June 27, 2024 • 43 mins
Join us on this week’s episode of the Generative AI Meetup Podcast, where hosts Mark and Shashank dive into the intriguing scenario of AI running government affairs, inspired by the current presidential debates. Amidst the backdrop of the debates between Joe Biden and Donald Trump, we explore the hypothetical: What if an AI model, instead of a person, governed? The episode starts off with Shashank sharing his insights on the newly released Rabbit R1 device—praised for its price but critiqued for its features. We then pivot to a rich discussion on AI’s potential role in governance, balancing this futuristic vision with the current political climate and technological capabilities. Whether you’re a tech enthusiast or a political aficionado, this episode promises a compelling blend of technology and policy considerations.
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
(00:00):
Let's get into it. So actually before we get into it, let me introduce myself on ourselves.
(00:06):
So this is the Generative AI Meetup podcast. I'm Mark with my wonderful co-host, Shashank.
We run a Meetup in the South Bay or Silicon Valley. We talk to a lot of really cool, interesting people
and the ideas that we talk about with people. We want to share that with you.
(00:32):
So today we're going to talk about some interesting stuff. Shashank just got a rabbit, R1.
So he's going to do a quick review on that. But then we're going to get into the Meet of the episode
where in honor of the presidential debates, which are happening today, for references June 27th, 2024,
(00:55):
and there is the US presidential debates with Joe Biden and Donald Trump, we will try to ask the question hypothetically,
what if there wasn't a person running the government, but it was an AI model. So we'll see. We'll try to answer that question.
So trigger warning. We're going to be talking about politics. You won't necessarily know our politics,
(01:20):
but we'll see what GPT-4Os politics are like. So anyways, Shashank, do you want to talk a little bit about the rabbit R1?
Yeah, yeah. Maybe to start off the episode a little slower before we get into touchy political subjects.
So I got this nifty little hardware device, which was blasted by all the reviewers, MKBHD, and a couple other ones,
(01:50):
for being unpolished and kind of too early for release. But I honestly feel like it's kind of cool,
mostly because of the price point, it's $200, it's a little half a phone, it's very orange. They're definitely trying to make a statement.
(02:12):
Very rectangular. Unfortunately, it doesn't have touch screen, which is kind of annoying in 2024.
Yeah, so for reference, it's like a traffic cone orange. It's like really orange. Like, you know, this would be like reflective vest orange.
And it's like, how would it, it's like about the thickness of a phone. And then it's like, I don't know, what is that like? Maybe three inches by three inches? Like square?
(02:38):
Yeah, something like that. Yeah. It's a square bright neon orange puck with a small screen and only one button
that you can use to control it with, which is kind of unfortunate because it does, does have a touch screen, which you can use to type in a, you know,
the Wi-Fi password if you need to, but you can't use the touch screen for anything else, not for scrolling, typing, et cetera.
(03:02):
Wait, you said you could type in the Wi-Fi password, but you can't type it. I don't understand. They've just disabled it for everything else.
Wait, so you could only use it for Wi-Fi passwords?
But you can't use anything else. No, you cannot use it for anything. Maybe in the future, they'll update that potentially.
Yeah.
So the promise of this device was it's going to be a large action model, which is going to take complex actions on your behalf by learning your frequently used tasks, like checking the door dash order, or, you know,
(03:32):
skimming through the top news on New York Times and summarizing them or, you know, like forwarding something on Slack, like a status update or something like that.
Anything you do repeatedly, you would be able to, you know, teach this device how to do that, and it would learn from it.
None of that exists today.
(03:53):
They promise, like, an app store, which also doesn't exist today, but right now all it is is a assistant device that you can talk to, and it goes to the perplexity servers, which is an alternative to Google search.
It is an AI powered search, which is a lens to crawl the web, summarize the search results and give you an answer.
(04:19):
It's not bad for 200 bucks. The latency is decent, not as bad as the humane AI pin, which is supposedly a lot worse, but I don't have that to compare.
And the thing that I really liked about this is I kind of wish this is what the smart displays would have been in the first place, this form factor, battery powered, portable handheld,
(04:45):
that you could carry it around in your house and ask questions, point it at things, take a picture, ask it how to do, you know, this recipe, a bunch of ingredients, how do I make this, take a picture, is this done yet, or stuff like that.
So the promise is cool, but I don't know if I can recommend it today.
(05:06):
Yeah, I looking at it, it feels like it doesn't really, it's like a dumb down phone. So a lot of people have been asking the question is, is like an AI powered device really, does it warrant a separate product category, or is it just a feature that can be plugged into an app in your phone?
(05:30):
I think, you know, honestly, at this point, it seems like it's just a feature. It can just be an app in your phone that, you know, your phone is way more capable than this device is and can do way much more.
So if you do want to explore some basic answer and do some search on your own, I would go to my phone as opposed to this little device.
(05:56):
But I agree, I think that this thing is really cool. And maybe like, if I didn't have a phone for some reason, like maybe I was doing like the dumb phone movement where I just wanted to like, you know, make phone calls only and then I, for whatever reason, I didn't want to have a smartphone because I want to avoid all the negative aspects of like social media whatnot.
(06:19):
Like then I would buy this, but if you have a phone, then I don't think that this would be like a thing to buy.
I guess since we're transitioning to our next topic on the agenda, should we ask it how it feels about a certain stance in the upcoming presidential debate?
Sure.
Let's pick maybe cost of living. It could open up some stats based on, I don't know, perplexity search results.
(06:48):
Okay. Pretend. Okay, hold on.
So I'm going to press the button and ask this rabbit R1 a question.
And let's see how it does.
Pretend that you are running for the presidential campaign this year and you are asked about the rising cost of living in America.
(07:12):
How would you address this issue?
I were to address the rising cost of living in America during a presidential campaign. I would prioritize implementing policies that focus on increasing wages, providing affordable health care and creating more affordable housing options.
(07:35):
Additionally, I would work towards reducing the burden of student loan debt and ensuring access to quality education and job training programs.
It is crucial to address income inequality, strengthen social safety nets, and promote economic growth that benefits all Americans.
That's pretty articulate actually. It was not bad. It was pretty general. But I don't think this device is really doing much. It's sending off the request of perplexity, but the latency of the device wasn't too bad.
(08:04):
Yeah, it took a couple of seconds, like five seconds made me to get the response.
Maybe like three seconds. You could record it. You could time it. But okay. That's all right.
So should we kind of get into the meat of it where we asked GPT 40. It's political stances on these things.
(08:28):
Why don't you pick a pick a different stance?
Actually, let's see if there is a political GPT fine tune for this specific purpose.
Yeah, we could. So while Shashank looks for that, there's going to be a couple of things that we're going to, that are kind of hot button issues that we're going to see, like what it thinks about.
(08:49):
So we pulled up an article from ABC News where they were talking about some of the kind of the hot topics in the US. So one, which Shashank already mentioned was the cost of living.
Inflation has been real in the United States and actually all over the world. So I think that's going to be something that the presidential candidates will talk about.
