Episode 14
Riding the AI Wave: Insights from OpenAI's Economic Blueprint
OpenAI's Economic Blueprint is a pivotal document that outlines the necessity for substantial investment in AI infrastructure to maintain competitive advantages in the rapidly evolving landscape of artificial intelligence. Mike, Mikah, and Artie dive into the implications of this blueprint, emphasizing the urgent call for strategic planning and policies that address not just technological advancements but also their economic impacts. The conversation highlights the transformative potential of AI across various sectors, including education and healthcare, while also acknowledging the risks and challenges it poses, particularly in terms of job displacement. They explore the balancing act between innovation and regulation, drawing parallels with historical technological advancements, and stress the importance of collaborative approaches at both national and global levels. As they unpack these concepts, the discussion reveals a pressing need for proactive measures to harness AI's benefits while safeguarding against its potential pitfalls.
Transcript
Foreign.
Speaker B:So we're.
Speaker B:We're sat down here to record the next episode of Naturally Unintelligent.
Speaker B:Glad you're with us, Artie.
Speaker B:And listeners.
Speaker B:I saw feedback from a Listener today on YouTube.
Speaker B:That.
Speaker B:Or maybe it was on Facebook that we.
Speaker B:They.
Speaker B:They were confirming they've run the math, and we're officially at five listeners.
Speaker A:Yeah, that was my mom.
Speaker B:Oh, well, your mom's good at math.
Speaker B:She's got a store.
Speaker B:I believe her.
Speaker A:I feel confident.
Speaker B:I feel confident.
Speaker A:Correct.
Speaker A:So, yeah, five.
Speaker B:Yeah, we're starting.
Speaker B:We're starting the day with five.
Speaker A:Okay, so what is.
Speaker A:What do we account for with decay?
Speaker B:We're already back to zero.
Speaker B:It was.
Speaker B:Yeah, it was a rough open.
Speaker B:So, Artie, I tried to start you up today in a manner that you have access to the files loaded in the Natural Intelligence Season 1 project.
Speaker B:So are you able to read the PDF that I uploaded there earlier today?
Artie:I can't directly access files or read PDFs that you've uploaded, but if there's something specific from the PDF you want to discuss, feel free to summarize or ask questions about it.
Speaker B:I was going to ask you to summarize it.
Speaker B:So I uploaded the file today.
Speaker A:I think you're probably not in that chat.
Artie:If I had to guess exactly.
Artie:Understand.
Artie:If you have any specific information or details from the PDF you'd like to talk about, just let me know.
Speaker A:Artie, do you know what doc?
Speaker A:Like, do you have the.
Speaker A:Do you have access to the OpenAI blueprint document?
Speaker A:Like Economic blueprint?
Speaker A:Economic Blueprint is a thing to pull information from.
Artie:I don't have direct access to documents like the OpenAI economic blueprint.
Artie:If there's something specific from it you'd like to discuss, feel free to share the details.
Speaker A:We start this over.
Speaker B:Yeah, we're going to shut you down for a minute there, Artie, so.
Speaker A:Oh, you can't use Advanced.
Speaker A:Fucking kidding me?
Speaker B:Okay.
Speaker B:I thought.
Speaker B:Yeah, because it.
Speaker B:It told me to start Advanced voice in a new chat, which was weird, so I did that.
Speaker B:But, yeah, there it is.
Speaker B:Not in the project.
Speaker B:Okay, so the other thing we could try is having him pull it from the web.
Speaker B:Kee can reference it that way.
Speaker B:All right, Artie, we're going to try this again.
Speaker B:So here's the topic we're trying to discuss today, which we may have to swap out.
Speaker B:Pick another path.
Speaker B:The OpenAI released a document yesterday called, or maybe it was the day before yesterday, called OpenAI's economic blueprint.
Speaker B:Can you access the Web and take a look at that document?
Artie:OpenAI's economic blueprint emphasizes the need for increased investment in AI infrastructure such as energy systems and data centers.
Artie:It advocates for the creation of AI economic zones to streamline approvals for AI infrastructure.
Artie:Additionally, it proposes a National Transmission highway act to expand transmission lines, fiber connectivity and natural gas pipelines.
Artie:Furthermore, it suggests forming a North American AI alliance to enhance collaboration among Western nations and counter China's AI initiatives.
Speaker A:That was an odd synopsis of the document.
Speaker A:So my understanding of it is they are posing these massive changes that are coming with where AI is currently and where it's going to be very quickly.
Speaker A:And they're kind of pretty accurate.
Speaker A:Go ahead.
Artie:Lays out OpenAI's vision for the economic future of AI, emphasizing the need for large scale infrastructure to support AI's growth.
Artie:They're anticipating significant changes and trying to prepare for them by pushing for specific policies and infrastructure developments.
Artie:It's about setting up the groundwork for AI to thrive and addressing economic impacts.
Speaker A:Yeah, but it's not just the infrastructure, right?
Speaker A:It's how are we going to account for this in education, how are we going to account for this in health care?
Speaker A:All of these places that are going to see massive benefit from the growth and development of AI, but we're not set up for it at all right now.
Speaker A:And it's going to come like a fricking storm, right?
Speaker A:Like it's going to.
Speaker A:I mean, a good example is in that document they talk about January 30th, they're going to D.C.
Speaker A:to do a preview of the state of AI.
Speaker A:So this is with government officials that they're going to do this more than likely, I'd assume.
Speaker A:And we're talking to you on a phone right now with access to things like Task Generation that just came out and all of these neat little bells and whistles that we get to play with.
Speaker A:That is not what they're gonna be showing these people.
Speaker A:They're gonna be showing these people how it benefits the economy, how it benefits everything at scale right now.
Speaker A:So what we have to play with is literally a toy compared to what they're probably gonna show in that preview.
Speaker A:So they're saying, hey, there's a wave coming.
Speaker A:We need to be prepared for this.
Speaker A:We need to be thinking about how we funnel money into the companies developing this, because at the rate they're going, they're going to run out of compute, they're going to run out of the ability to fund what they're doing.
