Menu
Sign In Search Podcasts Charts People & Topics Add Podcast API Pricing
Podcast Image

AI squared: AI explained

Ai in government plus why GPT 5 kinda sucks

16 Aug 2025

14 min duration
1829 words
3 speakers
16 Aug 2025
Description

Ai in government plus why GPT 5 kinda sucks

Audio
Featured in this Episode
Transcription

Chapter 1: What is the main topic discussed in this episode?

3.018 - 7.887 Ayush

Welcome back to AI Squared, where two minds explore one intelligent future. I'm Ayush.

0

8.869 - 21.608 Michael

And I'm Michael. So far, we've looked at how AI is changing healthcare, education, and creativity. But today, we're diving into something that affects all of us, whether you work in tech or not. Government and public policy.

0

22.409 - 35.992 Ayush

AI is being used to build smart cities, scan faces in public, monitor traffic patterns, and even help write laws. But it also raises serious concerns about privacy, bias, and surveillance.

0

36.012 - 42.903 Unknown

So today we're unpacking the role AI plays in the government, how it's being used, what's at stake, and why it matters more than ever.

0

44.005 - 55.028 Ayush

Let's start with something futuristic. Smart cities. These are cities where data and AI help optimize everything from traffic lights to trash collection.

55.048 - 65.78 Michael

Cities like Singapore, Barcelona, and Dubai are already using AI to monitor traffic, reduce energy use, and manage public transport, even responding to emergencies.

66.46 - 67.622 Ayush

It sounds amazing.

Chapter 2: How is AI transforming government and public policy?

67.922 - 73.508 Ayush

Smoother commutes, cleaner streets, faster emergency response. But what's the trade-off?

0

75.057 - 88.521 Michael

Usually, it's the ethics of data. To work, smart cities need a constant stream of surveillance, cameras, sensors, and location tracking. And that opens up major questions about who's watching and for what reason.

0

89.483 - 98.179 Ayush

And who controls the tech? Is it the city? A private company? What happens when that data is sold, leaked, or misused?

0

99.458 - 110.99 Michael

There's also the additional issue of unequal implementation. Wealthier neighborhoods may see better service while underserved communities may face over surveillance without the same benefits.

0

111.01 - 119.799 Ayush

That's a good point. If smart cities aren't inclusive by design, they might actually deepen inequality, even if they look more high tech.

121.06 - 129.875 Michael

Exactly. Efficiency can't come at the cost of equity. Now let's talk about something a little more controversial. AI and policing.

131.077 - 151.165 Ayush

Facial recognition is one of the most widely debated technologies today. It's used in airports, city centers, and even schools. But most importantly and most recognized, it's used to unlock your phone. But it's been shown to misidentify people, especially people of color, at alarming rates.

152.628 - 162.566 Michael

And it's not just facial recognition. License plate readers, audio surveillance, predictive algorithms, all of them are being quietly deployed in major cities.

163.748 - 175.905 Ayush

Predictive policing sounds like sci-fi, but it's already being used in cities like Chicago and Los Angeles. These cities analyze historical crime data to predict where crimes are likely to occur.

Chapter 3: What are smart cities and how do they utilize AI?

198.372 - 204.34 Michael

There's almost no public input. These technologies are adopted before citizens even know they exist.

0

204.843 - 215.882 Ayush

The good news, some cities are fighting back. San Francisco banned facial recognition. New York has proposed the Public Oversight of Surveillance Technology Act.

0

217.345 - 234.688 Michael

And additionally, advocacy groups like the ACLU and Algorithmic Justice League are pushing for stronger regulations and algorithmic transparency. And now let's shift gears to something more administrative, government services.

0

235.749 - 249.649 Ayush

AI is being used to process taxes, flag benefits of fraud, and even automate visa applications. In Estonia, for example, government paperwork is almost entirely digital and AI-assisted.

0

253.308 - 264.128 Michael

That can make things faster and more efficient. Fewer lines, quicker approvals, but automation also risks removing the human touch and thought from systems that affect people's lives.

265.01 - 276.852 Ayush

Take the case of Michigan's unemployment insurance fraud system. Thousands of people were falsely accused by an AI-driven algorithm and forced to repay money they never owed.