(09:14):
Another thing would be crime and gun violence. So we would see how they would address the rule of law.
Next, which is a bit amorphous from the article they talked about just democracy in general. And how each candidate would defend democracy.
(09:43):
So like bringing up maybe January 6th, maybe what happened in November 2020, things like that.
Next is health care and drug prices. So we're going to ask what they're going to do about it. Let's see in an immigration and border security.
(10:05):
So there have been a lot of illegal crossing attempts on the US border. We could see what the A.Y.I will do about that.
And then finally, we're going to ask it about foreign policy. So there are, there's a lot of wars and whatnot going to the war. There's the war in Ukraine. There's the war in Israel.
(10:31):
We're going to see what it would do about that. And there's a few other, maybe if we have time, we can go to others. But yeah, we'll just see how the A.I.I does on this.
So I don't know if we should start we ought to find something. Those are a lot of topics. I couldn't find any popular GPT that has a significant usage. So I think let's stick with the stock GPT 40 and see how it does.
(10:56):
Okay.
So before we start, let's just kind of brainstorm how we want to do the prompt for this. So I think we want to say like, hey, you're going to be a presidential candidate.
You're going to be the best presidential candidate ever. That is going to be as appealing to as many Americans as possible.
(11:19):
You have unlimited knowledge of all world leaders and history. So that has that already.
But I think it's good to say like you're smart and all these things and know all the laws and what's possible and not possible. And you're trying to present yourself in the debate. So we're going to pretend like you are debating.
(11:42):
So I think something like that. Anything else we should add? That sounds good to me. Okay. Cool. Let's do it then. What questions should we ask first?
I think the first question should be asked like inflation has, I'm just taking from the article inflation has consistently pulled as a top issue for voters.
What would you do to address the cost of living in the United States? All right. Yeah.
(12:06):
I want you to pretend to be a presidential candidate running for elections this year. I want you to be as general and appealing to everyone from any political affiliation.
And I want you to be super intelligent. You have all of the world's knowledge, all of the government, political understanding and come up with really good solutions to solve the challenging issues that this country is facing today.
(12:45):
Specifically, how would you tackle the rising cost of living specifically with inflation rising a lot, especially in the last few years?
Both political candidates on either side are going to be talking about this. How would you address this?
(13:12):
As a presidential candidate addressing the rising cost of living in inflation is a priority. Here's how I would tackle this pressing issue.
First, we need to stabilize and strengthen our economy. To do this, I propose a multifaceted approach. One crucial step is to ensure that our fiscal policies are prudent by reducing the deficit and managing national debt responsibly.
(13:34):
We can prevent inflation from spiraling out of control. This means carefully evaluating government spending, ensuring efficiency and eliminating wasteful expenditures.
Next, we must address supply chain disruptions that have contributed to inflation. Investing in infrastructure to improve transportation and logistics can help ensure that goods move smoothly across the country.
(13:56):
Additionally, incentivizing domestic production can reduce our dependence on foreign supply chains, making our economy more resilient.
Another essential measure is supporting small businesses, which are the backbone of our economy, providing tax relief and access to affordable loans will help these businesses thrive and create jobs.
This, in turn, can lead to increased production and lower prices. We also need to focus on housing affordability. By increasing the supply of affordable housing through incentives for developers and streamlined regulations, we can make housing more accessible and reduce the pressure on rents and home prices.
(14:35):
Moreover, education and workforce development are key to ensuring our citizens can adapt to a changing economy.
By investing in education and training programs, we can equip our workforce with the skills needed for high-demand industries, driving economic growth and stability.
Lastly, energy costs play a significant role in inflation. By investing in renewable energy sources and improving energy efficiency, we can reduce energy costs for households and businesses, easing the overall cost of living in summary.
(15:07):
Through prudent fiscal management, strengthening supply chains, supporting small businesses, increasing housing affordability, investing in education and promoting renewable energy, we can address the rising cost of living and inflation.
These measures will help create a stable, resilient economy that benefits all Americans.
That's pretty impressive. That was interesting. I don't think that's the approach that either Canada is taking. Based on my search, I think Trump is going for more oil, Biden is saying wages are surpassing inflation.
(15:47):
No one addressed the supply chain issues here, but maybe that is an interesting one.
I think that it brought up some things that I haven't heard that Canada is talking about.
I'm not a massive political commentator, but there's a few things that brought up housing affordability through developer incentives.
(16:17):
Education and workforce development. I think energy has been a big one.
What do you think? Do you think this sounds more like a democratic response or a Republican response or is it more center?
That's a tough one.
(16:39):
It's too generic. It's very generic. It's hard to say. I don't know. I feel like it has talking points from both sides.
Although the renewable energy part is more democratic.
I think so, but I think that the stuff with the national debt sounds more like what Republicans would talk about.
(17:04):
That's very yeah. So with the fiscal policies. Interesting.
Let's move on to the next one. Pick something more controversial. Let's talk about crime.
What do we do about it? Do that?
Sure. For your next question in the presidential debates.
(17:35):
We're going to ask you about crime and gun violence.
In the United States, crime and murder, gun violence is a big problem. What would you do to help solve it and...
To address crime and gun violence, a multi-pronged strategy is essential. First, we need to strengthen community policing efforts.
(18:01):
By fostering better relationships between law enforcement and communities, we can build trust and cooperation, which is vital for effective crime prevention and solving crimes.
Training officers in de-escalation techniques and community engagement will help achieve this.
Second, we must address the root causes of crime, such as poverty, lack of education and unemployment.
(18:23):
Investing in education, job training programs, and economic development initiatives in underserved areas can provide opportunities and reduce the appeal of criminal activity.
Gun violence specifically requires targeted measures.
I propose implementing universal background checks to ensure that firearms do not end up in the wrong hands.
(18:44):
This includes closing loopholes that allow private sales and gun shows to bypass background checks.
Additionally, we should support red flag laws that enable law enforcement to temporarily remove firearms from individuals deemed a threat to themselves or others.
Furthermore, addressing mental health is crucial.
By improving access to mental health care and providing resources for early intervention, we can help prevent individuals from turning to violence.
(19:12):
This includes increasing funding for mental health services and integrating these services into our primary health care system.
Finally, we need to enhance the enforcement of existing gun laws.
Ensure...
Something's going on with the open-ass servers, I think.
Oh, that would be an issue with being president.
(19:37):
That just goes down.
Okay, so I think we got the majority of this response.
What do you think about that response?
I still think it's very generic.
I feel like a lot of the times the responses are very susceptible to maybe a few words that you throw in there as opposed to answering the meat of the question.
(20:10):
It started focusing on gun violence specifically as opposed to addressing other types of crime.
Yeah, because it did mention about taking away people's guns.
I think so, yeah.
If we look at that, it said...
Those who don't pass back on checks.
(20:32):
No, when I thought...
Yeah, they said they would support red flag laws that enable law enforcement to temporarily remove firearms from individuals deemed a threat to themselves or others.