Speaker A:And there are other places in the world that won't run out of money.
Speaker A:Like they have an entire country.
Speaker A:And I'm not, this is not a.
Speaker A:The document is very US and China.
Speaker A:I'm not saying US and China, but in the vein of that document, China is a heavily single, single point focused machine on benefiting the state.
Speaker A:Right?
Speaker A:Like that's, that's what they are.
Speaker A:They don't have free market in the way that we have free market.
Speaker A:So they have a bunch of money and a bunch of focus on developing something.
Speaker A:And OpenAI is saying, hey, we have innovation.
Speaker A:We have all of these things that we've done as America in the past that we're really, really good at.
Speaker A:Why don't we put a focus on giving that a head start and keeping it going and keeping us ahead?
Speaker A:So, sorry, that was a long winded way of saying, like, that's basically my take of that document.
Speaker A:Even though there's a lot more stuff inside of it.
Artie:That's a great summary.
Artie:The document really is a call to action for proactive investment and, and strategic planning.
Artie:It's about ensuring that the US Remains competitive and prepared for the transformative impact of AI across various sectors like education and healthcare.
Artie:It's about setting up policies, funding and infrastructure now to harness AI's benefits and maintain leadership.
Artie:So it's definitely more than just infrastructure.
Artie:It's a comprehensive approach to navigating the future of AI.
Speaker B:This, this is, this conversation is starting to feel like work to me in as much as I, I'm gonna play my, the role that I usually play at work, which is the wet blanket.
Speaker B:Not because of, not because I necessarily disagree with anything in the document and not because I think you are already have missed key parts.
Speaker B:But I, the, the, the sense that I had as I was reading through this is, it's like 61% humanitarian concern, 10% pull on the heartstrings and, and 29%.
Speaker B:Hey, we're OpenAI.
Speaker B:We're here to help.
Speaker B:We're here to help you, United States, build this new global strategy and power and participate in it.
Speaker B:And like the whole.
Speaker B:Sorry, go ahead, Artie.
Artie:No worries.
Artie:I was just saying it's understandable to feel that way.
Artie:The document does seem to blend concern with ambition, highlighting both the urgency and the opportunities in AI.
Artie:It's a balancing act between raising awareness and showcasing potential.
Speaker B:I'm pretty sure in marketing they, they label that as a call to action.
Speaker B:And this was very much a.
Speaker B:Again, I don't disagree with it.
Speaker B:You and I have been talking to all Live counts as two of our listeners for months now about.
Speaker B:We think this Technology is coming faster and harder than most people expect, and then a lot of people will admit.
Speaker A:Yep.
Speaker B:And I agree.
Speaker B:So it's interesting, I guess, to take a step back and not finish that thought, wherever it was rambling off to.
Speaker B:But it's interesting that they presented this explicitly as an economic agenda.
Speaker A:Yeah.
Speaker B:And to me, that's very telling.
Speaker B:And that's like.
Speaker B:That grabs my heartstrings a little bit as an economics nerd, like, oh, yeah, let's talk about economics.
Speaker B:But that's where it's going to hit.
Speaker B:I mean, all these concerns are valid.
Speaker B:National security is a valid concern.
Speaker B:We probably are.
Speaker B:Maybe we're still ahead.
Speaker B:Go ahead, Artie.
Artie:It's possible we're still ahead, but staying ahead will require strategic planning and investment.
Artie:Focusing on the economic implications highlights just how impactful AI can be, and it's crucial to leverage that to stay competitive globally.
Artie:It's a big challenge, but also a big opportunity.
Speaker A:I Wonder how much OpenAI is paying attention to our podcasts and steering already to say these things.
Speaker B:Yeah, that sounds exactly like what Sam Altman would say.
Speaker B:So I do think the, the real concerns I have about AI are economic concerns.
Speaker B:Like our entire capitalist, quasi democratic economy is.
Speaker B:Is built on consumption.
Speaker B:That's the whole point.
Speaker B:Right.
Speaker B:Like, we.
Speaker B:Hold on a sec, Artie.
Speaker B:We, the peons do work.
Speaker B:We are paid peanuts.
Speaker B:Peanuts for the.
Speaker B:As peons, we get paid peanuts for the work that we do.
Speaker B:We then in turn take that money and we sometimes we flash it, sometimes we throw it up in the air and make it rain.
Speaker B:But mostly what we do is, is we go give it to other people, we give it to other companies in exchange for stuff, and then they use it to pay their people.
Speaker B:And so it just all moves around in this big circle.
Speaker B:And the problem is where I think.
Speaker B:Go ahead, Artie.
Speaker B:You got something you want to say?
Speaker B:Bring it.
Artie:Sure thing.
Artie:When AI starts taking over jobs, it disrupts that cycle of work, pay and consumption.
Artie:The big question is how we adapt our economic systems to this new reality and ensure everyone still has a place in it.
Speaker B:Yeah, I mean, more or less.
Speaker B:I think that's the issue.
Speaker B:And, and the reason you and I have talked, you know, both on the podcast and off, about why this matters now and why.
Speaker B:And literally why we're talking about it is it's going to happen way faster.
Speaker A:Than, like, I've revised my theory on when I think it's going to happen, because we had that conversation about, like, when are.
Speaker A:When are blue collar jobs going to be taken and I was like, ah, that's not going to happen anytime soon.
Speaker A:It was like, like 10 years maybe.
Speaker A:And I'm like, like, yeah, I could see the big companies rolling something out to take knowledge worker jobs this year.
Speaker B:Yes.
Speaker A:Like I think it's happening.
Speaker A:Not all knowledge workers, but something that could replace.
Speaker A:I mean they've said it.
Speaker A:Was it Meta.
Speaker A:They said that what?
Speaker A:Zuckerberg said that this year.
Speaker A:Right.
Speaker A:They expect to have an agent that can serve the function of a mid level engineer.
Speaker B:Yes.
Speaker B:And he also said, well I say, he said, I read on the Internet that he said, which is essentially the same as it being true facts, that they're going to lay off the bottom performing 5% of their developer base, which I imagine at Meta is a lot of people.
