China’s AI and robot revolution

WEBVTT - China’s AI and robot revolution China's artificial intelligence and robotic scene is moving really fast. China's new five year policy blueprint, released in March, lays out the country's ambitions to aggressively adopt AI and dominate emerging technologies such as quantum computing and humanoid robots. On this episode of the Business of Tech, I'm joined by two New Zealanders who've just had a front row seat to that transformation. They've come back with stories of hospitals, banks, councils and factories already running AI and automation at industrial scale, and a robotics industry that's exploding in both ambition and the sheer number of players.

I'm your host, Peter Griffin, and in this episode, we're heading to China, not literally, but through the eyes of two Kiwi technologists. They've just completed an AI discovery tour across some of the country's most advanced AI and robotics hubs. My guests are Auckland based AI entrepreneur mng Chuk, co founder and CTO of Element X, and christ Church AI engineer Blake Harkness, who advises New Zealand businesses on how to adopt AI and runs the Young Kiwis in AI community. The pair joined a group that spent a week in China recently as part of an AI Forum New Zealand and New Zealand China Council delegation.

They visited the likes of robotics maker Unitry, as well as hyperscalias like Ali, Baba, Cloud, Baidu, and also Zai, a large language model developer, as well as the China Construction Bank and chin Hua Agent Hospital, which apparently is the world's first AI controlled hospital. They saw everything from sovereign AI offering zaim governments, to live deployments of AI agents and banking and public services, to humanoid robots and robotic dogs. One of the big themes in this conversation is just how visible AI has become and everyday life in China, from super apps that bundle payments, chat and AI translation to hospitals where triage chatbots greet you at the door and robots dispense prescriptions.

AI isn't something people are piloting in China, It's already woven into the fabric of daily services from millions of users over there. China may not be the leader in AI development yet, but it certainly is a leader in the adoption of AI. So we'll dig into China's rapidly advancing AI stack, including its strength in open source, large language models and emergence of sovereign AI offerings designs so countries can train and host models on their own infrastructure. That open source strategy means that while geopolitical tensions limit how much Chinese proprietary tech flows into Western markets and vice versa, Chinese models and tools can still be adopted and adapted by organizations globally, including here in New Zealand and then as the robotics Ming and Blake, as I said, visited Unitary and other makers in a sector that now boast something like over one hundred humanoid robotics firms, plus a growing ecosystem of quadrupeds and specialized robots for search and rescue, asset maintenance, and logistics.

But running through all of this is attention we need to grapple with. In New Zealand and across the West. China is racing ahead in robotics and key aspects of AI, but rising geopolitical mistrusts, data sovereignty concerns, and national security rules mean that much of that technology may never be directly available in our markets. At the same time, there are real opportunities for collaboration, whether that's around open source model, sovereign AI buildouts, or joint projects in areas like health, infrastructure and manufacturing where both sides stand to benefit.

So stay with us as we explore what China's AI sprint looks like on the ground, what it means for New Zealand's own AI ambitions, and how we should think about partnership in a world where technology and geopolitics are increasingly intertwined. Blake and Men, Welcome to the business of tech. How are you both doing? Yeah?

So very good. Yeah. I mean you've just come back from China where you had a front row seat into what is going on in China in the area off AI and robotics and technology in general. So I'm really keen to get your perspectives on this because I've been seeing these really amazing updates on LinkedIn from the likes of Graham Mullow from tech Instead, who was on a trip and others who were traveling around China and obviously saw some really mind blowing stuff.

So keena get into that. But first just one of learn a little bit more about yourselves. Ming I've had something to do with you in the in the past on AI related sort of case studies and that sort of thing. But Blake, I haven't met you before.

Tell us a bit about your background. Yeah, awesome. I did a megatronic engineering degree and about halfway through that CHEPT came out and I checked all my exams and assignments through and realized everything I was being taught AI could answer. I was like, oh wow, Like what am I going to do for a career?

Am I going to find a role? So I kind of went all into AI from there. So at the moment, I run my own consultancy. How can I SAI helping out businesses run across New Zealand implement and utilized AI.