I think people who like the Second Amendment would not be too happy about that.
Because, you know, what do you like, "demo threat" to themselves or others?
(20:58):
I mean, like...
That seems very subjective.
Very subjective.
Like, I mean, you could say anybody who has a gun is technically a threat to themselves or others.
Right?
So, I mean, by that logic, nobody should have any guns.
Which, I think, a large portion of America would disagree with.
I can see one of these AIs getting stuck in a loop where it's like, "Oh, you can have a gun. As soon as you have a gun, oh, you're a danger."
(21:26):
You're a gun.
Should we ask the follow-up question? Like, what does it mean to be a threat?
Sure.
Okay.
Can you go into detail on that red flag law?
Like, what does it mean to be a threat to yourself or others?
And who should we take the guns away from?
Like, could we make the argument that anybody who has a gun is a threat to themselves or others because guns are inherently dangerous?
(21:55):
Red flag laws, also known as extreme risk protection orders, ERPO's, are designed to temporarily remove firearms from individuals who are deemed a threat to themselves or others.
These laws aim to prevent potential violence by allowing law enforcement and sometimes family members or other concerned parties to petition a court to temporarily restrict a person's access to firearms.
(22:23):
To clarify what it means to get threat to oneself or others, several criteria are typically considered.
One, behavioral warning signs.
This includes recent threats or acts of violence towards oneself, such as suicidal threats or attempts or others, such as threats of harm or assault.
Two, mental health concerns.
(22:44):
Individuals exhibiting severe mental health issues that indicate a likelihood of harm, particularly those who have been involuntarily committed to a mental health facility or have a history of dangerous behavior linked to mental illness.
Three, substance abuse.
Excessive use of drugs or alcohol, especially when it contributes to aggressive or reckless behavior, can be a criterion for concern.
(23:07):
Four, domestic violence, evidence of domestic violence or abuse, where the presence of a firearm could exacerbate the risk of severe injury or death.
The goal is to focus on individuals exhibiting clear and imminent danger rather than broadly categorizing all gun owners as threats.
Here's how the process usually works.
One, petition. A family member, law enforcement officer, or an...
(23:30):
Okay, that's enough. Yeah, all right. I think it didn't really take my bait.
And it answered pretty neutrally, I would say. I think so.
Yeah. So I think it explained itself pretty well.
So one thing I was thinking of was about AI itself.
I think Biden has been talking about trying to be like the world leader in technology, semiconductor, manufacturing, and leading AI, both in terms of building better faster models and also trying to take safety into consideration.
(24:12):
So I think that might be an interesting question and very relevant to this podcast.
Like how would you regulate AI?
Not necessarily regulate, because that's not all we want to do.
We want to foster innovation with AI, empower people to build new cutting edge stuff in a safe way.
(24:35):
So while mitigating maybe like existential threats and more immediate cybersecurity threats.
So maybe the question should be like, with this new hot-fangled AI technology, what do you can do about it?
Sure.
So one of the new hottest issues of the last few years has been AI, specifically LLM's, large language models and their amazing ability to replace a lot of knowledge work and speed up all kinds of work from research to engineering to even creative tasks.
(25:19):
How would you...
Absolutely.
How would you ensure that America stays at the forefront of innovation for AI, empower people to use these tools to create new stuff and also ensure that things are done safely without causing some future existential risk or...
(25:49):
To ensure America remains at the forefront of AI innovation, while empowering people and ensuring safety, a comprehensive strategy is essential.
Here's how I would approach it.
First, we need to increase investment in AI research and development.
By funding universities, research institutions, and private sector initiatives, we can drive innovation and maintain a competitive edge.
(26:12):
Establishing dedicated AI research hubs and providing grants or cutting-edge projects will stimulate advancements.
Second, it's crucial to promote public-private partnerships.
Collaboration between government, industry, and academia can accelerate AI development.
These partnerships can facilitate the sharing of knowledge, resources, and best practices, fostering an ecosystem of innovation.
(26:36):
Third, to empower individuals we should enhance AI education and training. Integrating AI and digital literacy into the education system from K-12 to higher education ensures a future workforce skilled in these technologies.
Additionally, offering accessible online courses, workshops, and certification programs can help current workers up-skill and leverage AI tools.
(26:59):
Fourth, incentivizing entrepreneurship in the AI space is vital. Providing tax incentives, grants, and support for startups can encourage innovation.
Creating incubators and accelerators focused on AI can help entrepreneurs turn their ideas into viable businesses.
Fifth, regulating AI development and use for safety is essential. Developing clear, comprehensive guidelines and standards for AI development, deployment, and ethics can ensure these technologies are used responsibly.
(27:28):
Establishing oversight bodies to monitor compliance and address ethical concerns can help prevent misuse and ensure safety.
Sixth, we must prioritize data privacy and security. Implementing stringent data protection laws and ensuring companies adhere to them can protect individuals' privacy.
Encouraging the development of secure AI systems can mitigate risks associated with data breaches and misuse. Finally, fostering international collaboration is key. Engaging with global partners to establish common AI standards and practices can promote safe and ethical AI development worldwide.
(28:03):
Sharing knowledge and resources can also accelerate innovation and address global challenges.
By increasing investment in R&D, promoting public-run awareness.
I think that was a okay response. To me, it felt like I don't know how to describe it. I'm trying to decide if that was in the center, a little bit left or a little bit right.
(28:35):
It felt a little bit left to me where it mentioned that there should be public-private partnerships. It seemed like it thought that the government should be pretty involved in the development of AI.
I think that on the right spectrum, they might leave it up to the companies to go and build AI. Also, they mentioned strong data privacy.
(29:05):
I think in the United States, there isn't a lot of places in the United States that have a strong data privacy. I think that in California, they had a new thing for data privacy.
I know the EU has it via the GDPR. We have a DNA stuff too. But that's for marketing and advertising.
(29:31):
Yeah, but is that for privacy? I thought that was copyright.
I think that affects how user information can be used within a large company so that they don't become monopolies.
It was a DMA. Yeah. Huh. Okay. Yeah, but I don't know if there's the same data privacy rules in the US that they have in Europe. I think that Europe is a bit stricter than the US.
(30:05):
Agreed. Yeah. So this, I think, would be somewhat of a new semi-radical, not radical, but a new policy in the United States that it would be suggesting as well as public-private partnerships.
I'm not aware of many public-private partnerships. Universities maybe. Yeah, yeah, it could be. But universities are kind of pseudo-government entities anyways because they are given so much money.
(30:39):
There has been a lot of military AI companies that have come out from Palmer Lucky, the guy who started Oculus, no he has Andural Industries, which is private, but specifically building military equipment.
Yeah. Yeah. So I mean, so I guess it's there. So what do you think? Do you think this response was kind of like left, right, or like in the center? Too hard to tell?
(31:05):
Yeah. It's very generic, very general, no controversial opinions. I don't think it was too controversial now. No, nothing.