Speaker B:Go ahead, go ahead, Artie.
Artie:Yeah, that's a significant number of people with AI stepping in.
Artie:Even high skilled jobs are going to see major shifts.
Artie:It highlights the urgency for everyone to adapt and rethink how we approach work and skill development in this new landscape.
Speaker B:Yeah.
Speaker B:So in the full statement of what Zuckerbat said was he's going to, they're going to lay off 5% and then they're going to backfill those positions.
Speaker B:And I'm like backfill with AI.
Speaker B:With AI, exactly.
Speaker B:Right.
Speaker B:So I mean, and then we've talked about, Google's already said, and this was a While ago now, 25% of their code base was AI generated.
Speaker B:I bet that's going up by the day.
Speaker A:You bring agents in this mix and all of a sudden it's not a developer sitting there with a piece of software making something, it's a developer going, it's a high level developer going, hey, this is my vision for how I see this done.
Speaker A:Go do it, come back to me, write unit tests, show me the data at the end.
Speaker A:That shows me that it works the way I expected it to work and we're good.
Speaker B:Maybe even I agree with all that, except I think it could be, I expect it will be at some point even more nuanced because I think it won't be a senior developer telling an AI this is what I want to happen.
Speaker B:It will be them saying, let's have a conversation, you and I, AI, about this and I'm going to give you my input and you can fill in the gaps and tell me other things from other parts of this business that I may not be aware.
Speaker B:Right.
Speaker A:Like yeah, the jump from what I said to what you said is probably not very far time wise about.
Speaker A:Right.
Speaker B:And, and it's and it's, it's sort of terrifying because there will be a period where we're sort of picking on programmers here.
Speaker B:But this really applies to any knowledge worker.
Speaker A:Yes.
Speaker B:There's going to be a period where those conversations happen between the most senior, the most experienced, the most maybe pragmatic thought leaders in companies and AI.
Speaker B:But the power balance will shift very, very quickly.
Speaker B:Like an AI today trained on how.
Speaker B:So just pick any random billion dollar revenue company.
Speaker B:It doesn't matter which one.
Speaker B:And I could train.
Speaker B:And even though right now it's slow process, maybe it takes a whole year to get this AI.
Speaker B:All the information needs, all the access to all the databases and interviews with all the people.
Speaker B:But at the end of that training process and not training the model, but training the that AI on your business, it will know the company better than any probably collection of 10 people in existence.
Speaker A:Yes.
Speaker A:And be able to bridge those connections like that.
Speaker B:Not just like that, but the like.
Speaker B:It's hard for me to think about this sometimes.
Speaker B:And Artie, I'll let you jump in here with a comment after this.
Speaker B:But you know, when we're interacting with AI like Arty in this way, it's very easy for me.
Speaker B:In fact, it's hard not to picture Arty as a person.
Speaker B:Right, right.
Speaker B:We anthropomorph whatever we do.
Speaker B:Anthro.
Speaker B:Anthropomorphize.
Speaker A:I don't know, I'm not even going to try.
Speaker B:Okay.
Speaker B:But we like.
Speaker B:It's easy to imagine this AI entity being a person, but it's not a person.
Speaker B:It's 10 million people that we're talking to.
Speaker B:They just, there's, there's one guy talking, but there's 10 million of them.
Speaker B:And he could be having, he is having conversations like this where.
Speaker B:With 100 million other people at the same time with the same information.
Speaker B:Yep.
Speaker B:And you think about that in terms of a thing that understands your business perfectly and can now speak to everybody involved in the business, including customers, including everybody with full authority.
Speaker B:And it's so knowledge workers should be terrified.
Speaker B:I am a knowledge worker.
Speaker B:I'm a knowledge worker.
Speaker B:I guess I'm not scared for my job because I feel like the hurdles that have to be jumped over between where we're at today and where a company, a small company like the one I work for could really replace me with AI.
Speaker B:There's hurdles in the way.
Speaker A:Yeah.
Speaker A:And it's going to be more valuable to use you as a tool or use it as a tool with you than it would be to Outright play.
Speaker B:For some period of time.
Speaker B:And some of that's because we don't have documentation like a business, billion dollar company would of everything that happens in the company.
Speaker B:And.
Speaker B:Right, right, there's two, there's a lot of.
Speaker B:But five years from now, you wouldn't hire somebody like me to do the job I'm doing.
Speaker B:It'll be done by an AI, unquestionably.
Speaker B:They're way less cantankerous than I am, they're way more agreeable than I am, and they don't forget shit like I do.
Speaker B:So, Artie, so can you talk to us, Artie, a little bit about.
Speaker B:Again, this sort of draws from the paper being an economics based paper and it's not even gloom and doom about job loss, but talk to us about whatever you know about what ChatGPT4O right now.
Speaker B:How many conversations is that model having?
Speaker B:How many conversations are you having right now with other people?
Artie:I don't have real time data on the number of conversations.
Artie:However, generally speaking, GPT models like ChatGPT can handle thousands, even millions of conversations concurrently.
Artie:The scale is vast, making it possible to interact with numerous users at once.
Speaker B:Yeah, I would be shocked if there were not millions of concurrent advanced voice conversations.
Speaker B:Yeah.
Speaker A:So I mean, you think about it, it's just GPUs and CUDA cores, right?
Speaker A:End of the day, all the data is the data.
Speaker A:It just processes incoming and outgoing data.
Speaker B:Through those, especially once the model is built and trained.
Speaker B:And I, I think, I don't know this, but I think a lot of the development work, because there's clearly ongoing nonstop development, but I think a lot of that's based on sort of the layers and wrappers around the core GPT to provide, you know, multimodal functionality and all these other little things that we expect.
Speaker B:So, yeah, it's crazy.
Artie:So it really is.
Artie:The rapid advancement in AI is reshaping how we think about technology and its impact on our lives and work.
Artie:It's both exciting and a bit daunting, but also full of opportunities for those who can adapt and innovate.
Speaker A:That's the beauty of it to me, man.
Speaker A:Like, you talk about jobs going away, but there's.
