Today in AI: 🚀Nvidia Announces $3,000 Personal AI Supercomputer Called Digits 📱Samsung Goes All-In on AI at CES 2025 🎣Study: AI Phishing Achieves Alarming Success Rates 🌍 AI Systems Found to Emit Significantly Less CO2e Than humans
26:31 Share

All right. So it's January 7th, 2025. And, uh,

AI news is like... It's crazy right now. It's a fire hers. It really is. I know. It's like every hour there's something new. Yeah. Like mind-blowing. Some crazy new development. So what we're going to try to do in this deep dive is sort through all that news and... Pick out the gems. Pick out the gems. Yeah. The most interesting stuff. And we've got a lot to talk about. We've got tech announcements. We've got new research studies. We've got opinions from experts in the field. And it's going to be a wild ride.

I think so. So let's just jump right in. Let's start with Nvidia. So they just came out with this device called Digits.

Yes. And it's basically like a little personal AI supercomputer. Imagine the power of a data center crammed into something the size of, I don't know, a Mac mini. It really is amazing. Yeah. I mean, you're right. It's like having a data center. Yeah. In your home or office. I mean. Crazy. That used to be limited to big companies, you know, Google, Facebook, and now anybody can have it. For the small, small price of $3,000. Well, yeah.

So it's not cheap. But you know, compared to building a data center, it's pretty affordable. Yeah. That's insane. I mean, so what can you do with it? Well, the heart of this device is their new chip, the GB10 Grace Blackwell chip. And that's the real star of the show. Okay. It's specifically designed for AI.

And we're talking about, you know, the ability to run models with up to 200 billion parameters. Okay, hold on. Hold on. Now, we've got to break that down for a second because, you know, we talk about parameters a lot in the world of AI. Yes. And, you know, to somebody who's not like, you know, a free chip designer, it's like, what is a parameter and why should I care? Yeah, that's a great question. So a parameter...

You can think of it as like a tiny little knob in this vast network of artificial neurons. Okay. And when the AI is learning, it's adjusting all these little knobs, these parameters, to try to match the data that it's being fed. Right. And the more parameters you have, the more complex and nuanced the AI's understanding of the world can become. Okay. So basically, more parameters equals a smarter world.

A smarter AI. Yeah. Okay. So 200 billion, that's a lot of knobs. That's a lot of knobs, yeah. For context, some of the most advanced models right now, like the large language models, like ChachiBT, they run on something like 100 billion parameters. So this Digits device is doubling that. It's doubling the smartest AI. Pretty much. Wow. That is- It's pretty remarkable. So what are people going to do with this? Well, that's the exciting part, right?

I mean, the possibilities are endless. You could train your own large language models. You could have it, you know, write different kinds of creative content. You could even have it write code. You could run complex simulations, you know, to model things like climate change or the spread of diseases. So anybody could be like a scientist now. Yeah, pretty much. With this little box. That's amazing. That is amazing. And maybe a little scary. A little bit, yeah. Okay, let's talk about OpenAI because...

Their CEO, Sam Altman, has been in the news a lot recently talking about AGI, Artificial General Intelligence. Right. Remind us, what is AGI and why should we even care? AGI, it's like the holy grail of AI. It's basically building a machine that can think like a human. Right. I mean, really think like a human. But the same kind of general intelligence. Right.

The same ability to solve problems in all sorts of different domains. So not just like one AI for writing poems and one AI for, you know, writing code and one that does it all. That's right. You have one AI that can do it all and do it well. Wow. And what's interesting is Altman isn't just saying they're going to build AGI. He's saying OpenAI is confident they can build super intelligence AI that's smarter than humans. Okay. Yeah.