278.047 - 288.263 Michael

In the Netherlands, a welfare fraud detection system was found to disproportionately target minority communities, and it was eventually ruled discriminatory by a Dutch court.

289.445 - 296.616 Ayush

So, while AI can streamline services, it must be designed with fairness and accountability from the start.

300.582 - 311.233 Michael

One promising idea, AI... upwards people, independent agencies that oversee government algorithms, just like courts oversee laws.

Chapter 4: What ethical concerns arise from AI in smart cities?

368.91 - 385.218 Michael

And we need better public understanding. AI policy isn't just for lawyers and tech experts. It's for all of us. Now, this all sounds big and far away, but it's not. You can advocate for AI policy in your city, school, or country.

0

386.279 - 397.373 Ayush

Just ask questions. Who owns your data? Who benefits from AI being used in your community? And demand answers and demand a seat at the table.

0

397.393 - 406.384 Michael

Educators can teach algorithmic literacy and students can organize awareness campaigns. Communities can demand audits and public input on AI projects.

0

407.005 - 412.833 Ayush

AI governance isn't just for politicians. It's for everyone.

0

412.853 - 421.705 Michael

One last idea. Citizen assemblies diverse groups of everyday people helping shape AI rules. It's happening in the UK, France, and even Brazil.

422.886 - 449.232 Ayush

When people who are affected by AI have a voice and it's now in use, the outcomes are fairer, smarter, and more dramatic. All right, this week, OpenAI released GPT-5, and there have been some mixed reviews about it. So this is a special bonus episode in the middle of our regular episode of AI Square. I'm Ayush.

450.634 - 465.868 Michael

And I'm Mikko. Today, we're going to tackle something stirring up a lot of conversation and controversy in the AI world, GPT-5, and why so many users are calling it underwhelming.

467.211 - 485.845 Ayush

The hype promised a revolution. Bigger models, smarter reasoning, next level creativity, and even more accurate real-time knowledge. We were told it would be faster, more context aware, better at following complex instructions, and capable of deeper reasoning than any AI before it.

487.929 - 503.461 Michael

But in reality, For a lot of people, GPT-5 feels slower, less accurate, and more restricted than they expected. And instead of delivering a groundbreaking leap, it feels like a step sideways or even backward in some areas.

Chapter 5: How is AI impacting policing and surveillance?

690.539 - 701.393 Michael

Some tech bloggers speculate this could be due to model compression, cost-cutting, or even resource sharing between multiple AI products that dilute GPT-5's raw power.

0

701.413 - 716.293 Ayush

We also have to consider OpenAI's growing focus on brand safety. The more widely AI is adopted, the greater the pressure to avoid controversial outputs, even if that means sacrificing depth and originality.

0

720.931 - 729.544 Michael

This all leads to a bigger question. What do we actually want? Do we want a safe, predictable assistant or a bold, sometimes chaotic, collaborative?

0

730.565 - 740.88 Ayush

Because GPT-5 might be showing us the trade-off we're going to see with every next-gen model. It's going to be more polished, but less willing to push the boundaries.

0

742.48 - 756.081 Michael

In this episode, we'll dig into real user reviews, explore technical possibilities for why GPT-5 feels different, and debate whether these limitations are temporary growing pain or the start of a new, more restricted era for AI.

757.183 - 770.344 Ayush

Whether you're disappointed, satisfied, or somewhere in between, GPT-5 has definitely made one thing clear. The conversation about what AI should be is far from over.

771.522 - 780.235 Michael

So stick around. This one's going to go deep. Pulling no punches and asking the questions a lot of people are thinking, not saying.

782.218 - 805.056 Ayush

All right. So what does this mean about GPT-5? Well, GPT-5 is, as long as I've used it, just a cleaner user interface. That's the only difference I've noticed. AI in government is powerful, but it's also political. It reflects the systems we already have, for better or worse.

806.878 - 814.287 Michael

The future of AI won't just be written in code. It'll be written in laws, debates, and the choices we make together.

Comments

There are no comments yet.

Please log in to write the first comment.