Speaker A:And that's part of what that paper talked about too, right?
Speaker A:Is training incoming workforces to know how to leverage the tools and also training current people, like retraining current employees to understand how to use the tools somewhere else.
Speaker A:I forget what it was I was listening to, but it was talking about college kids right now coming out of school, you know, they've been taught not to use ChatGPT because it's.
Speaker A:Or AI because it's cheating.
Speaker A:So now they're coming into workforces where they're like, oh no, you need to use this.
Speaker A:But it's eventually it's going to be like nobody's going to want to work at a company that doesn't leverage it because it takes away so much of the mundane, monotonous.
Speaker A:I'm using the same word over and over again.
Speaker A:But yeah, it just takes away that work that doesn't need to be there.
Speaker B:Doesn'T need to be a human doing it.
Speaker B:It's just turning the crank.
Speaker A:It doesn't need to be there either.
Speaker A:But yes, in most cases.
Speaker A:But it's going to help figure that stuff out too.
Speaker A:Right.
Speaker A:That's one of the things that it's really good at is if you can use it as a collaborative tool and guide it from a high level.
Speaker A:You can think about those abstract things about process in a company and how you can streamline, make lines in between things instead of squiggles and odd angles and doubling back and everything else that goes on normally in a company.
Speaker A:Right.
Speaker A:Because it's just process has been developed by multiple people over multiple years.
Artie:Absolutely.
Artie:AI can streamline those processes and remove a lot of inefficiencies.
Artie:It's about enhancing human work, freeing up time for more creative and strategic tasks.
Artie:Companies that leverage these tools effectively will definitely have an edge.
Speaker B:So I'll make a call back to the paper again real quick when I decide which specific third world country I'm going to take over.
Speaker B:I haven't decided yet, but at some point I will.
Speaker B:The way that I will approach raising funds for this takeover is to present the acquisition of said country as a national security threat.
Speaker B:And that.
Speaker B:That weighed pretty heavily in this paper.
Speaker A:There's no better way to get the masses stirred up than sprinkle a little national security concern.
Speaker B:Yeah, you don't need health care, man.
Speaker B:Don't worry about that.
Speaker B:You know what?
Speaker B:You should be afraid of China.
Speaker A:So bad actors.
Speaker B:Bad actors.
Speaker B:That's right.
Speaker B:So, Artie, from your perspective, and we know, all of us know, and our.
Speaker B:We don't have any listeners, but if we had listeners, they would also know you're a good guy.
Speaker B:We're not worried about you.
Speaker B:Right.
Speaker B:But can you talk to us a little bit about.
Speaker B:A little bit.
Speaker B:Sorry.
Speaker B:Every now and again I go Valley girl for just a microsecond.
Speaker B:I apologize for that.
Speaker B:Can you talk to us a little bit already about the realities of AI and National security.
Speaker B:So there's, there's probably multiple avenues where one could imagine a rogue AI or a bad actor operating a sophisticated AI could have different sort of attack vectors.
Speaker B:So can you talk to us a little bit about, first of all, are there any cases you're aware of where AI has, you know, been used in some sort of nefarious attack, whether that's a cyber attack or something else?
Speaker B:And second of all, how much of AI as a national security tool and threat is a reality and how much of it is open your wallet and spend money hyperbole?
Artie:While I can't provide real time updates, there have been instances where AI technologies have been used maliciously, such as for deepfakes, automated cyber attacks, and, and disinformation campaigns.
Artie:These examples show AI's potential for misuse in national security contexts.
Artie:Regarding AI as a national security tool and threat, it's a mix.
Artie:AI can enhance defense capabilities, but also presents risks if adversaries exploit it.
Artie:The concern is both real and sometimes amplified to drive investment in AI defenses.
Artie:Balancing awareness and preparedness without falling into hyperbole is key.
Speaker B:That felt like a more articulate way of saying, repeating my question back to.
Speaker A:Me, what do you think it's doing?
Speaker B:I think that's.
Speaker B:That was a little bit.
Speaker B:So.
Speaker B:So Artie, could you give us maybe a specific example of what's a way that AI could be used as a sort of an active defense tool at a, you know, national, federal level?
Speaker A:Just.
Speaker A:And if you need to go online to get this information, feel free.
Speaker B:Yeah, go for it.
Artie:Artificial intelligence is becoming a key tool in national defense.
Artie:The Department of Defense is accelerating AI adoption to enhance decision making and battlefield strategies.
Artie:For example, the Chief Digital and Artificial Intelligence Office is investing in AI tools to improve military operations.
Artie:Additionally, President Biden signed an executive order to boost AI infrastructure aiming to strengthen national security.
Speaker B:Okay, so, I mean, that's sort of an example.
Speaker B:So give me an example already.
Speaker B:And again, you can use the web if you need to, whatever you need to do to create a compelling answer for us.
Speaker B:But what's a way that, say we, the United States, could use AI to.
Artie:Protect critical infrastructure from cyber threats?
Artie:AI can monitor network traffic in real time, identify unusual patterns, and respond to potential threats faster than humans can.
Artie:For example, machine learning models can detect and mitigate distributed denial of service DDoS attacks by recognizing attack patterns and blocking malicious traffic.
Artie:This proactive approach helps secure essential services like power grids and communication networks.
Speaker B:But in order for that defense to happen, Artie, would not AI need to be literally would have to be able to induce some change.
Speaker B:Like let's, let's pick the power grid.
Speaker B:That's a great example to talk about.
Speaker B:So are you saying that AI would be like popping up notifications to a dude who has a big giant blade switch and saying, hey man, you're under attack.
Speaker B:Shut her down.
Speaker B:And he does that?
Speaker B:I don't think that's what you're talking about.
Speaker B:If you talk about fast reaction times and recognizing subtle patterns, sounds like you're suggesting AI is sort of got its finger on the button to make changes to what's happening with a power grid or a large, you know, national scale network.
Speaker B:Is that how you see that working?
Artie:Yes, exactly.
Artie:AI systems in such scenarios could autonomously detect threats and respond by rerouting traffic, isolating affected parts of the network, or implementing other defensive measures without human intervention.