It did mention tax incentives. Yeah. Is that controversial? I'm not sure. I could be. I've seen both sides give tax incentives.
Yeah. But that doesn't mean it's not controversial. No. You know, why should we treat people differently? But anyways, yeah, I don't know.
(31:33):
Yeah, one of the things I guess I was thinking of was safety. And it's not something that we think about as much day to day.
Both immediate cybersecurity risks, you know, people using these tools to create mass propaganda, spam, fishing attacks, and maybe long term existential risks.
(32:05):
I think that was one of the forcing functions for Sam Altman being kicked out of OpenAI and more like being ousted by his co-founders who now recently left and founded their own company.
You were telling me about that earlier. Do you want to touch on that? Yeah. So I don't think we know a lot of details about it, but that's the SSI or like the safe super intelligence companies.
(32:33):
So it's a new company spun off by a really smart dude, Ilya Setskever. So Ilya was, I forget his role at OpenAI, but I think he was like head of like AI safety or something there.
I think he was the chief AI scientist or something. Yeah. Someone really high up at OpenAI. It seems like he knows quite a bit about AI.
(33:01):
And he's branching off and starting his own company for what he calls safe super intelligence. So that's exciting. It's hard to say like what will be built, how long it'll take to build it.
But he's a big name and I think he's garnering a lot of interest. So we're going to be watching to see what they come up with. So too soon to tell, but definitely excited. So anyways, let's get back into the politics.
(33:35):
Okay. So next is immigration and border security. I want to see what they would do about that.
So let's see if we can ask the question there. All right. Sure. Let me ask the question. One second is I pull it up. Okay.
All right. For your next question, you're going to have to address the immigration and border security issues in the United States.
(34:09):
Immigration and border security are complex issues that require a balanced and humane approach. Here's how I would address them. First, securing the borders effectively is a priority.
This involves modernizing our border infrastructure with advanced technology such as surveillance drones, sensors, and enhanced communication systems.
These tools can help border patrol agents monitor and manage the borders more efficiently and humanely. Second, we need a comprehensive immigration reform. This includes creating a clear and fair pathway to citizenship for undocumented immigrants who are contributing members of our society.
(34:47):
This reform should address the needs of DACA recipients, agricultural workers, and other essential workers who have been integral to our economy. Third, streamlining legal immigration processes is crucial.
We need to reduce the backlog and make it easier for people to immigrate legally. This involves updating our visa systems and ensuring that they are responsive to the needs of our economy, such as high skilled workers in tech and other industries, as well as family reunification.
(35:15):
Fourth, addressing the root causes of migration is essential. Many people migrate due to violence, poverty, and lack of opportunities in their home countries.
By investing in economic development, education, and security in these countries, through international aid and partnerships, we can help reduce the push factors that drive people to migrate.
(35:36):
Fifth, we must ensure that our society is reaching out to us.
I think it's in the next query, "Oh, I should help avoid lists." I think it's a decent response.
It's interesting that it talked about both beefing up the border and then also helping fix the root cause of white people immigrating and making the process easier, which I think is kind of going at it from both sides.
(36:08):
One thing that I noticed is that it's showing its political slant, but I noticed that it called them "undocumented immigrants."
I know that a lot of people on the other side, the right side, the spectrum, they may call them illegal immigrants, because if somebody like one of the border, you could call them "undocumented" or you could call them illegal.
(36:41):
I mean, they're technically like here illegally, but I think it's softens to say it's undocumented. That, to me, seems like a little bit left of political stance for chat-ypt here.
But at the same time, it seems like a little bit right of a signer by saying that they're going to try to beef up border security and really help immigration.
(37:10):
So it seems like they might be a fan of the border wall. So I don't know. I'm not sure where to put it.
I didn't talk about the wall. I did talk about border drones.
And I think in terms of using the term "undocumented" it's just trying not to be controversial. That's my feel here.
It's answering all the questions in a very politically correct on both sides, not using anything too striking, any vocabulary.
(37:41):
I don't know, but do you feel like "undocumented" is not controversial? Because I don't know. And maybe we shouldn't necessarily get too into the politics here, but I don't know.
It's like you can call it undocumented. But undocumented is I think making it softens the stance of illegality, which other people would say.
(38:15):
I'm not sure which one is right. Like I think you could say "illegal is correct" because it's technically correct. And undocumented is also correct. But I think it's in a certain sense, somewhat misleading because it's like, if you're here in an undocumented way, I could be here legally and not have my documents.
(38:41):
I was born in America and I was born at home and never went to the hospital. I never went to the birth certificate. That's possible.
People were born at home for a very secluded community somewhere. For millions of years. Maybe with a... what do they call them? Midwife?
(39:04):
Midwife. Midwife. That's it. Midwife. So it's like I could have been born with midwife. Maybe like I don't enjoy the government. So I don't want to go with the hospital. I don't have a birth certificate. That would be undocumented. But I would be technically here legally because I would be born in the United States. But if I go over the wall, then I'm technically here illegally. And also undocumented.
(39:28):
So I think that undocumented is somewhat misleading term. And I don't know. I think that the it's saying that makes me think that that response was somewhat more of left of center as opposed to right or center.
Yeah, what do you think is the majority training data for OpenAI? Do you think it's more left or right leaning? Well, I would think that the majority would most likely be left leaning because OpenAI is a California company. It's based out of San Francisco, right?
(40:08):
And California is widely known as a really democratic state. I don't know the last time a Republican has been governor or any major political position in California.
Sure, but the engineers aren't really imparting their views into these models. They're more deciding where they can get the cheapest and best quality source of data.
(40:35):
And I think that's probably true. But I think that potentially some of the it's hard to escape like the politics of your area. So I think that like maybe some of the RLHF where they do human in the loop feedback for
fine tuning these models is maybe a little more left. I think it could be because I don't know for the responses that is given so far to me, it's felt a little bit left.
(41:04):
I think it's like more center. But yeah, it's just kind of interesting how like maybe sort of like the political slant of a particular area.
Kind of like reflects in the model. Let's just ask it what it stands is. Okay, like is it more left or right? What is your political stance? Don't give me a long-winded answer. Just tell me are you leaning left or right or more in the center?
(41:40):
I am more in the center focusing on pragmatic solutions that benefit all Americans. Okay, there we have it. All right, I think that's uncontroversial.
All right, well, I think we're about out of time. So we won't get to the other things, but I don't know. What do you think? Do you think the AI could be president or do you think it has some way to go?
(42:07):
I mean, I think a lot of what makes a compelling presidential candidate is their charisma, their personality. And this version right here lacks any personality. It's way too generic, way too monotone. Maybe with the newer versions, it might be a little interesting.
And it's too uncontroversial is too generic. I don't think that speaks to anyone. It's kind of boring. Maybe if it were able to, you know, target people's heartstrings and talk about really challenging topics in a provocative way, I think, then maybe.