But I also run young qs and I helping bridge that gap from young people get into AI careers enrolled. It's brilliant. Yeah, well, yeah, I'm sure a lot of people are having that realization as well. They might be two or three years into an engineering degree or something like that, thinking they're going to be doing software engineering and then discovering seeing what's happening with Claude alone and get hub Copilot and thinking that might not be an option.

For me. So yeah, I think that story is going to resonate with people. Man, you're the co founder and chief technology officer of element X. Tell us about what element x does.

Yeah, absolutely, so. Yeah. Elemen X was actually born out of University wiklinn when me and my co founder Daniel another business partner Richard as well, we started a company about back in twenty thirteen, and that was around when not so much when chat GPT was out, when Alex Nett and your network's deep learning started becoming a thing, and kind of since then, we saw a bit of an opportunity to bridge i guess cutting it, cutting edge research that's coming out of the university's to businesses and through our kind of research academic background and also you know, involvement in i guess the startup community, we saw a real gap there in terms of bringing this technology and putting into business hands.

So yeah, hence why elman X we've since then been helping putting AI into businesses hands, especially large enterprise prizes where it can get quite complex with security requirements and also a whole bunch of otter factors as well. Yeah, you've got some great case studies of projects you've undertaken. You're involved with the AI Forum as well, so showing some sort of leadership there, which is great. Tell Us about this AI Discovery tour off China, how did that come together and what really was the the thing that you're keen to get out of it?

Yeah, I mean this Day I Discovery Tour, I mean it was a really really packed program, really interesting companies. They're highlighting some of the best of AI robotics and robotics in China. So we visited quite a wide range of robotics companies or which had you know, their own focuses. We went to some of the hyperscalas and the frontier LLM labs building I guess the top Chinese coding models, just top Chinese lllms in general, and quite a lot of different different organizations like the municipal which is our city councils essentially, and also you know a hospital where they're really integrating the stuff into the operations.

Yeah, and Blake, what's the sense when you sort of arrived in China? Was at your first trip? Yes, it was how visible is AI in everyday life? You know?

Last time I went there, it was really all about you know, the souper wraps like we Chat and now everyone using QR codes to pay for things. It was the start of the electric vehicle revolution, so we were seeing lots of evs and electric buses and that it's very much now about AI. How visible is it in everyday life? Yeah, I think it's insanely visible.

Whereas we got off into the airport, the first thing we saw was a huge Olympic ex AI banner in the main kind of central area there as well. As we were driving around to the hotel, we just look out and see AI on different buildings just kind of stand out as well. I'm not sure what these companies are doing, but they're doing some form of AI. The other thing, like you mentioned is super apps, as we have.

There's all in one apps that do anything that you want to do in China there and almost all of them have some form of AI. They've got AI for translation now which makes it easier for searching app. It's just embedded everywhere. Yeah, and I guess is it pretty clear that they're on their own tech stack when it comes to AI.

We've got obviously ten cents are huge over there. As a Hyperscala app developer, Ali barber, You've got work buddy, which is sort of like the co pilot of choice. Up there was this a whole new ecosystem of apps and services that you might not necessarily be using here in New Zealand. Absolutely, and child Blake feels this as well.

But when we kind of got there, you basically had to switch out all of the apps that we're using, from you know, mapping to communication to ordering things. So it's a completely different ecosystem and a super interesting one as well, because I mean a lot of people and including myself, kind of see China as have skipped few steps when you know, during their technology adoption, especially you know, with payment systems, they kind of skipped the credit card phase I suppose, and went straight into mobile payments, you know, and even terms of mobile computing, they kind of skipped laptops, went straight to mobile phones and everybody's you know, doing things on their mobile phone.

So you had a completely different ecosystem and almost a different trajectory I suppose that they took. And I guess one other interesting observation at the end of the trip as well, a couple of companies talking about I guess how they're helping Kiwi businesses market into China and they're talking about these these platforms there. It's a little different then, I guess what we're used to because each of these platforms like bike dancers, TikTok, you know, and some of the yellow providers, they're kind of almost closed ecosystems in the massive ecosystems and themselves, which is you know a little bit different from what we're used to, where what we're using is you know, spread over quite a lot of different platforms.

And I think you know, in New Zealand in many respects, we're still sort of in part mode for a lot of the you know, really innovative uses of AI. But Blake, I think what you saw there visiting hospitals and factories and the like, you know, they're deploying it sort of add an industrial scale already. Yeah. Absolutely.