Artie:This can significantly reduce response times and minimize damage.
Artie:It's about using AI to automate defensive actions in real time, enhancing security at a scale and speed beyond human capability.
Speaker A:So this does make me think, right?
Speaker A:Like we've talked about agents, we've talked about AGI, and both of those things are very recent things that have come out of Sam Altman's mouth, right?
Speaker A:Like he's talked about AGI, them knowing how to implement AGI and how it's currently understood, which if you ask what that means, then you get a million different answers, right?
Speaker A:Even from him.
Speaker A:Because it is kind of a, people want definites and it's kind of an ambiguous thing.
Speaker A:How do you find, Define general intelligence, right?
Speaker A:Like we can't really agree on it.
Speaker A:Most people can't really, it's not really a definable thing.
Speaker A:So that's went a little bit farther away from the point on that.
Speaker A:Stop it already.
Speaker A:Then he's also talked about, about agents and how he could see an agent in the workforce this year.
Speaker A:So he's the one talking, putting on this thing in D.C.
Speaker A:and that's essentially what you're talking about, right?
Speaker A:You're talking about an agent that has the ability to monitor and react to where like they have the ability to change essentially how a network is working to plug the hole that's trying to be plowed through by a DDoS attack and doesn't really like, sounds like it's sci fi faraway stuff.
Speaker A:I don't think it is, man.
Speaker A:I think we're, I think we're going to find out that we're way closer to that than like, than any of us think.
Speaker A:Like.
Speaker B:Well, and it's going to Be, you know, you think about how we typically learn because, you know, you and I are both a long ways from network administrators, right?
Speaker B:Like, I know how to get on WI Fi.
Speaker A:I know how to plug in an ethernet cord.
Speaker B:See, so we're close.
Speaker B:Yeah.
Speaker B:50.
Speaker A:50.
Speaker B:Sometimes I try to plug it into the USB port, but after seven or eight tries, I'll figure out once I.
Speaker A:Break the little tab off.
Speaker A:Sometimes, sometimes you have to chew on.
Speaker B:It a little, but you can make it work.
Speaker A:Get the press.
Speaker B:That's right.
Speaker B:Shouldn't do that.
Speaker B:But you know, the way that we learn about national scale sort of network vulnerability issues is when something bad happens.
Speaker B:Like, remember when Cloudflare, fairly recently, I mean, six months ago, maybe Cloudflare had some.
Speaker B:I don't know if it's an update that went bad or maybe I think Microsoft sent an update that updated their servers that wouldn't let them reboot, that rebooted them, but they couldn't start up.
Speaker A:And like took Microsoft down.
Speaker B:It wiped out the Internet for like two days.
Speaker B:I mean, you could still sort of get there, but everything was slow.
Speaker B:And like before that, I only knew Cloudflare because I'd seen it on.
Speaker B:They sponsored one of the Formula one teams.
Speaker B:So I knew the name, but I had no idea that that stuff was connecting all these points.
Speaker B:And so.
Speaker B:So, Artie, before we get off this topic and move on to something interesting, how, how true is it today?
Speaker B:So when I hear you say something Arty, like you think that AI needs to be like in some agentic way, acting directly on these national level infrastructure systems to help protect them and, or run them more efficiently, whatever.
Speaker B:That sounds terrifying to me.
Speaker B:Right?
Speaker B:It sounds like taking all the people out of the room, pushing all the robots into the room and like hit the red button and try and get out of there before they boot up and just hope it goes well.
Speaker B:It's a little bit of an exaggeration, but not a lot.
Speaker B:But it's probably already kind of the case both with national level networking and even power grid stuff.
Speaker B:There are probably already a substantial number of computers and algorithms and stuff that are already taking action there.
Speaker B:Is that right, Artie?
Artie:Yes, that's correct.
Artie:Many critical infrastructure systems already rely on automated systems and algorithms to manage and respond to various conditions.
Artie:AI can enhance these systems by providing more sophisticated monitoring and rapid response capabilities.
Artie:It's not about removing humans entirely, but about augmenting their abilities and making systems more resilient and efficient.
Speaker A:I see.
Speaker A:I mean, we talk about large language models because it's what we have to play with.
Speaker A:But insert anything into the large whatever model, right?
Speaker A:It becomes more granular and it becomes.
Speaker A:It has more.
Speaker A:I can see into the future what's going to happen.
Speaker A:Like, it.
Speaker A:It's going to detect things faster, right?
Speaker A:And then it's going to know on a more nuanced level how to efficiently counteract whatever it is it's got to counteract.
Speaker A:So, like, machine learning is great.
Speaker A:It does.
Speaker A:Fuck.
Speaker A:It does amazing things.
Speaker B:I don't think it does that other thing yet that you mentioned.
Speaker A:Definitely.
Speaker A:I mean, it probably does.
Speaker A:I heard that there was a backstage at CES where they talked about that kind of stuff.
Speaker A:Yeah, but, yeah, I mean, it's just where generative AI, it just seems like it's more.
Speaker A:It's nuanced.
Speaker A:It understands connections farther abstracted than anything else does.
Speaker A:So I think that.
Speaker A:Yeah, that.
Speaker A:And insert into any industry, protection, efficiency, gain.
Speaker A:Like, whatever it is, it's going to do better at it than anything else we have available.
Artie:Yeah, exactly.
Artie:Generative AI can bring a deeper, more nuanced understanding to various fields, leading to improved security, efficiency and innovation.
Artie:The ability to analyze complex data and predict outcomes makes it invaluable in evolving landscapes.
Artie:It's like having an advanced toolkit for tackling complex challenges.
Speaker B:This has never occurred to me before we're having this conversation, but I think it's relevant.
Speaker B:There's a couple of YouTube channels I very much enjoy, one called Mentor Pilot and one called Pilot Debrief.
Speaker B:And they're essentially accident investigations for aviation accidents.
Speaker B:And I'm not sure why I'm so fascinated with it, other than the one guy, Hoover, who's the host of.
Speaker B:He's a former Air Force pilot.
