Kilda. I'm Chelsea Daniels and this is the Front Page, a daily podcast presented by the New Zealand Herald. Silicon Valley has been rattled by a low cost Chinese AI, with the startup claiming their deep Seek technology can emulate the performance of chat GPT at a fraction of the cost. It's launched, shook share markets and spawned allegations from Open Ai that their Chinese rivals used its work and models to make their own artificial intelligence.
This is all.
Happening as the new US administration appoints an AI and crypto star so the online currency can thrive.
So what is the.
Future of AI? And will we be buying all of our coffees in crypto sooner rather than later? Today on the Front Page, science and tech journalist Peter Griffin is with us to run through the latest advancements and shake ups in the tech space. First off, Peter, when we talk about AI, how would you explain what it actually is to someone who may have no idea?
AI is a feel that's been around for many decades, and I've been writing about it my entire career twenty five years or so. It's developed and become more powerful as a result of the computing power that can be applied to it. So you know, there are various forms of AI that are sort of domain specific. We've seen, for instance, computer vision. If you go into a supermarket, now a camera can identify your face potentially, and that's
been used for security. There are all sorts of different types of AI, but the one that is really current and attracting a lot of attention and investment and hype is of AI, and that is all around language using sophisticated computer models and processing power to process vast amounts of information and then to serve you intelligent answers based on that processing and that training off the model that has been done with literally everything that's out there on
the internet. So think about things like chat, GPT, clawed from Anthropic mid Journey, which is creating images based on text and puts. That's the one that has captured the imagination of billions of people around the world right at the moment.
Can you give us an idea of who the big players are in the AI spaces, particularly in the American sense. I'm thinking Open AI, Nvidia, the two that have become kind of household names at this point.
Hey, yeah. When open ai released chat gpt in November twenty twenty two, that was a game changer in terms of being able to use a chatbot and get really
high quality, intelligent answers. And that went from everything from just asking it for recipes for food that you're interested in putting some ingredients, get some recipes all the way through to computer coders were able to use chatchept to basically create computer applications from scratch without having to do very much manual program and you just tell chatchpt what you want in terms of a software application and it will do a pretty good job of trying to come
up with that. So open ai not a listed company, private company, but has attracted billions of dollars of investment, including from Microsoft, and the technology underpinning chat chept is now used for Microsoft Copilot service, which has rolled out to one hundreds of millions of computers as well. Other than that, you've got Meta, which is the owner of Facebook and Instagram. They have an open source AI effort
called Lamma. So open source means that rather than it being proprietary and them holding onto everything and keeping it close to their chest, they're basically giving away the source code to their language model, so anyone can use it. So that's a different approach, but it's not an uncommon approach in the history of software development. You've got other players like Amazon and Anthropic, You've got the French company Mistral, and then you've got all these Chinese companies which we
don't know very much about. Deepseeker is the one that has popped onto the radar just in the last couple of weeks, but there is a whole ecosystem of companies in China working on this as well.
Deep Seeker also said it only took two months and less than six million dollars chump change to develop the model, which is again much cheaper than other companies like Meta aka Facebook have spent on their AI models, So it's cheaper to make, cheaper to use. Generally, there is just a kind of beautiful irony to AI companies just being completely undercut by this, considering that the entire business model
of AI is to undercut people. You've been hearing NonStop from aibros about how people's jobs are going to be redund and people to be replaced by AI, and then, as Twitter user at Techborg beautifully put it, I can't believe chat GPT lost its job to.
AI, so Chinese AI that deep Seek one that dropped recently. The most notable part of that, I guess, is that the researchers claim that they were able to train their AI for around six million dollars US right, which is a fraction of what it costs to generate open AI. How significant is the arrival of deep Seek.
It is very significant, and this is reflected in what we've seen in the markets. The market to spook about this, particularly around stocks like in Video, which have really accumulated one hundreds of billions of dollars worth of value on the back of selling. In Video's case, the AI chips, the high performance chips that are used for AI workloads.