I think one of the most mind blowing ones was the hospital for me, This one that we visited served around five million patients every single year, which is very close to the population of New Zealand. And this is just one of hundreds that they have around and so the whole kind of process was using AI and automation to really speed up that efficiency. As you go in, you speak to an AI chat blot that could speak any language to you, that does the first triage. Everything, as you mentioned, is done off a QR code.

Did you go and sit and weight the scans, MRIs and cts. They reduce the time for doctors to analyze them by eighty percent, using AI to help that piece out as well. Once you're done with the doctors, you've got your scans, you've got a prescription that's also an automated system that robotics takes them all off the shelves and delivers them straight to you, all of a QR code. So efficiency is at almost every single point journey.

Wow, and that's exactly the sort of efficiency we need with an aging population, major strain on health sector resources. So that's the transformation we need to go through. Did you get a sensething of the models and a tech that's underpinning this. You know, we had the Deep Seek moment, you know, a couple of years ago, which really showed the world how quickly things are moving in China.

Did you get a sense of the models that they're using, whether they're customized ones or whether they're using the equivalent of the Claudes and the open ais in China to build on top of. Yeah, so I think it was quite a quite a mix. So, I mean, definitely there's a lot of great domestic models they've been building, and the model labs like z Ai that were talking to, we're very much aware of, you know, well, they've got strong ambitions and they're very much aware of how they're placed against some of the US models as well, the US model labs like the open AI and fropics, so and you know, and they're doing a really impressive job in terms of closing that gap in open source.

China's really been leading on the open source large language model piece and it feels like it's you know, eminent where that gap completely closes. I think we're only a few months away now. So yeah, so obviously I think a lot of the homegrown models are being widely used. I do suspect also some of the US models are being used as well.

So yeah, but yeah, no, definitely definitely, really really strong in that development. And I guess the advantage of open sources, like with deep seek, you know, Western companies can adopt it. I use perplexity and deep Seek was built into it. From basically week two that it was available, so I can draw on on deep seek and it's not centrally hosted, so it concerns about stuff going through service in Beijing isn't an issue.

It's sort of central to do. You think too, the speed at which China can move is that open source ecosystem they're fostering. Yeah, I think it's actually quite a clever strategy, that open source ecosystem, especially since you know there's a bit of concerns obviously when you're using tech China they've got different kind of national security rules and you know there's obviously a bit more fear sometimes especially in the Western markets, but about you know, using some of the technology, but going open source means that you know, they provide models that you can deploy in your own hardware and in your own systems.

That even when we're talking to Zai, they offered a sovereign AI service essentially where you know, you bring your infrastructure, you bring your team, you bring your data, and they will kind of act as consultants to help you train up your own sovereign model. And you know that way you really own that whole stack yourself, and you know your country sales are really interesting approach I found. And yeah, I mean the open source. I'm a big believer of open source and I think it's great for the world.

And you know, especially we've seen actually even just over the weekend, how some proprietary models from US company could you know it could be shut down over night. I think that was quite a mind blowing thing for me that I didn't expect at all. When we're talking with Zai as well, that you kind of get that Western view that AI, and especially in China. You're kind of losing a bit of that data access.

You're not really sure what's how it's been used. But for them to in the presentation say, hey, we're actually working with I think was a Singaporean government potentially to actually set up sovereign AI for them, that was quite unexpected. That's incredible and it's good to hear because we're having increasingly having discussions about what is sovereign AI? Is it the models, is it where the data is resident So you know, five years ago we weren't really talking about data sovereignty really that we were welcoming to hyperscalers in here.

So now we're actually having that discussion. I think which is a infection on where the geopolitical situation has gone in the world. You know, a lot of companies and particularly government application sensitive data they want to have it maybe on locally owned operated infrastructure, and that now is going to apply to artificial intelligence. So how do we And that's a really intriguing model that ZAI is adopting, where they're saying, we will help you build sovereign AI on your own infrastructure using our intelligence.

We haven't really seen that an operation from the big I guess western AI providers, at least in this part of the world. What was the sense she got Blake around agentic AI and the extent to which that's being used, because you know, once again we're pretty well penetrated with AI assistance co pilots across most government departments here. Most enterprises might be using the enterprise version of open AI or claud or co pilot, but it's a bit patchy in terms of where agents are being used for, you know, to automate workflows end to end.