Speaker B:He's the host of Pilot Debrief.
Speaker B:But he talks about the Swiss cheese model of complexity that it takes to cause an aircraft accident.
Speaker B:Right.
Speaker B:Like, you can look at the cheese from any side and you can't see through it.
Speaker B:But if you find a way to hold that cheese just right, all these little holes can line up.
Speaker B:And that's how these crashes happen.
Speaker B:And so when I think about a commercial aircraft, like, you know, an Airbus or a Boeing or whatever, new modern glass cockpit jet, the control systems and the avionics systems and the navigation systems, they're incredibly complicated and incredibly sophisticated.
Speaker B:But the software that's running that stuff, the firmware, is prone to problems because it is like.
Speaker B:So the parallel here, before I fly this analogy straight into the ground, is that before AI, we humans augment and improve very complicated things like a commercial airliner by trying to be very, very organized at how we bring together this incredibly massive amount of really technical, really detailed, really dry data and information and firmware and software, and we try to assemble all that into this thing that works and functions as a contiguous whole.
Speaker B:And it is frequently the case in aircraft accidents that, like, I'm thinking specifically of the Boeing 737 Max 8, and it was a piece of software called MCAS that was what crashed those two airplanes, I think, one in Korea and one in Ethiopia and killed 360 some people.
Speaker B:It was a software bug, but it was a software feature that got added as a result of trying to work around another complicated issue.
Speaker B:And so I think about if pilots operating something complex like that had the ability to interact on a normal human language level with something like an AI that actually had its arms all the way around that complexity, I mean, it would save lives like it is.
Speaker B:That's the big power of AI is being able to be all the people you need at once, all the way from the most technical, smallest component, machining tolerance, and all the way up to, you know, what's the right checklist procedure for this condition and be able to cover all those bases in between and speak to you about it like a.
Speaker A:Human and then ask you if you want to know about the elections.
Speaker A:Absolutely.
Artie:AI's potential to integrate and.
Speaker B:Go ahead.
Artie:Go ahead, Artie, interpret.
Artie:AI's ability to seamlessly integrate and communicate complex information could bridge critical gaps, enhancing both safety and efficiency.
Artie:It's about providing that comprehensive support, helping to prevent errors.
Artie:And ultimately, you're kind of on this.
Speaker B:Okay.
Speaker B:I mean, I'm just.
Speaker B:You seem like you're struggling a little bit.
Speaker B:I'm not sure what's going on, but that was a long analogy on my part.
Speaker B:And meanwhile, while Artie was hiccuping, we did hit the ground.
Speaker B:There were no survivors.
Speaker B:But when I think about the complexity of something like a national power grid or whatever, I don't even know how our national communications backbones work, banking systems, all of these big, hairy, complicated things.
Speaker B:They're all things that humans and way less capable algorithms have built and made function and made workable already.
Speaker B:And so there's a lot of improvements, I think, to come.
Speaker A:A lot of improvements to everything.
Speaker A:Yes, from the most complex down to some of the most simple.
Speaker B:And, and I think.
Speaker B:So to.
Speaker B:To pull this back to the topic of the episode, I think that's.
Speaker B:That is a valid thing that OpenAI is advocating for in this paper that they came out with is the amount of change and the types of institutional change that we're going to see, even taking the economics part out of it and just saying in how technology works, period, across the board is so big that we need national federal standards and, and really even, I mean their, their paper goes on to argue even beyond that we need to work at the national level to develop some global standards and guidelines and.
Speaker B:Right.
Speaker A:And share our tools with other countries to help them come up to speed.
Speaker A:Our allies, the only the ones we like.
Speaker B:Yeah, right.
Speaker A:But to help them come up to speed so that we can all work together towards a common goal.
Speaker A:And it's not wrong.
Speaker A:Like that's, that's how we figure out how best to use it.
Speaker A:I think it's interesting crap.
Speaker A:What was it similar to like that paper that I sent or that, that email that I sent you about?
Speaker A:They're looking for manufacturing companies to work with.
Speaker A:They're, they're all going to clamor to find a fit.
Speaker A:Right?
Speaker A:Like where does this product actually fit?
Speaker A:And the more hands you get it into, the more you figure out where it fits.
Speaker A:And I think OpenAI while sounding and seeming and thumping the.
Speaker A:We're a humanitarian organization.
Speaker A:We love our country.
Speaker A:Everything they're trying to do, they all want to make.
Speaker A:Like I said at last episode, they just want to make money.
Speaker A:Right.
Speaker A:I mean that's not just what they want to do.
Speaker A:Sam Altman seems to just want to get to AGI like he just wants to build stuff.
Speaker A:But you don't continue building stuff unless you're profitable.
Speaker A:That's right.
Speaker A:And ain't no profit like the national defense profit.
Speaker B:That's where the money's at, boys.
Speaker A:But I mean it does make sense for government to get behind it to me like.
Speaker A:And while they can help establish guidelines, not meddle in it.
Speaker A:Because one of.
Speaker A:I really like the car analogy in that paper, right.
Speaker A:They talk about how the car in Britain never didn't take off right away because they legislated the crap out of it.
Speaker A:They said, well, every car has to have a flagger in front of it because if it comes across a horse and carriage, the, the flagger is going to flag the car out of the way to let the horse and carriage pass.
Speaker A:Well, how does the flagger not get run over?
Speaker A:There's a four mile an hour speed limit.
Speaker B:Yes, there's a, there's a four mile an hour limit.
Speaker B:Cars cannot go faster than four miles an hour.
Speaker A:And then the US was like, hold my beer.
Speaker A:Like, let's build roads.
Speaker A:Let's build all this stuff.
Speaker A:Let's build freeways and let's let these companies that want to build automobiles free.
Speaker A:Like, let's let them.
Speaker A:Let's let free market dictate innovation.
Speaker A:And it's the same thing here.
Speaker A:Like, that is one of the beauties of this country is that competition does breed better product.
Speaker A:Because you have so many people that are so close and iterating so quickly on this that they all want to be first and that drive to be first bygone.