Other companies that are running AI services have taken a hit as well, because if what deep seek is claiming is true, you know, this company is able to do what those companies do at a fraction of the cost, and you know, so that completely undermines the business model potentially off those companies. There's a lot of caveat around this.
While deep Seek is a company producing open source models, it means that experts can look at those models, look through the source code benchmark the performance against others, and that's been done in recent weeks, and you know a lot of experts are saying, you know, this is legitimate. What there is a question mark over is the exact circumstances they use to build and train that model. They claim they did it for as little as six million dollars.
Is that the only cost. There's accusations that they have basically distilled all of the data from other models. Open ai is investigating at the moment the potential that deep Seek actually took all of the data out of open ai and used it to train its model. So data is still to wash out. There's not a lot of transparency often around Chinese companies and how they operate, and you've got the looming presence of the Chinese Communist Party
in the background. There's geopolitical issues here, there's competitive tensions, so a lot of this has still to wash out. But it looks as though, according to the experts, that this is a more efficient way of doing AI and that fundamentally changes everything, and I think in a good way as well. Because AI has been too expensive through energy intensive, we need breakthroughs to make it more accessible to more people.
Yeah, I've seen some commentary around how the US has invested significantly in the tech sector of the last two decades in particular, yet China continues to outshine those US creations. I mean, just look at how TikTok overtook all US social media platforms and basically reinvented the game. There have we been sold alive? Do you think by the US given the receipts we're getting perceived receipts I suppose about how much China is pumping into it rather than the US.
Look, I don't think it should be any surprise that China has massive capability in tech. You know, it was prioritized by the government. We have a communist party there that fues five year plans that says we want to buy twenty thirty b the world leader and AI. It has the freedom to do that, and every aspect of the economy and society jump into line and follow that. So we saw well over a decade ago the rise
of Ali Baba and ten cents. You know, these Chinese tech giants, they sort of hit a rough patch with the Communist party as they got too powerful. People like Jackiyan the equivalent really of Elon Muskin in China got too big for his boots and was sort of his operations were curtailed a bit by the Chinese government, So that the government there is happy for tech to thrive and flourish long as they have a visibility into everything
and apps like we Chat. They love the development of that because they can see every interaction, every financial transaction, every social media message that goes across that network. So they love tech as long as they have control it and it serves their purposes. So there's a booming industry there. It hasn't been so visible to us except when a
company like tektok pops up. But what we've seen with some of their tech giants like TikTok and Huawei is when they get big enough to go international, there's this tension around whether they are a national security risk to Western governments, and that's why we've seen Huawei basically banned from use in countries like New Zealand, Australia and the US, and we've seen the span of Tektok as well.
Donald Trump announced a five hundred billion dollar AI infrastructure program called Stargate prior to this change coming out is deep seak going to impact that potentially.
I mean, if this is born out, you know where this leads to. And I think the more interesting thing is what comes next because you know, all of the companies are pouring over a deep Seat to see how they did it. You Meta reportedly has set up a war room to get their top engineers to look at this. So then you know there will be reverse engineering if Deep Seat undertaken by these engineers, and hopefully they will
create much more efficient models based on that. But if you look at the investment last week up to five hundred billion dollars underwritten by soft Bank, Oracle and Open Ai involved. So Donald Trump was very proud about that, but will that investment actually stack up? Will it actually happen? Now, if these innovations actually flow through to American companies, we could see that POTENTI being done for a fraction of
the cost. And that's where you know, I maintain it's a really good thing because we've been on this sort of scary trajectory with AI where there's this race to build data centers and to build nuclear modular reactors to power those data centers. Because all of these CEOs of these tech companies are looking at each other and going, if we're going to win this race, we need to have more and more capacity, more and more data, more chips, and they're spending up big on that. If you don't
have to do that, that is a good thing. It means that there's less energy required, which is more sustainable. So hopefully you know a few companies outside of China, because let's face it, government departments, big US companies, New Zealand companies are not going to rely on Chinese developed
large language models. So hopefully this innovation will flow through to the Western world that they will incorporate it in and we will get much more efficient models that we're going to use every day, which costs a lot less because at the moment, something like Copilot, which is Microsoft Smart Assistance that you can use in Microsoft Word and Excel. Now that costs between thirty and forty dollars a month for one subscription for one person. It's eye wateringly expensive.