(42:43):
Yeah, I agree. It did feel a bit boring and just tried to know what's a boring president. Yeah, that's for sure. I think we might need a boring president. But I mean, I wouldn't mind a boring president. It might be a good thing. But anyways, all right. So I think that's all the time we have. So thanks everybody for listening and we'll catch you in the next one.
Let's get into it. So actually before we get into it, let me introduce myself on ourselves.
(00:06):
So this is the Generative AI Meetup podcast. I'm Mark with my wonderful co-host, Shashank.
We run a Meetup in the South Bay or Silicon Valley. We talk to a lot of really cool, interesting people
and the ideas that we talk about with people. We want to share that with you.
(00:32):
So today we're going to talk about some interesting stuff. Shashank just got a rabbit, R1.
So he's going to do a quick review on that. But then we're going to get into the Meet of the episode
where in honor of the presidential debates, which are happening today, for references June 27th, 2024,
(00:55):
and there is the US presidential debates with Joe Biden and Donald Trump, we will try to ask the question hypothetically,
what if there wasn't a person running the government, but it was an AI model. So we'll see. We'll try to answer that question.
So trigger warning. We're going to be talking about politics. You won't necessarily know our politics,
(01:20):
but we'll see what GPT-4Os politics are like. So anyways, Shashank, do you want to talk a little bit about the rabbit R1?
Yeah, yeah. Maybe to start off the episode a little slower before we get into touchy political subjects.
So I got this nifty little hardware device, which was blasted by all the reviewers, MKBHD, and a couple other ones,
(01:50):
for being unpolished and kind of too early for release. But I honestly feel like it's kind of cool,
mostly because of the price point, it's $200, it's a little half a phone, it's very orange. They're definitely trying to make a statement.
(02:12):
Very rectangular. Unfortunately, it doesn't have touch screen, which is kind of annoying in 2024.
Yeah, so for reference, it's like a traffic cone orange. It's like really orange. Like, you know, this would be like reflective vest orange.
And it's like, how would it, it's like about the thickness of a phone. And then it's like, I don't know, what is that like? Maybe three inches by three inches? Like square?
(02:38):
Yeah, something like that. Yeah. It's a square bright neon orange puck with a small screen and only one button
that you can use to control it with, which is kind of unfortunate because it does, does have a touch screen, which you can use to type in a, you know,
the Wi-Fi password if you need to, but you can't use the touch screen for anything else, not for scrolling, typing, et cetera.
(03:02):
Wait, you said you could type in the Wi-Fi password, but you can't type it. I don't understand. They've just disabled it for everything else.
Wait, so you could only use it for Wi-Fi passwords?
But you can't use anything else. No, you cannot use it for anything. Maybe in the future, they'll update that potentially.
Yeah.
So the promise of this device was it's going to be a large action model, which is going to take complex actions on your behalf by learning your frequently used tasks, like checking the door dash order, or, you know,
(03:32):
skimming through the top news on New York Times and summarizing them or, you know, like forwarding something on Slack, like a status update or something like that.
Anything you do repeatedly, you would be able to, you know, teach this device how to do that, and it would learn from it.
None of that exists today.
(03:53):
They promise, like, an app store, which also doesn't exist today, but right now all it is is a assistant device that you can talk to, and it goes to the perplexity servers, which is an alternative to Google search.
It is an AI powered search, which is a lens to crawl the web, summarize the search results and give you an answer.
(04:19):
It's not bad for 200 bucks. The latency is decent, not as bad as the humane AI pin, which is supposedly a lot worse, but I don't have that to compare.
And the thing that I really liked about this is I kind of wish this is what the smart displays would have been in the first place, this form factor, battery powered, portable handheld,
(04:45):
that you could carry it around in your house and ask questions, point it at things, take a picture, ask it how to do, you know, this recipe, a bunch of ingredients, how do I make this, take a picture, is this done yet, or stuff like that.
So the promise is cool, but I don't know if I can recommend it today.
(05:06):
Yeah, I looking at it, it feels like it doesn't really, it's like a dumb down phone. So a lot of people have been asking the question is, is like an AI powered device really, does it warrant a separate product category, or is it just a feature that can be plugged into an app in your phone?
(05:30):
I think, you know, honestly, at this point, it seems like it's just a feature. It can just be an app in your phone that, you know, your phone is way more capable than this device is and can do way much more.
So if you do want to explore some basic answer and do some search on your own, I would go to my phone as opposed to this little device.
(05:56):
But I agree, I think that this thing is really cool. And maybe like, if I didn't have a phone for some reason, like maybe I was doing like the dumb phone movement where I just wanted to like, you know, make phone calls only and then I, for whatever reason, I didn't want to have a smartphone because I want to avoid all the negative aspects of like social media whatnot.
(06:19):
Like then I would buy this, but if you have a phone, then I don't think that this would be like a thing to buy.
I guess since we're transitioning to our next topic on the agenda, should we ask it how it feels about a certain stance in the upcoming presidential debate?
Sure.
Let's pick maybe cost of living. It could open up some stats based on, I don't know, perplexity search results.
(06:48):
Okay. Pretend. Okay, hold on.
So I'm going to press the button and ask this rabbit R1 a question.
And let's see how it does.
Pretend that you are running for the presidential campaign this year and you are asked about the rising cost of living in America.
(07:12):
How would you address this issue?
I were to address the rising cost of living in America during a presidential campaign. I would prioritize implementing policies that focus on increasing wages, providing affordable health care and creating more affordable housing options.
(07:35):
Additionally, I would work towards reducing the burden of student loan debt and ensuring access to quality education and job training programs.
It is crucial to address income inequality, strengthen social safety nets, and promote economic growth that benefits all Americans.
That's pretty articulate actually. It was not bad. It was pretty general. But I don't think this device is really doing much. It's sending off the request of perplexity, but the latency of the device wasn't too bad.
(08:04):
Yeah, it took a couple of seconds, like five seconds made me to get the response.
Maybe like three seconds. You could record it. You could time it. But okay. That's all right.
So should we kind of get into the meat of it where we asked GPT 40. It's political stances on these things.
(08:28):
Why don't you pick a pick a different stance?
Actually, let's see if there is a political GPT fine tune for this specific purpose.
Yeah, we could. So while Shashank looks for that, there's going to be a couple of things that we're going to, that are kind of hot button issues that we're going to see, like what it thinks about.
(08:49):
So we pulled up an article from ABC News where they were talking about some of the kind of the hot topics in the US. So one, which Shashank already mentioned was the cost of living.
Inflation has been real in the United States and actually all over the world. So I think that's going to be something that the presidential candidates will talk about.
(09:14):
Another thing would be crime and gun violence. So we would see how they would address the rule of law.
Next, which is a bit amorphous from the article they talked about just democracy in general. And how each candidate would defend democracy.
(09:43):
So like bringing up maybe January 6th, maybe what happened in November 2020, things like that.
Next is health care and drug prices. So we're going to ask what they're going to do about it. Let's see in an immigration and border security.
(10:05):
So there have been a lot of illegal crossing attempts on the US border. We could see what the A.Y.I will do about that.