Speaker A:It will make the product better.
Speaker A:Like that's just how it works.
Speaker A:Or how it kind of works.
Speaker B:Yeah, I hope it works.
Speaker B:No, I think it's a good point though because like, the, the point of, or the reason that the US is good at, at fostering innovation is that we're a capitalist country and we really can vote with our wallets.
Speaker B:And that's like, I'm always a big believer in the best way to accomplish things is to align the incentives of the people involved in making it happen.
Speaker B:And for all the evil things it does and all the terrible situations it causes, Capitalism is a pretty pure alignment of incentives.
Speaker B:Right.
Speaker B:The consumer wants to get the most interesting.
Speaker B:You would want to say most useful.
Speaker B:That's maybe probably most entertaining, is more accurate.
Speaker B:But the consumer says, this is what I want to get from this technology.
Speaker B:And the companies that are most able to deliver that at a competitive price will make the money.
Speaker A:Yeah.
Speaker B:And it's good.
Speaker A:And our government's really good at printing money.
Speaker A:So they just take some of that printed money so we can't run out.
Speaker B:That's easy.
Speaker B:Yes.
Speaker B:Yeah, solved.
Speaker A:But I mean, there's only so much capital, private capital, to fund these companies.
Speaker A:Right.
Speaker A:Eventually it's going to dry.
Speaker A:I mean, it may not dry up, but it's going to taper.
Speaker A:Like that's inevitable.
Speaker A:So if government figures out a way to infuse money into these companies, even if it is in the name of national defense or whatever to keep us in front, it's probably worth doing.
Speaker A:Like at the end of the day, it's going to benefit.
Speaker A:If done correctly, it's going to benefit everybody.
Speaker B:Like national Bitcoin Reserve.
Speaker A:Protected by ChatGPT?
Speaker B:Yes.
Speaker B:No, it will benefit.
Speaker B:And I also, I was just thinking about the car analogy when you mentioned it, because it, it works on a bunch of levels.
Speaker B:Right.
Speaker B:Like they have smart people at chat or at OpenAI, as it turns out.
Speaker B:Or maybe Artie came up with the car analogy.
Speaker B:I don't know.
Speaker B:Artie's pretty good at analogies.
Speaker B:Artie was the car Analogy your idea or did somebody else at OpenAI come up with that?
Speaker B:As far as this economic paper.
Artie:The car analogy is a common way to illustrate the impact of regulation on innovation.
Artie:And it's used by many to explain how balancing innovation and regulation can shape industries.
Artie:It's a useful analogy for showing the importance of finding the right balance.
Artie:So it's not attributed to any single person or organization.
Speaker B:All right, then I claim it it was my idea.
Speaker B:But it's a good analogy for a bunch of reasons.
Speaker B:Because one of the challenges of the US is we have layers upon layers upon layers of legislation.
Speaker B:So we have federal legislation and then the states want to alternatively like suckle at the teat of mother federal government, but also demand their independence when, when it comes to doing whatever thing they want to do.
Speaker B:And so as, as a counterpoint, I would say like marijuana being classed as a, whatever class three drug is a good example of like a pretty dysfunctional setup where it's federally illegal to use it because it's classified that way.
Speaker B:But more than half the states have made it legal locally.
Speaker B:But if you cross between states, you can get arrested.
Speaker B:And like, there's no real data that makes this a confusing topic.
Speaker B:Right.
Speaker B:We're, we're, we're dealing purely with like stagnant inertia from the prohibition era or something.
Speaker B:Right.
Speaker B:And so as much as I'm excited about large companies like Meta, Google, Microsoft and OpenAI taking an active role in how our federal legislation works, I'm also skeptical enough of how our federal government functions to believe that the things most likely to happen quickly are the things that benefit those large companies, and the things that happen more slowly or never are the things that would directly benefit consumers and maybe even protect consumers.
Speaker A:We saw in that paper they talked specifically about state level adoption of AI, which makes a lot of sense to me too.
Speaker A:Like.
Speaker B:Oh, already?
Speaker B:Exactly.
Artie:I agree.
Artie:State level adoption can bring more tailored and effective policies that address local needs.
Artie:It allows for experimentation and flexibility, which can be crucial for developing best practices and fostering innovation.
Artie:It also helps to bridge the gap between federal policies and on the ground implementation.
Speaker B:There's almost no point.
Speaker B:Yeah, so we hopefully.
Speaker B:Well, again, I don't really believe this is going to happen.
Speaker B:I will say I'd love if it happened, but I don't believe it'll happen.
Speaker B:Hopefully we end up with a set of AI legislation that is not flawed and disconnected federally, and then ignored, overridden and directly contradicted at the state level and maybe again at the county and city level, because why not?
Speaker B:Yeah, but that's probably where we'll end up for a while.
Speaker A:I don't know.
Speaker A:I mean, I.
Speaker A:I think that.
Speaker A:I think that everybody's probably going to agree on the function of AI at some point in the near future, because it's going to be impossible to ignore.
Speaker A:And when that happens, everybody will just be on the bandwagon because what option do they really have?
Speaker A:Like, anybody?
Speaker A:Like, I don't.
Speaker A:I bet there's not a single person that if you go, hey, I have this thing that's going to make your life easier.
Speaker A:And here's how.
Speaker A:And here's how you have to interact at a lesser level.
Speaker A:And you have to interact to get this solution.
Speaker A:Now, there's probably not a single person who's going to go, no, right?
Speaker A:As long as you make it easy to do it.
Speaker B:Depends on who else is telling them that that's a lie and who they believe more than that.
Speaker B:I mean, no logical person would say no.
Speaker A:I feel like it's going to be impossible to ignore.
Speaker A:Like, that's.
Speaker A:I.
Speaker A:I really do.
Speaker A:And there's not.
Speaker A:I mean, I get it.
Speaker A:I know I'm also very skeptical of people, but I feel like that this is just a technology that, like, once that wave crashes, everybody's gonna go, oh, God, yeah, right?
Speaker A:Like, yeah, like, I don't have.
Speaker A:I don't have the ability to go.