Maybe call centers are at the front line of that. But do you get the sense that a lot of Chinese companies and government departments are already deploying agents. Yeah, it's a hard thing to tell. So one of the ones we met with was one of the banks, the big four banks, and they said they had about four hundred AI deployments across the business.

We didn't really get too much detail of what those AI deployments actually looked like in regards, but we know that through many with I do inside of their app, they have kind of their agentic phone control as well that is accessed to kind of do all that search and tool in for you. It was quite hard to get those real example use cases. I think a lot of them would focus on that higher level piece in there. I think I was going to touch on that same thing pretty much, and especially that example of the China Construction Bank where yeah, they had quite a lot of deployed applications.

Yeah, they didn't quite say what they were. However, it was interesting. I think one of the delegates asked a question where the answer and I think it was around, you know, how do they provide that platform or how do they encourage adoption throughout the bank? And I think the answer there was around you know that how they create they created a platform where the employees can experiment with their own use cases.

So it was actually you know, up to the employees themselves, the staff to find the use cases rather than fully being topped down in that regard. And it was kind of a let's see a bit of an organic adoption in the sense that people would you know, create these little agents that would do something for them, and they would publish it so that all of the staff could use it. And you know, ones that kind of picked up traction, you know, they stayed around, they could see there was usage, and obviously there was it was solving a problem, and you know some that were maybe less used, you know they were they kind of just faded away, so to kind of let the employees almost discover the use cases themselves.

And that's probably what led to the almost four hundreds of so did they talked about the. Other one on there probably is around the equivalent to the councils over there, and we had a tour of all of the different AI pieces that they had to them. Quite interesting ones was they had little self served chaosks for law and so you could go and ask this AI chatbot, Hey, this is a situation that I'm having. What is the best way of.

Approaching and dealing with this with the local regulation the laws, and it actually help bide those answers out. But also health screening as well, where you'd get your fingerprint scanned, you'd have a photo of your tongue, all of these kind of pieces that are available just to the general public, and that kind of also boosts I think those use cases when people can see, hey, this is actually benefiting me where I asked for this benefit. That's what you're saying, thing about sort of letting employees' experiment and then if it scales and if it's successful and it proves its its investment case, it could be scale across the organization.

We've seen some New Zealand companies like one end Zed done that to a very successful degree. What was essentially got like you know, China's famous for its five year plans, so It's political system allows it to think very long term and to put a lot of resources into areas it prioritizes. And China has said by twenty thirty, you know, it wants to be the world leader in artificial intelligence. Do you see that manifesting when you go around businesses and councils and that there is clearly a determined push to embrace AI.

Especially when we're looking at the bigger companies that we visited, like the bank for example, that was a big part of it. They're going, here's what the plan is. This is all the stuff that we're going to be doing to align with that plan and keeping it at that high level piece. The other thing that was quite interesting as well is the even though the competition between the big banks, they knew how many agents the other banks been deployment.

There's almost a bit of that competition sharing knowledge between each other, but also aligning that straight to the plan, whereas I think a lot of the more startup culture aligns to it, but they have their own ambitions as well. Yeah, and definitely I think my feeling was quite a coordinated effort nationwide kind of kind of feel that you know, that plan being distilled down, you know, to the banks for example, which you know which which I think it definitely does feel like it will help the nation as a whole advance into that, you know, into basically to their targets.

And I guess probably interesting as well in terms of how business is done in China. I think generally businesses are it's in their best interest to align with the you know, the the government strategy. I think someone one of the one of the businesses that we mentioned even mentioned that specifically was that you know, you know, when you are aligned, you do get you know, a lot of a lot of support. So that kind of you know, helps with that alignment as well.

Yeah, and we saw that, you know, in the Chinese government, you know, probably over a decade ago, really prioritized digital services and e commerce and that, and so the bids in the ten Cents and Ali Barber's became massive and sort of then drifted away from the government in terms of alignment, and we saw the you know, the backlash that that created. So I guess the government's thinking, how do we sort of keep control of AI but allow innovation to flourish. And when you're putting a lot of resource into everything from the universities, I mean, the papers that Chinese researchers are publishing and AI you know is some world leading stuff going on there, but you also are seeding the startup ecosystem but also giving the word to government departments and enterprises you need to get on board with this.