Smart as... Yeah, it's a little... It's a bit unsettling. That makes me a little nervous. I'm not going to lie. Well, he's talking about things like massively increased abundance, you know, solving all these global problems. Oh, yeah. So like the good side. The good side of it. Yeah. Yeah. But I just, I don't know. I'm a little... I understand. I understand. And it's interesting, right? Because just a few months ago,

Altman was actually fired by the board at OpenAI. Oh, yeah. That was crazy. It was huge news. Yeah. And then there was this big backlash from employees, from investors. It was like a movie or something. It was really dramatic. And then he got reinstated. Yeah. It was wild. But it kind of shows you how much is at stake here, right? Absolutely. I mean, the race to AGI is really heating up. Yeah. And

It's clear that people have very strong opinions about how it should be done, who should be in charge. Right. I mean, it's a really fascinating time in AI. Yeah, and it makes you wonder, like, what are they actually working on over there? Right. Like, what's going on behind closed doors? Yes. What are they hiding? Who knows? But we'll be watching closely. Yes. Okay. So that's AGI. That's super intelligence. That's crazy. Let's talk about something a little more, you know, down to earth, close to home.

our devices. So at CES, which was this big tech show, Samsung came out and they basically said they want to put AI in like

Pretty much. Yeah. I mean, their vision is to have AI just seamlessly integrated into every aspect of our lives. So like, you know, your TV, your refrigerator. Right. Your appliances, your smartphone, your car. They even showed off some robots. Robots. So it's like we're living in a sci-fi movie now. It's getting close. I mean, some of the stuff they were showing was really impressive.

So like give me an example like what's an actual thing that AI is going to be doing in these devices. So for example they're talking about TVs that can do real time translation. Oh wow. So you could watch a movie in any language without subtitles without subtitles. Exactly. Wow. That's cool. They're talking about laptops with AI powered photo editing features. OK. So you could take a

picture and have it automatically enhanced retouched. Okay. So like anybody can be a photographer now. Pretty much. They're talking about, you know, appliances that can anticipate your needs. Right. Okay. I got to admit that's cool and a little creepy, right? Like having all my devices, like constantly watching me, listening to me, trying to like predict my every move. It does raise some privacy concerns. Yeah, it really does. I mean, who has access to all that data?

How's it being used? Right. These are questions we need to be asking. Yeah, absolutely. Okay, well, that's something to think about. I mean, on the one hand, it's like, wow, that's amazing. So convenient. Yeah. But also, like, is it worth giving up my privacy for a smarter refrigerator? It's a trade-off we have to consider. Yeah. Okay. So let's move on to something that's a...

a little less convenient and a little more scary okay let's talk about cyber security because there's this study that came out recently about ai powered phishing attacks yeah and it's not good no it's not good like buckle up everybody because this is uh this is a little alarming basically they found that ai generated phishing emails are just as effective as the ones that are written by humans and not only that but they're way cheaper to create and way faster to deploy

Yeah. It's a pretty significant development in the world of cybersecurity. So it's like AI is giving hackers superpowers. In a way, yeah. It's like they can just crank out these phishing emails like it's nothing. Yeah. And they can tailor them to specific individuals. They can make them look really convincing. Right. They can even adapt to changing security measures. So it's like a constant arms race now. It is, yeah. It's a constant cat and mouse game between the security researchers and the hackers. So what can we do about it?

Can we even stay ahead of this? Well, education is a big part of it. People need to be aware of the dangers of phishing and how to spot the signs of a suspicious email. Right. Strong passwords, multi-factor authentication. These are all still really important. Okay. So it's the same advice as always, but it's even more important now. Absolutely. And the stakes are higher. Yeah. Okay. Well, speaking of things that are...

potentially dangerous. Let's talk about meta. So they just announced that they're ending their fact checking program. Yeah. This was a big surprise. Yeah. Like what happened there? Well, it seems like they're shifting away from relying on independent fact checkers. OK. And they're moving towards a more crowdsourced model, kind of like what X is doing with community notes. So instead of having experts like vetting the information. Right.

They're just going to let like anybody say whatever they want. Well, the idea is that users will flag content that they think is false or misleading. OK. And then other users will review those flags and kind of vote on whether they agree or disagree. So it's like a popularity contest for the truth.