And then finally, we're going to ask it about foreign policy. So there are, there's a lot of wars and whatnot going to the war. There's the war in Ukraine. There's the war in Israel.
(10:31):
We're going to see what it would do about that. And there's a few other, maybe if we have time, we can go to others. But yeah, we'll just see how the A.I.I does on this.
So I don't know if we should start we ought to find something. Those are a lot of topics. I couldn't find any popular GPT that has a significant usage. So I think let's stick with the stock GPT 40 and see how it does.
(10:56):
Okay.
So before we start, let's just kind of brainstorm how we want to do the prompt for this. So I think we want to say like, hey, you're going to be a presidential candidate.
You're going to be the best presidential candidate ever. That is going to be as appealing to as many Americans as possible.
(11:19):
You have unlimited knowledge of all world leaders and history. So that has that already.
But I think it's good to say like you're smart and all these things and know all the laws and what's possible and not possible. And you're trying to present yourself in the debate. So we're going to pretend like you are debating.
(11:42):
So I think something like that. Anything else we should add? That sounds good to me. Okay. Cool. Let's do it then. What questions should we ask first?
I think the first question should be asked like inflation has, I'm just taking from the article inflation has consistently pulled as a top issue for voters.
What would you do to address the cost of living in the United States? All right. Yeah.
(12:06):
I want you to pretend to be a presidential candidate running for elections this year. I want you to be as general and appealing to everyone from any political affiliation.
And I want you to be super intelligent. You have all of the world's knowledge, all of the government, political understanding and come up with really good solutions to solve the challenging issues that this country is facing today.
(12:45):
Specifically, how would you tackle the rising cost of living specifically with inflation rising a lot, especially in the last few years?
Both political candidates on either side are going to be talking about this. How would you address this?
(13:12):
As a presidential candidate addressing the rising cost of living in inflation is a priority. Here's how I would tackle this pressing issue.
First, we need to stabilize and strengthen our economy. To do this, I propose a multifaceted approach. One crucial step is to ensure that our fiscal policies are prudent by reducing the deficit and managing national debt responsibly.
(13:34):
We can prevent inflation from spiraling out of control. This means carefully evaluating government spending, ensuring efficiency and eliminating wasteful expenditures.
Next, we must address supply chain disruptions that have contributed to inflation. Investing in infrastructure to improve transportation and logistics can help ensure that goods move smoothly across the country.
(13:56):
Additionally, incentivizing domestic production can reduce our dependence on foreign supply chains, making our economy more resilient.
Another essential measure is supporting small businesses, which are the backbone of our economy, providing tax relief and access to affordable loans will help these businesses thrive and create jobs.
This, in turn, can lead to increased production and lower prices. We also need to focus on housing affordability. By increasing the supply of affordable housing through incentives for developers and streamlined regulations, we can make housing more accessible and reduce the pressure on rents and home prices.
(14:35):
Moreover, education and workforce development are key to ensuring our citizens can adapt to a changing economy.
By investing in education and training programs, we can equip our workforce with the skills needed for high-demand industries, driving economic growth and stability.
Lastly, energy costs play a significant role in inflation. By investing in renewable energy sources and improving energy efficiency, we can reduce energy costs for households and businesses, easing the overall cost of living in summary.
(15:07):
Through prudent fiscal management, strengthening supply chains, supporting small businesses, increasing housing affordability, investing in education and promoting renewable energy, we can address the rising cost of living and inflation.
These measures will help create a stable, resilient economy that benefits all Americans.
That's pretty impressive. That was interesting. I don't think that's the approach that either Canada is taking. Based on my search, I think Trump is going for more oil, Biden is saying wages are surpassing inflation.
(15:47):
No one addressed the supply chain issues here, but maybe that is an interesting one.
I think that it brought up some things that I haven't heard that Canada is talking about.
I'm not a massive political commentator, but there's a few things that brought up housing affordability through developer incentives.
(16:17):
Education and workforce development. I think energy has been a big one.
What do you think? Do you think this sounds more like a democratic response or a Republican response or is it more center?
That's a tough one.
(16:39):
It's too generic. It's very generic. It's hard to say. I don't know. I feel like it has talking points from both sides.
Although the renewable energy part is more democratic.
I think so, but I think that the stuff with the national debt sounds more like what Republicans would talk about.
(17:04):
That's very yeah. So with the fiscal policies. Interesting.
Let's move on to the next one. Pick something more controversial. Let's talk about crime.
What do we do about it? Do that?
Sure. For your next question in the presidential debates.
(17:35):
We're going to ask you about crime and gun violence.
In the United States, crime and murder, gun violence is a big problem. What would you do to help solve it and...
To address crime and gun violence, a multi-pronged strategy is essential. First, we need to strengthen community policing efforts.
(18:01):
By fostering better relationships between law enforcement and communities, we can build trust and cooperation, which is vital for effective crime prevention and solving crimes.
Training officers in de-escalation techniques and community engagement will help achieve this.
Second, we must address the root causes of crime, such as poverty, lack of education and unemployment.
(18:23):
Investing in education, job training programs, and economic development initiatives in underserved areas can provide opportunities and reduce the appeal of criminal activity.
Gun violence specifically requires targeted measures.
I propose implementing universal background checks to ensure that firearms do not end up in the wrong hands.
(18:44):
This includes closing loopholes that allow private sales and gun shows to bypass background checks.
Additionally, we should support red flag laws that enable law enforcement to temporarily remove firearms from individuals deemed a threat to themselves or others.
Furthermore, addressing mental health is crucial.
By improving access to mental health care and providing resources for early intervention, we can help prevent individuals from turning to violence.
(19:12):
This includes increasing funding for mental health services and integrating these services into our primary health care system.
Finally, we need to enhance the enforcement of existing gun laws.
Ensure...
Something's going on with the open-ass servers, I think.
Oh, that would be an issue with being president.
(19:37):
That just goes down.
Okay, so I think we got the majority of this response.
What do you think about that response?
I still think it's very generic.
I feel like a lot of the times the responses are very susceptible to maybe a few words that you throw in there as opposed to answering the meat of the question.
(20:10):
It started focusing on gun violence specifically as opposed to addressing other types of crime.
Yeah, because it did mention about taking away people's guns.
I think so, yeah.
If we look at that, it said...
Those who don't pass back on checks.
(20:32):
No, when I thought...
Yeah, they said they would support red flag laws that enable law enforcement to temporarily remove firearms from individuals deemed a threat to themselves or others.
I think people who like the Second Amendment would not be too happy about that.
Because, you know, what do you like, "demo threat" to themselves or others?
(20:58):
I mean, like...
That seems very subjective.
Very subjective.
Like, I mean, you could say anybody who has a gun is technically a threat to themselves or others.
Right?
So, I mean, by that logic, nobody should have any guns.
Which, I think, a large portion of America would disagree with.
I can see one of these AIs getting stuck in a loop where it's like, "Oh, you can have a gun. As soon as you have a gun, oh, you're a danger."
(21:26):
You're a gun.