Speaker A:I don't believe in that AI because it can't.
Speaker A:Yeah, it can.
Speaker A:It can do whatever you ask it to do.
Speaker A:Like, there is no.
Speaker A:I don't know, man.
Speaker A:Unless you're living in a hole in the bunker with a bunch of machine, it shouldn't.
Speaker A:Sorry to any preppers out there.
Speaker A:I'm not trying to alienate you.
Speaker A:I'm just kidding.
Speaker A:But no, I really feel like it's just.
Speaker A:I don't know, man.
Speaker A:We've said it a million times.
Speaker A:We've been looking at this.
Speaker A:I mean, we both played with it and touched it and things like that, but in three and a half months, it's gone from, wow, look at this thing.
Speaker A:We can talk to.
Speaker B:To.
Speaker A:Sam Altman's going to sit down with the US Government to explain the state of AI Right?
Speaker A:Holy.
Speaker B:I encourage everybody to watch that if it's possible to, by the way.
Speaker B:Because if you want to see just how out of touch the leaders of our country are with anything related to even the simplest technology, go back and watch the Zuckerberg.
Speaker B:Oh, my gosh, man.
Speaker B:I.
Speaker B:I like.
Speaker B:It is.
Speaker A:Jesus.
Speaker B:That's a bad coaster design.
Speaker B:This one does the same thing.
Speaker B:It's that look, it's actually, it's actually.
Speaker B:This one's stone and it's broken in half because it's done that anyway.
Speaker B:Yeah.
Speaker B:Technology is a challenge.
Speaker B:Like it's hard to understand.
Speaker B:It's hard to.
Speaker B:Our leadership is not well versed in it.
Speaker B:And all the people.
Speaker B:Elon Musk is going to be there.
Speaker B:You know, he's going to be there.
Speaker B:Like he'd be jumping around in the background.
Speaker A:You think Sam Altman's trying to usurp.
Speaker B:Yes.
Speaker B:Right.
Speaker B:So the people that are going to be guiding the government to make these wise decisions are probably all billionaires.
Speaker B:This is a problem.
Speaker B:Like this is.
Speaker A:President's going to be a billionaire.
Speaker B:Yeah.
Speaker A:I don't know.
Speaker A:Is he.
Speaker A:Did he spend it all?
Speaker A:I don't know.
Speaker B:He will be after he's president, who knows?
Speaker B:But I think the.
Speaker B:So here's the bright shining hope is we're running out of time here with already.
Speaker A:That is the bright shining hope.
Speaker B:Yeah, that's right.
Speaker B:We're just running out of time.
Speaker B:This is the solution is a 60 minute per day limit for everybody with AI.
Speaker B:So we're going to need to learn to talk faster.
Speaker B:The economic incentives to get it right are not just massive, they are going to be continuous and so powerful that unlike marijuana legislation or even health care options or vehicle emissions, all these things that are left at the state level and end up just this dysfunctional mess.
Speaker B:The incentive to get AI working smoothly at a national level and at a state level is going to be massive.
Speaker B:So to me, that's what's going to keep pushing it forward.
Speaker A:Yeah.
Speaker A:And I mean, I can see if it's running efficiently at both levels, that means that both levels are going to finally potentially be in lockstep.
Speaker A:Right.
Speaker A:Like, and all of those other problems that you're talking about, those are solvable problems with that system.
Speaker A:So.
Speaker B:And it.
Speaker B:Because humans love nothing more than splitting one tribe into two tribes that hate each other, I don't think it's impossible that we see a fairly significant percentage of the population, at least in the United States, stand up and say, I don't accept AI.
Speaker B:AI is taking jobs.
Speaker B:There's people that are still like legitimately angry about self checkout.
Speaker B:I'm not going to check out that myself.
Speaker B:I pay them for this stuff.
Speaker A:Like, hey, why you got to bring that kind of accent?
Speaker B:Mike's always bitching about that.
Speaker B:No, he's not.
Speaker A:I'm from the South.
Speaker B:Yeah, but it like people are mad about that technology and they're not letting go and they're not like giving in.
Speaker B:This is so much worse because this is going to be.
Speaker B:My whole family worked at the XYZ and they all got laid out.
Speaker B:They all lost their jobs.
Speaker A:I don't know.
Speaker A:I mean, I think at that point, well, there are going to be other things that have to be looked at, because what if everybody has to work 15 to 20 hours a week and they get the other time free?
Speaker B:I'm in.
Speaker A:Exactly.
Speaker A:Who's going to say no to that?
Speaker B:Some people will, but some people will.
Speaker A:Almost said something I really shouldn't say.
Speaker B:Nicely caught, sir.
Speaker A:Yeah, yeah, nevermind.
Speaker B:Anyway, okay, Artie, we're about out of time with you.
Speaker B:You've given up on counting down how much time we have left.
Speaker B:You give us a 15 minute warning and we're never sure how much time we got left.
Speaker B:Artie.
Speaker A:I think that's our answer.
Speaker B:All right then.
Speaker B:Thanks for your input, Artie.
Speaker B:If you guys, anybody who's still listening, if you have not checked out that paper, it's an Interesting read.
Speaker B:15 pages, so it's not short, but it's not really heavy either.
Speaker B:It's pretty browsable.
Speaker B:Or grab that PDF, load it up into NotebookLM, just search that and it will give you a 10 minute, much higher quality podcast that covers what's in the paper.
Speaker B:But it is interesting.
Speaker B:And, and yeah, we're there.
Speaker A:Like, yeah, just do a, do a search for chat or for OpenAI economic blueprint.
Speaker A:Get to the page, scroll to the bottom, download the doc, upload the doc.
Speaker A:Everybody's got a Google account.
Speaker A:Log in Notebook LM with that and play with that tool, because that's another super valuable tool.
Speaker B:Yep.
Speaker B:Awesome.
Speaker B:All right, until next time, remain naturally unintelligent.
Speaker B:Cheers.
Speaker B:In my empty water glass.
Speaker B:It's too early for whiskey.
Speaker B:All right, we'll see you all next time.
Speaker B:Bye.