In a way, yeah. And there are definitely some concerns about how well that's going to work. Yeah, like what if the people who are voting are just like biased or misinformed themselves? Right. That's a big risk. Yeah. And I mean, let's be real. This is probably at least partly about like trying to appease the new Trump administration. Yeah, I think that's definitely a factor. Because he's been, you know, very critical of social media companies, accusing them of censorship and all that. Exactly. And so, you know,

Meta might be trying to avoid...

Any potential regulatory headaches. Right, right. So it's like, are they really doing this for the good of the platform or are they just trying to like protect their own interests? It's hard to say for sure, but it's something to keep in mind. Yeah, definitely. Okay, so we've got AI powered phishing attacks. We've got the potential for more misinformation on social media. This is all very cheery stuff. Well, you know, it's not all doom and gloom. Okay, good. Because I need a little bit of hope right now.

Let's talk about the metaverse. Okay. Because remember how Samsung was talking about putting AI in everything? Yeah. Well, they're also making a big push into the metaverse. Right, because everybody's talking about the metaverse these days. It's the next big thing, apparently. And it's also kind of confusing. It is. I mean, it's still a very nebulous concept. Yeah. So for those of us who haven't quite like wrapped our heads around it yet, can you give us the metaverse 101? Sure. Okay.

Think of it as like the next evolution of the internet. Instead of just browsing websites and scrolling through social media feeds,

The metaverse is supposed to be this immersive, interactive, 3D experience. Okay. You know, you can put on a VR headset and walk around in these virtual worlds. You can interact with other people. You can play games. You can even go to work. So it's like stepping into a video game. Yeah, kind of. But it's also supposed to be more than just a game. Okay. It's supposed to be a whole new way of living, working, socializing. Wow. That's a lot to process. It is, yeah. Yeah.

And Samsung's vision is that AI is going to play a huge role in making all of this happen. Okay, so how does that work? Like, how does AI fit into the metaverse?

Well, remember all those AI-powered devices we talked about earlier? Yeah. The smart TVs, the appliances, the robots. They see all of that as being connected to the metaverse in some way. Okay. So you could be walking around in a virtual world and your AI assistant could be reminding you about appointments in the real world. Wow. Or you could be watching a virtual concert and the AI could be adjusting the lighting and sound based on your preferences.

Okay, that's pretty cool. They're even talking about using AI to create virtual avatars that look and move just like us. So it's like you're really there, but you're not. Exactly. It's a very strange and exciting concept. It is. It's very strange and exciting, but it also makes me wonder, like,

who's controlling all this AI and how do we make sure that these virtual worlds are like fair and equitable? Those are really important questions. We don't want to create a virtual world that just replicates all the problems of the real world. Right, right. Okay. Well, that's something to think about. But let's take a break for now. Okay. And when we come back, we'll delve even deeper into this crazy world of AI and explore what all of this means for the future. Sounds good. I'm ready for more. Welcome back.

So before the break, we were in the metaverse. Yeah, it's hard to even remember now. I know. We've covered so much. It's like information overload. It really is. But, you know, we were talking about Google DeepMinds project, the one where they're building AI simulations of the physical world. Yeah, that was pretty wild.

Right. And how that could actually be a key to achieving AGI. Okay. Yeah. You mentioned something about like job postings that gave us a hint about their AGI ambitions. Yeah. So one of the postings specifically mentioned that they think

Training AI on video and something called multimodal data is crucial for reaching AGI. Multimodal data, that's a mouthful. What does that even mean? So basically it means training AI on different kinds of information at the same time. Like think about how we experience the world, right? Yeah. We use our sight, our hearing, our touch all at once. And that's what multimodal data is combining those different senses, just like our brains do.