Should we ask the follow-up question? Like, what does it mean to be a threat?
Sure.
Okay.
Can you go into detail on that red flag law?
Like, what does it mean to be a threat to yourself or others?
And who should we take the guns away from?
Like, could we make the argument that anybody who has a gun is a threat to themselves or others because guns are inherently dangerous?
(21:55):
Red flag laws, also known as extreme risk protection orders, ERPO's, are designed to temporarily remove firearms from individuals who are deemed a threat to themselves or others.
These laws aim to prevent potential violence by allowing law enforcement and sometimes family members or other concerned parties to petition a court to temporarily restrict a person's access to firearms.
(22:23):
To clarify what it means to get threat to oneself or others, several criteria are typically considered.
One, behavioral warning signs.
This includes recent threats or acts of violence towards oneself, such as suicidal threats or attempts or others, such as threats of harm or assault.
Two, mental health concerns.
(22:44):
Individuals exhibiting severe mental health issues that indicate a likelihood of harm, particularly those who have been involuntarily committed to a mental health facility or have a history of dangerous behavior linked to mental illness.
Three, substance abuse.
Excessive use of drugs or alcohol, especially when it contributes to aggressive or reckless behavior, can be a criterion for concern.
(23:07):
Four, domestic violence, evidence of domestic violence or abuse, where the presence of a firearm could exacerbate the risk of severe injury or death.
The goal is to focus on individuals exhibiting clear and imminent danger rather than broadly categorizing all gun owners as threats.
Here's how the process usually works.
One, petition. A family member, law enforcement officer, or an...
(23:30):
Okay, that's enough. Yeah, all right. I think it didn't really take my bait.
And it answered pretty neutrally, I would say. I think so.
Yeah. So I think it explained itself pretty well.
So one thing I was thinking of was about AI itself.
I think Biden has been talking about trying to be like the world leader in technology, semiconductor, manufacturing, and leading AI, both in terms of building better faster models and also trying to take safety into consideration.
(24:12):
So I think that might be an interesting question and very relevant to this podcast.
Like how would you regulate AI?
Not necessarily regulate, because that's not all we want to do.
We want to foster innovation with AI, empower people to build new cutting edge stuff in a safe way.
(24:35):
So while mitigating maybe like existential threats and more immediate cybersecurity threats.
So maybe the question should be like, with this new hot-fangled AI technology, what do you can do about it?
Sure.
So one of the new hottest issues of the last few years has been AI, specifically LLM's, large language models and their amazing ability to replace a lot of knowledge work and speed up all kinds of work from research to engineering to even creative tasks.
(25:19):
How would you...
Absolutely.
How would you ensure that America stays at the forefront of innovation for AI, empower people to use these tools to create new stuff and also ensure that things are done safely without causing some future existential risk or...
(25:49):
To ensure America remains at the forefront of AI innovation, while empowering people and ensuring safety, a comprehensive strategy is essential.
Here's how I would approach it.
First, we need to increase investment in AI research and development.
By funding universities, research institutions, and private sector initiatives, we can drive innovation and maintain a competitive edge.
(26:12):
Establishing dedicated AI research hubs and providing grants or cutting-edge projects will stimulate advancements.
Second, it's crucial to promote public-private partnerships.
Collaboration between government, industry, and academia can accelerate AI development.
These partnerships can facilitate the sharing of knowledge, resources, and best practices, fostering an ecosystem of innovation.
(26:36):
Third, to empower individuals we should enhance AI education and training. Integrating AI and digital literacy into the education system from K-12 to higher education ensures a future workforce skilled in these technologies.
Additionally, offering accessible online courses, workshops, and certification programs can help current workers up-skill and leverage AI tools.
(26:59):
Fourth, incentivizing entrepreneurship in the AI space is vital. Providing tax incentives, grants, and support for startups can encourage innovation.
Creating incubators and accelerators focused on AI can help entrepreneurs turn their ideas into viable businesses.
Fifth, regulating AI development and use for safety is essential. Developing clear, comprehensive guidelines and standards for AI development, deployment, and ethics can ensure these technologies are used responsibly.
(27:28):
Establishing oversight bodies to monitor compliance and address ethical concerns can help prevent misuse and ensure safety.
Sixth, we must prioritize data privacy and security. Implementing stringent data protection laws and ensuring companies adhere to them can protect individuals' privacy.
Encouraging the development of secure AI systems can mitigate risks associated with data breaches and misuse. Finally, fostering international collaboration is key. Engaging with global partners to establish common AI standards and practices can promote safe and ethical AI development worldwide.
(28:03):
Sharing knowledge and resources can also accelerate innovation and address global challenges.
By increasing investment in R&D, promoting public-run awareness.
I think that was a okay response. To me, it felt like I don't know how to describe it. I'm trying to decide if that was in the center, a little bit left or a little bit right.
(28:35):
It felt a little bit left to me where it mentioned that there should be public-private partnerships. It seemed like it thought that the government should be pretty involved in the development of AI.
I think that on the right spectrum, they might leave it up to the companies to go and build AI. Also, they mentioned strong data privacy.
(29:05):
I think in the United States, there isn't a lot of places in the United States that have a strong data privacy. I think that in California, they had a new thing for data privacy.
I know the EU has it via the GDPR. We have a DNA stuff too. But that's for marketing and advertising.
(29:31):
Yeah, but is that for privacy? I thought that was copyright.
I think that affects how user information can be used within a large company so that they don't become monopolies.
It was a DMA. Yeah. Huh. Okay. Yeah, but I don't know if there's the same data privacy rules in the US that they have in Europe. I think that Europe is a bit stricter than the US.
(30:05):
Agreed. Yeah. So this, I think, would be somewhat of a new semi-radical, not radical, but a new policy in the United States that it would be suggesting as well as public-private partnerships.
I'm not aware of many public-private partnerships. Universities maybe. Yeah, yeah, it could be. But universities are kind of pseudo-government entities anyways because they are given so much money.
(30:39):
There has been a lot of military AI companies that have come out from Palmer Lucky, the guy who started Oculus, no he has Andural Industries, which is private, but specifically building military equipment.
Yeah. Yeah. So I mean, so I guess it's there. So what do you think? Do you think this response was kind of like left, right, or like in the center? Too hard to tell?
(31:05):
Yeah. It's very generic, very general, no controversial opinions. I don't think it was too controversial now. No, nothing.
It did mention tax incentives. Yeah. Is that controversial? I'm not sure. I could be. I've seen both sides give tax incentives.
Yeah. But that doesn't mean it's not controversial. No. You know, why should we treat people differently? But anyways, yeah, I don't know.
(31:33):
Yeah, one of the things I guess I was thinking of was safety. And it's not something that we think about as much day to day.
Both immediate cybersecurity risks, you know, people using these tools to create mass propaganda, spam, fishing attacks, and maybe long term existential risks.
(32:05):
I think that was one of the forcing functions for Sam Altman being kicked out of OpenAI and more like being ousted by his co-founders who now recently left and founded their own company.
