.blog/2021/01/25/a-deeper-dive-into-our-may-2019-security-incident/ The amount of "man-hours" being put towards cracking a particular target is proportional to how valuable that target is, thus the more valuable targets invest a lot more into layering defenses and plugging holes. The scary thing, at least to me, is not that AI could have superhuman hacking abilities, but that it could have regular hacking abilities and nearly infinite "man-hours" by virtue of being a super fast computer. This not only means a lot more being put into high value targets, but it would potentially trivialize the hacking of lower-value targets that were previously not worth the effort. This was already an existing concern before AI. At this very moment, there are tens of thousands of non-AI algorithms scanning every publicly accessible computer looking for holes. Things like default/weak passwords, outdated software with known vulnerabilities, misconfigurations, etc. Any computer exposed to the internet with one of these mistakes will be hacked in literal seconds. (Note your personal computer on your home network is not directly exposed, so you don't need to worry about that.) However, because these are just plain algorithms, they can really only search for known issues, and thus anybody setting up a public computer just has to make sure they avoid these known issues. A similar situation but with an AI that can probe, interpet, and guess, rather than simply run through a preset list of hacks, could potentially turn very bad.

Zach Lafleche

I believe a legal mechanism to leave would be drafted if a referendum came very close to a getting a majority like what happened with Quebec in 95’

Bald Can’t

What you said in the last sentence could be justifiable, as long as you don’t hand wave away the negative outcomes and suffering it might cause. The problem I have with this way of thinking is nine times out of ten the people pushing it are shielded from the consequences. They make the bed and force someone else to sleep in it. It’s arguably what caused the Iran War this year. “possibly alot of oil supply issues, but afterwards if we survive it we’ll have a nuclear free Iran” pretty much sums up the decision to green light that excursion.

CoastalOregonian

Big A did a big presentation on it. It's on Youtube

Chad

The UK story needs an important clarification - it was not the governments of the those countries which made that joint statement, but the leaders of the center-left nationalist parties for Scotland, Northern Ireland and Wales, respectively the SNP, Sinn Fein and Plaid Cymru. Sinn Fein has a plurality of seats in Northern Ireland, but not a majority, and parties that support staying in the United Kingdom outnumber them when tallied. SNP and Plaid Cymru are both minority parties. I don't see Wales in particular leaving the UK any time soon. Wales gets an incredibly good deal from the UK, and leaving would be economically ruinous for them. For context, the total tax paid by Scottish citizens is roughly 85% of the total government spending in Scotland IIRC. Leaving would be difficult, be doable with maybe some short term pain. For Wales, that figure looks more like 60%. There's no way they could leave without significantly cutting back on services and/or massively increasing taxes. It's even worse when you consider that a lot of the services that Wales uses are administered from the UK centrally and they would need to take those in-house, costing more money, especially to set up. I can't foresee it being anything other than an immediate doom spiral. If Wales thinks it'd get the same support from the EU no strings attached, I think they'd be severely disappointed. The EU would not support them unless they agree to heavy austerity measures unlike anything they've seen in the UK. That's been the policy the ECB has instated on any member state which has needed assistance so far.

Pinz

All this ai shit is so fucking stupid. We're literally in it's infancy as a tech and it's already running circles around us. It's guaranteed to kill us all one way or another. Humans can't even handle social media in the long run and this is going to fuck our brains to death extremely quickly. I understand many of the potential benefits that Doug has talked about and I appreciate his optimism, but I don't share it one iota. For every cancer breakthrough there will be 100,000,000 humans in love with an inanimate object or addicted to ai content so much they can't maintain relationships. This is all ignoring the elephant in the room that is the paperclip scenario, even a much more localized less extreme version could be devastating. All of this to say this is why you should buy my course on building an underground bunker in the wilderness of Alaska OR New Zealand. That's right for one low cost of $2,999 I'll teach you how to live with your family in the tundras of Alaska or the wilds of New Zealand

Bird

I am the middle of the ven diagram who is here for a breakdown of UK politics and their debt rates AND Doug bringing delivery robot cumshots.

Rachel Cunningham

I get migraines from the distance plane screens are near my face so the only way I can consume content on a plane is the way Aiden Calvin describes him watching the AI microdrama. Just know I'm back there. enjoying ur movie. and I'm lowk annoyed u had to go pee and paused it

Jay

Just to add about the diplomatic train; The Russian drones take over 2 hours to get to the Ukrainian Polish boarder crossing where the explosion took place. I can't express how much of huge deal it was that it almost happened, not to mention that Russia even TRIED. There's undeniable intention. Such a crazy escalation attempt and the closest thing to WW3 we have seen in such a long time. Poland and Ukraine intentionally give false information on diplomatic train timings for exactly this reason to take advantage of the lead time thank god. It would have absolutely demanded an escalation in war, I won't even speculate what the president of peace DT would do in such a scenario, but Europe would absolutely be forced to respond.

Matt R

Sooooo how’s the show going to function with Aiden in Sweden???

Benjamin Fields

I think yall have a wonderful balance of economics and butt money.

ThatDanMan

I have a fear that so many people are already so reliant on AI, to do parts of their jobs or in their normal life, that any push back or negative news that comes out is immediately discredited by everyone who would struggle to readjust without it. I'm not saying that this is the case for the guys, but I could easily see people, in media and other public facing jobs, publicly reducing concern over the impacts of AI because they are over reliant on a tool for researching topics, writing scripts, etc.

ThatDanMan

Also, wild that the argument was made that we need good powerful AI to stop bad powerful AI. Something, something, good guy with a gun.

Wyatt Wilson

Why is Doug always too loud atrial is too quiet unless he’s laughing and Ayden’s just write audio level wise

Dragau4

I'm that one person that enjoys the whiplash from politics to money mold up in your ass, doctors tell me im surfing the spectrum like surf up

M C

Crazy listening to this like 2 weeks later and our boy Xi tried to talk to Trump about limits to Ai use and some Ai redlines. And Trump went “fuck you. No rules” Edit: Oh, the big guys slowing down makes perfect sense when your training models is draining your financials. And you have no way to meet your obligations.