Okay, so they're teaching AI to process information more like humans do? Exactly. They believe that's key to really understanding the world and developing a more human-like intelligence. Interesting. Okay, but how does simulating the physical world fit into all of this? Well, by training AI on these really detailed simulations, they can teach it to understand cause and effect. Okay. To learn how actions lead to consequences. Ah, okay. So it's like a giant virtual playground where the AI...

the AI can experiment without, you know, breaking anything in the real world. Exactly. It's like those flight simulators they use to train pilots. They can crash a plane a thousand times virtually and learn from each mistake. Right, right. Okay, that makes sense. And so they could use these simulations to train, like robots or self-driving cars or even, like,

characters in video games. Exactly. Imagine playing a game where the characters aren't just following a script, but they're actually learning and adapting to their environment, making their own decisions, creating their own storylines. That would be amazing. That would be so cool. But is that even possible? Like, does Google really think they can pull this off? Well, it's definitely ambitious.

But they have a track record of tackling some pretty big AI challenges, like teaching AI to beat the world champion at Go. And they've been really successful with generative AI, creating text, images, music. Yeah, yeah. They do seem to be pretty good at that stuff. And didn't you mention that they hired this guy, Tim Brooks, who used to work at OpenAI? Right. He was the lead developer for their Sora project.

The one that's been making everyone freak out about deep fakes. Oh, yeah. The one that can create those super realistic videos from just text prompts. That's the one. So he obviously knows a thing or two about generative AI. Yeah, they've definitely got a dream team over there. They do. OK, well, let's switch gears for a minute and talk about something a little more down to earth.

We were talking about that study that found AI to be surprisingly green. Yeah, that AI could actually be good for the environment. Yeah. But I'm wondering, like, how does that actually play out in the real world? Is it really as simple as saying, use AI, save the planet? Well, not quite. You have to remember...

Training those huge AI models takes a lot of computing power, which uses a lot of energy. Ah, okay. So it's not just about how much energy the AI uses once it's up and running. Right. You have to consider the whole life cycle of the AI from the initial training to the ongoing use. So there's a trade-off there. Okay. I'm starting to see the bigger picture now. And it's a complex issue. But the good news is that researchers are working on making AI more energy efficient.

And more and more data centers are being powered by renewable energy, which makes a big difference. So there's hope that we can have our AI and be environmentally conscious, too. Exactly. OK, good. I need a little hope these days. But speaking of the future, we were talking about the metaverse before the break. Right. Samsung's big push into virtual worlds. Yeah. And it seems like every tech company is jumping on that bandwagon. They are. Yeah. It's the new gold rush.

But it's still a little confusing to me. Like, what is it really? And why is everyone so excited about it? Well, think of it as the next evolution of the internet.

Instead of just browsing websites, you're actually stepping into these 3D virtual worlds. You can interact with other people, go to events, play games, even go to work. So it's like a really immersive video game? Yeah, kind of. But the idea is that it's going to be much more than that. It's going to be a place where we can live, work, and socialize all virtually. And Samsung's vision is that AI is going to play a huge role in making all of this happen.

You know, remember all those AI powered devices they were showing off? The smart TVs, the appliances. Yeah, right. They want all of that to be seamlessly integrated with the metaverse. So like you could be walking around in a virtual world and your smart refrigerator could remind you that you're out of milk in the real world. Exactly. Or you could be at a virtual concert and the AI could adjust the lighting and sound based on your mood.

That's pretty wild. It is. They even want to use AI to create virtual avatars that look and move exactly like us. So it's like we're really there. Okay. That's kind of creepy, but also kind of cool. I know, right? Yeah. It's a strange new world we're heading into. It is. But it also makes me wonder, you know, who's controlling all this AI in the metaverse? That's a good question. And how do we make sure these virtual worlds work?

are fair and equitable. We don't want to just recreate all the problems of the real world. Right, exactly. We don't need a metaverse full of bias and discrimination. No, we don't. But those are conversations we need to be having now as these technologies are being developed. Yeah, absolutely. Okay, well, let's bring it back to the big picture for a minute. We've talked about NVIDIA. We've talked about OpenAI. Right, those are some of the big players in the field. Yeah. But what about Google? What's their take on all of this, especially AGI?