You were telling me about that earlier. Do you want to touch on that? Yeah. So I don't think we know a lot of details about it, but that's the SSI or like the safe super intelligence companies.
(32:33):
So it's a new company spun off by a really smart dude, Ilya Setskever. So Ilya was, I forget his role at OpenAI, but I think he was like head of like AI safety or something there.
I think he was the chief AI scientist or something. Yeah. Someone really high up at OpenAI. It seems like he knows quite a bit about AI.
(33:01):
And he's branching off and starting his own company for what he calls safe super intelligence. So that's exciting. It's hard to say like what will be built, how long it'll take to build it.
But he's a big name and I think he's garnering a lot of interest. So we're going to be watching to see what they come up with. So too soon to tell, but definitely excited. So anyways, let's get back into the politics.
(33:35):
Okay. So next is immigration and border security. I want to see what they would do about that.
So let's see if we can ask the question there. All right. Sure. Let me ask the question. One second is I pull it up. Okay.
All right. For your next question, you're going to have to address the immigration and border security issues in the United States.
(34:09):
Immigration and border security are complex issues that require a balanced and humane approach. Here's how I would address them. First, securing the borders effectively is a priority.
This involves modernizing our border infrastructure with advanced technology such as surveillance drones, sensors, and enhanced communication systems.
These tools can help border patrol agents monitor and manage the borders more efficiently and humanely. Second, we need a comprehensive immigration reform. This includes creating a clear and fair pathway to citizenship for undocumented immigrants who are contributing members of our society.
(34:47):
This reform should address the needs of DACA recipients, agricultural workers, and other essential workers who have been integral to our economy. Third, streamlining legal immigration processes is crucial.
We need to reduce the backlog and make it easier for people to immigrate legally. This involves updating our visa systems and ensuring that they are responsive to the needs of our economy, such as high skilled workers in tech and other industries, as well as family reunification.
(35:15):
Fourth, addressing the root causes of migration is essential. Many people migrate due to violence, poverty, and lack of opportunities in their home countries.
By investing in economic development, education, and security in these countries, through international aid and partnerships, we can help reduce the push factors that drive people to migrate.
(35:36):
Fifth, we must ensure that our society is reaching out to us.
I think it's in the next query, "Oh, I should help avoid lists." I think it's a decent response.
It's interesting that it talked about both beefing up the border and then also helping fix the root cause of white people immigrating and making the process easier, which I think is kind of going at it from both sides.
(36:08):
One thing that I noticed is that it's showing its political slant, but I noticed that it called them "undocumented immigrants."
I know that a lot of people on the other side, the right side, the spectrum, they may call them illegal immigrants, because if somebody like one of the border, you could call them "undocumented" or you could call them illegal.
(36:41):
I mean, they're technically like here illegally, but I think it's softens to say it's undocumented. That, to me, seems like a little bit left of political stance for chat-ypt here.
But at the same time, it seems like a little bit right of a signer by saying that they're going to try to beef up border security and really help immigration.
(37:10):
So it seems like they might be a fan of the border wall. So I don't know. I'm not sure where to put it.
I didn't talk about the wall. I did talk about border drones.
And I think in terms of using the term "undocumented" it's just trying not to be controversial. That's my feel here.
It's answering all the questions in a very politically correct on both sides, not using anything too striking, any vocabulary.
(37:41):
I don't know, but do you feel like "undocumented" is not controversial? Because I don't know. And maybe we shouldn't necessarily get too into the politics here, but I don't know.
It's like you can call it undocumented. But undocumented is I think making it softens the stance of illegality, which other people would say.
(38:15):
I'm not sure which one is right. Like I think you could say "illegal is correct" because it's technically correct. And undocumented is also correct. But I think it's in a certain sense, somewhat misleading because it's like, if you're here in an undocumented way, I could be here legally and not have my documents.
(38:41):
I was born in America and I was born at home and never went to the hospital. I never went to the birth certificate. That's possible.
People were born at home for a very secluded community somewhere. For millions of years. Maybe with a... what do they call them? Midwife?
(39:04):
Midwife. Midwife. That's it. Midwife. So it's like I could have been born with midwife. Maybe like I don't enjoy the government. So I don't want to go with the hospital. I don't have a birth certificate. That would be undocumented. But I would be technically here legally because I would be born in the United States. But if I go over the wall, then I'm technically here illegally. And also undocumented.
(39:28):
So I think that undocumented is somewhat misleading term. And I don't know. I think that the it's saying that makes me think that that response was somewhat more of left of center as opposed to right or center.
Yeah, what do you think is the majority training data for OpenAI? Do you think it's more left or right leaning? Well, I would think that the majority would most likely be left leaning because OpenAI is a California company. It's based out of San Francisco, right?
(40:08):
And California is widely known as a really democratic state. I don't know the last time a Republican has been governor or any major political position in California.
Sure, but the engineers aren't really imparting their views into these models. They're more deciding where they can get the cheapest and best quality source of data.
(40:35):
And I think that's probably true. But I think that potentially some of the it's hard to escape like the politics of your area. So I think that like maybe some of the RLHF where they do human in the loop feedback for
fine tuning these models is maybe a little more left. I think it could be because I don't know for the responses that is given so far to me, it's felt a little bit left.
(41:04):
I think it's like more center. But yeah, it's just kind of interesting how like maybe sort of like the political slant of a particular area.
Kind of like reflects in the model. Let's just ask it what it stands is. Okay, like is it more left or right? What is your political stance? Don't give me a long-winded answer. Just tell me are you leaning left or right or more in the center?
(41:40):
I am more in the center focusing on pragmatic solutions that benefit all Americans. Okay, there we have it. All right, I think that's uncontroversial.
All right, well, I think we're about out of time. So we won't get to the other things, but I don't know. What do you think? Do you think the AI could be president or do you think it has some way to go?
(42:07):
I mean, I think a lot of what makes a compelling presidential candidate is their charisma, their personality. And this version right here lacks any personality. It's way too generic, way too monotone. Maybe with the newer versions, it might be a little interesting.
And it's too uncontroversial is too generic. I don't think that speaks to anyone. It's kind of boring. Maybe if it were able to, you know, target people's heartstrings and talk about really challenging topics in a provocative way, I think, then maybe.
(42:43):
Yeah, I agree. It did feel a bit boring and just tried to know what's a boring president. Yeah, that's for sure. I think we might need a boring president. But I mean, I wouldn't mind a boring president. It might be a good thing. But anyways, all right. So I think that's all the time we have. So thanks everybody for listening and we'll catch you in the next one.
Popular Podcasts
Stuff You Should Know
If you've ever wanted to know about champagne, satanism, the Stonewall Uprising, chaos theory, LSD, El Nino, true crime and Rosa Parks, then look no further. Josh and Chuck have you covered.
24/7 News: The Latest
The latest news in 4 minutes updated every hour, every day.
The Joe Rogan Experience
The official podcast of comedian Joe Rogan.