If that can be reduced to three or four dollars a month, now that's good for everyone.
Moving on to crypto. It's the world's worst kept secret. I suppose that Trump has been championing it. Recently, Trump media announced plans to invest up to two hundred and fifty million dollars US into investment accounts and bitcoin and similar cryptocurrencies or crypto related securities. How does Trump's vested interest in something affect the way the rest of the market reacts to it.
Well, first of all, it's a huge comfort of interest. You know, both he and his wife Millennia launched mean coins in the lead up to the inauguration, which massively spiked and value, and that's really dangerous territory because you know, they could collapse in values. So he could have supporters of Donald Trump who they jump on board that bandwagon
and then end up losing a lot of money. He looks to profit significantly if the value of these coins go up, which they have, So that's a huge conflict of interest. But you know, in terms of his philosophy and his approach to crypto, it's very much influenced by the PayPal mafia that now surround him, and some of them are people like Elon Musk, David Sax who's his crypto, so are Peter feel who actually is a New Zealand citizen but the founder of talent Heer and one of
the co founders of PayPal. A lot of these people sort of have a libertarian bent, so they really like cryptocurrencies, and they've been talking extensively to Donald Trump, influencing as thinking about it. But because they've been influential, and he is talking about low regulation of crypto and even having a strategic reserve for the US actually buying buying coin or seizing that bitcoin that has been taken from criminals
and building a strategic reserve out of that. That is what has led to in the last part of twenty twenty four, the massive spike in crypto fancis, particularly Bitcoin. It's why the market is point at the moment people are talking about this becoming finally potentially a means of exchange where people are using it to buy and sell things rather than just hoarding it as an investment. All of that really is on the fact of this sort of crypto friendly environment that Trump is creating.
Coin the other day, do you intend to continue selling products that benefit yourself personally while you're president?
Well, I don't know if it benefited.
I don't know where it is, I don't know much about it other than I launched it. I heard it was very successful.
I haven't checked it.
Where is it today? Thank you made a lot of money, sir?
How much million dollars? It seems like the last several days several billion.
That's peanuts for these guys.
Well, he faced criticism for launching that crypto meme coin on the eve of his inauguration. Is he able to do something like that because the space is so unregulated, Like, you can't manipulate the share market, can you in the same way, it seems that you can do with crypto?
Right. Well, you know, there's much more scope for things like insider trading, technical manipulation, all those sorts of things in a largely unregulated world of crypto. And we've seen that many many times where these coins have been set up and pumped up, hyped up by influencers and celebrities. And then there's the inevitable so called drug pool where once the price goes up, the founders are the people who own a big chunk of the coins, and second they can pull their money out. The job is done
for them. They've made their money and they don't care about anyone else, and usually the pricing collapses, confidence is lost. That's the thing about crypto. It's all based on confidence in these coins, and that goes for Bitcoin as well, and it's very sentiment driven, so it's very volatile. At the moment, sentiment is very positive around crypto, but it's
a little bit like deep Seat. You know, a new coin could come in that's much more efficient, that runs better applications, and suddenly the crowd will shift to that one and the old ones are not looking so good. So everyone needs to be really worry about about getting
into crypto. But underpenning it is technology called blockchain technology, which employs things like smart contracts which allow for a much more efficient way off exchanging things digitally, and we haven't seen that take off in a big way yet. So hopefully, aside from all the speculation and investment that is happening in crypto, we will see this year a
bit of an uptick in development of those technologies. And New Zealand has just announced that this year it will create a sandboxer regulatory sandbox to allow innovators to experiment with AI and blockchain and financial services in a safe way. So it's an opportunity for us to look at where the value is in crypto and blockchain and actually do it responsibly.