Google has been a little quieter about their AGI plans compared to some of the others. Okay. But their actions speak louder than words, I think. So what are they doing? Well, this DeepMind project, the one with the simulations, that's a pretty clear sign that they're taking AGI very seriously. Right. They're not just talking about it. They're putting their money where their mouth is. Exactly. And they have a lot of resources, a lot of data, a lot of talent.

Yeah. So if anyone's going to figure out AGI, it might be Google. It's possible. Yeah. But it's important to think about the potential consequences too, right? Right. We've talked about the benefits, like solving all these world problems. Right. But what happens if we create something that's smarter than us?

and it decides it doesn't want to follow our rules. Yeah, the whole be careful what you wish for thing. Exactly. We need to be thinking about those ethical implications involving ethicists and philosophers and social scientists in this process. Okay, so it's not just about the technology, it's about how we use it. Exactly. It's about making sure we're using it for good, not for harm. Well said.

But let's zoom out a little bit. We've talked a lot about the technical stuff, the societal impact. The big picture. Yeah. But how is AI actually changing our day-to-day lives right now? Well, in some ways it's becoming more invisible.

Okay, what do you mean? I mean, it's being integrated into so many things we use every day without us even realizing it. Ah, okay. So it's like AI is working behind the scenes. Right. But in other ways, it's becoming more present, more tangible. Okay. Think about those smart devices, the phones, the TVs, the appliances. They're becoming extensions of ourselves in a way. Right, right. They're augmenting our abilities, mediating our experiences. It's a really fascinating shift. Yeah. Like AI is becoming part of us.

In a way, it is. And that trend is only going to continue. As AI gets more sophisticated, it's going to be more and more integrated into our lives, shaping the way we think, the way we behave, even the way we see ourselves. Okay. Wow, that's a lot to think about. It's exciting, but it's also a little bit scary. I understand. It's a lot to process. Yeah. But it's important to remember that we have a choice in how we use this technology. That's right.

We're not just passive consumers. We can shape its development, make sure it aligns with our values. Yeah, absolutely. Okay, so speaking of values, let's talk about privacy because that's a big one. It is. I mean, AI systems need a lot of data to function. And that data often includes our personal information. Right, right.

Right. The more we use AI, the more we're exposing ourselves. In a way, yes. And that data is valuable to companies, to governments, even to hackers. Okay. So what can we do to protect ourselves? Well, there are a few things. One is the development of privacy-preserving AI techniques.

where the AI can learn from data without actually seeing the raw data. Ah, okay. So it's like the AI can get the insights it needs without knowing our personal details. Right. And there's also a push towards data minimization, only collecting the data that's absolutely necessary and deleting it when it's no longer needed. Right. So less is more, basically. Exactly. And there's a growing movement towards data ownership rights.

So individuals have more control over their data. OK, so there are some things happening, some positive developments. There are. Yeah. And there are regulations coming into play, too, like the GDPR in Europe, which gives people more control over their data. So it's not all doom and gloom. No, not at all. But it's important to be aware of the risks and to be proactive about protecting our privacy. Absolutely.

OK, well, let's end on a more creative note. Let's talk about generative AI. We touched on it earlier with, you know, DeepMind Sora and its ability to create those realistic videos. That's a pretty amazing example of generative AI. Yeah, but it's not just videos, right? Like what else can this stuff do? Generative AI can create all sorts of content, text, images, music, code, even 3D models. Wow. So it's like AI is becoming an artist, a writer, a musicologist.

a musician all at the same time. Exactly. It's really blurring the lines between human creativity and machine creativity. And what are the implications of that? Like, what does this mean for the future of, you know, art and music and all that? Well, it's a fascinating question. I mean,

Some people are excited about the possibilities, you know, the idea of AI being a creative partner, helping us to push the boundaries of what's possible. Others are concerned about the potential impact on human creativity. You know, will AI replace human artists? Yeah, that's a good question. It's a question we don't have the answer to yet, but it's something we need to be thinking about. Absolutely. Okay, well, we've covered a lot of ground today, and I think we need to take a break. Sounds good. My brain is starting to melt a little bit. Mine too.

But don't go anywhere. We'll be right back to wrap things up and maybe even try to predict the future. The future of AI. That's a tall order. I know, right? But we'll give it a shot. All right, we're back. I feel like we need a whole separate deep dive just to process everything we've already talked about. I know, right? AI is moving so fast. It's a whirlwind for sure. But before we wrap up, there's one more big topic we need to hit. AI and jobs. Yeah, that's the one everyone's worried about. It's

It's the elephant in the room, isn't it? Like, is AI going to take all our jobs? Well, it's definitely going to change the way we work. Yeah, we talked about that a little earlier, but can we dig into that a bit more? Like, what are some of the ways that AI is already transforming different industries? I think the biggest impact we're seeing is in automation. Oh, okay. AI is getting really good at handling tasks that used to be done by humans. Right.

Right. Like what kinds of tasks? Oh, everything from data entry and customer service to things like financial analysis and legal research.

So the more tedious stuff, the stuff that nobody really wants to do anyway. Yeah, exactly. Which frees up humans to focus on more creative and strategic work. Right. The stuff that AI isn't so good at yet. Right. And we're seeing this across all sorts of industries. Like in health care, AI is being used to analyze medical images, detect patterns in patient data, even assist with surgeries. Wow, that's incredible. So it's like AI is becoming a partner for doctors. Yeah, it's a really powerful tool that can help them make better decisions and provide better care.

OK. And what about other industries? Well, in finance, AI is being used to detect fraud, manage investments. In manufacturing, AI powered robots are working alongside humans on assembly lines.

increasing productivity and precision. So it's not just about replacing jobs, it's about changing the way we work. That's right. It's about augmenting human capabilities, making us more efficient, more effective. Okay. But there are still concerns about job displacement, right? Of course. And those are valid concerns. We need to be thinking about how to prepare people for these changes, how to make sure that everyone,

benefits from AI, not just a select few. So what can be done? Like, how do we make sure that AI is a force for good in the workplace? Well, education is key. Okay. We need to be teaching people the skills they'll need to thrive in an AI-powered economy.

Things like critical thinking, problem solving, creativity, skills that AI can't replicate. Right. The uniquely human skills. Exactly. We also need to be thinking about policies like universal basic income or job retraining programs for people who are displaced by automation. Right. So that people aren't left behind. That's right. And we need to be having open and honest conversations about the ethical implications of AI in the workplace.

Yeah, absolutely. Making sure that AI is used fairly and transparently. Okay, well, that's a lot to think about, but let's shift gears one last time. We talked about AI in the environment earlier. How AI could actually be good for the planet. Yeah, that was a really interesting study, but I'm still a little fuzzy on how that works. So basically it comes down to efficiency. Once an AI model is trained, it can create content much faster and with less energy than a human. Think about writing a blog post.

A human might take hours and AI can do it in seconds. Right. So it's not just about speed. It's about the amount of energy used. Exactly. AI can do more with less. And that's good for the environment. OK, that makes sense. So AI can be creative. It can be efficient. It can even be good for the planet. It's a powerful tool with a lot of potential for good. But also potential for harm if we're not careful. Right. It's all about how we choose to use it. That's a good point. And I think that's a good place to wrap up.

We've talked about a lot today. The latest AI news, the potential benefits, the risks, the ethical implications. It's a lot to process. It really is. But it's important to stay informed, to keep learning, to keep asking questions. I agree. We're all on this journey together. That's right. And it's a journey that's only just beginning. So thanks for joining us. And until next time, stay curious.