So you mentioned the PayPal mafia, and we know that Donald Trump is appointed former PayPal executive. You mentioned David Sachs as the White House AI and Crypto SAR. The move came after Bitcoin sword above one hundred thousand dollars US for the first time. How significant do you think this move is.
Yeah, well, it's David Sacks and Elon Musk and others who are very much in driver. So, Sir, Donald Trump doesn't know anything fundamentally about cryptocurrency. He started turning up at conferences when he was on the campaign trail, crypto conferences literally he was speaking at because the people who are advocates for this technology were funding his campaign and
he felt obliged to do that. And he has an obligation now to repay all of those billionaires who helped him on his campaign, literally funded his campaign and use their networks to spread support for him. So the payback now is a low regulation environment that allows them to move fast, make lots of money. And that has been the sort of the story of Silicon Valley over the last twenty years. It was starting to return actually under
the last Trump administration. That's where we started to see that the first big cases from the Department of Justice against Google and Amazon and Meta and Microsoft. It really accelerated under the Biden administration. Now we face the prospect of a lot of that being undone and going back to a sort of more of a lass a fair environment. He's already overturned Present Biden's executive order on AI, so
that's a really good indication. The SEC the Security and Exchange Commission, was a very hardline on crypto under Biden. So we'll see that dismantled as well as part of Trump's move to weed out the deep state and all the people who think differently from him and his coterie of tef bros. So I think we'll see just a much lower regulatory environment. And some would say at the moment that's what America needs to let its innovators move
quickly and be unencumbered by red tape. But we've sort of seen the downside of that, which is this massive inequality that it's created with a lot more power and money concentrated into the hands of people like Elon musk All.
We're here across all levels of government to ensure New Zealand takes a strategic, coherent approach to AI. I have established a cross party AI Caucus to support this and have tasked my officials to scope work on AI roadmap and risk management based guidance for business a by digitizing government portfolio. I have similarly tasked officials to advance work to support safe AI uptake at innovation in the public service to make it more efficient and responsive to the needs of New Zealanders.
These changes to AI and crypto overseas, how do you think it will affect New Zealand innovators?
Is it a positive thing?
And should the government do more to regulate both of those sectors as a result.
I think it's an opportunity for New Zealand. But we're at a bit of a fork in the road and this really came to light last week where the government laid out its tide for how it wants to reorganize our science and innovation sector. It's dis establishing Callahan Innovation after twelve years, so there's a lot of change going on at once to create a new scientific institution that will focus on artificial intelligence, on quantum computing, and on
synthetic biology so basically gene editing technologies. These are areas that we have sort of missed the boat on. Other countries like Australia have gone bag on AI and Quantum put a lot of money into them over the last twenty years, so they're a much better place to use these technologies for economic development and to answer some of
the big problems that they have. But you know, we do have an opportunity, particularly if the barrier to using these technologies balls, particularly around AI, we have the opportunity to use models like deep seek to come up with applications that help our businesses become really competitive and sell things to the rest of the world. So our edge is that we can do things cheaply. If you look at Peter Beck and rocket Lab, you know they were never going to be able to compete with SpaceX or NASA.
They found a way to do things much more cheap and that's what attracted attention from the rest of the world. You know, low cost launches to the lower Earth's orbits. So that's got to be our mindset. How do we take all of these tools and open source tools are great because they're basically free. We can take these tools, build on top of them and create something off value and that goes for blockchain and crypto technologies as well, but we just don't really have that much capability. We've
really hollowed out our science sector. We've got some great business people, but these are areas that to have an edge, you really have to have people, talent, expertise, and it's been a one way street really, people leaving rather than coming here in the last couple of years anyway, since COVID. We need smart people to be here and to develop stuff here.
Thanks for joining us, Peter, Thanks Chelsea. That's it for this episode of the Front Page. You can read more about today's stories and extensive news coverage at enzadhrald dot co dot MZ. The Front Page is produced by Ethan Sills and Richard Martin, who is also a sound engineer. I'm Chelsea Daniels. Subscribe to The Front Page on iHeartRadio or wherever you get your podcasts, and tune in on Monday for another look behind the headlines.