It's quite a powerful thing. Yeah, absolutely, I think it was an Internet plus plan which I should learnt after the trip that they had, and a few other things that you know, that really boost the e commerce and you could see, well, we could definitely see that e commerce was extra. It was on steroids to me, like you know, and how how you know, widely propagated that was. And you'd mentioned some companies got too big as a result, and I think we, you know, we might see the same thing as well as some of the companies might get a little bit too big and too much power centralized in one company.

So yeah, yeah, the e comma side was mind blowing as well that many were Ali Baba Cloud. One of their core offerings to help small businesses is sitting them up with live streaming and so they'll go out on a farm and they'll help set them up to actually live stream them shopping and going, hey, this is the stuff that I'm selling, talk about it while they're doing the work. And that's how they drive a lot of sales for small businesses, which is the complete opposite here in New Zealand.

Oh that's huge, just just thousands and thousands of channels sort of selling by those shorts of means. Yeah, it hasn't really taken off in the ways. It will be interesting to see if it does. One other thing I'm reading about, I don't know if you saw any evidence off was the rise of so called digital employees and China.

They're basically AI agents that have job titles. They even have their own KPIs. Did you hear any talk or did you see any of that in use. I didn't hear too much in terms of the employee side.

But as we're talking about the live streaming, one of the examples we got given when we're at Baido was AI generated live streams, and so they would take clones of celebrities, obviously with their permission, they would have them sitting there on a live stream like a talk desk, and they'd have all the comments streaming through, and a I would actually answer all the comments whilst trying to sell the product as well. The interesting thing there is that they actually clearly state in the live stream that it is AI generated, but they found almost no effect on the sales that compared to like a real person actually doing it and having that notice there, which is quite an interesting one in terms of that digitization of people.

Yeah, that was super interesting. Like when they mentioned digital humans now our business does work with you know it providers like Unique and we've been building digital humans at avatars, but a lot of those applications were more one to one where you're asking you know us in question. It's giving advice guidance. This was one two millions, right like you know and and applied to an already booming industry of live streaming, e commerce, and yeah, Blake mentioned already it sounds like it's been working really well.

So a big part of the trip was visiting the likes Uni Tree and robotics makers. I think there's something like a hundred humanoid robotic companies in China now, so like EVS, they've just gone big on robotics. I guess, you know, we see the sort of humorous, the viral sort of videos of Chinese robots is stumbling and smashing into hundreds of pieces as they do these sort of sports tournaments and races that they're doing. That's sort of the I guess, the funny end of it.

But they are did serious about robotics. What did you see, you know, what were some of the most impressive things you saw at the lights of Uni Tree when it came to robotics. I guess to me, maybe wonder well. One of the things I was quite impressed with is how almost you know, production or consumer ready they looked.

You know, I could probably have one of these in my household. I couldn't right now because there's still some ways to go in terms of using in general purpose environments like like a house. But you know, compared to some of the older humanoids that have seen maybe like ten years ago, bulky lots of wires, tevered, you know, deeds were really really slick. I think that was my first first impression.

Yeah, Passive Blake, any thoughts from you as well? Yeah, I think the humanoid ones is the hard piece. Like Asming mentioned that there's so many aspects to having a full body, legs and everything, and the use cases are quite small at the moment. With how well they can be created.

For example, there's ones that are in the pharmacy, but they only need wheels to move around and move they don't need the full leg components as well. I think the hardest piece is that simulation data compared to the real world with so many different environments and places that it could be working in, and so the big push for a lot of them is actually, how do we get this real world data? Do we tape a phone to someone's head and kind of record them doing different actions? Do we use via headsets full body control?

Getting that data piece? The real practical use cases I think come down to almost like there was robotic dogs, which was quite interesting, but also the price of them. One of the ones that has a forty light camera. You can buy a pbcheck at the moment for about five thousand New Zealand dollars.

And there's a lot of things around search and rescue, asset maintenance. If it's on a substation at an electrical distribution center, no, that's fairy weird. There's a lot of use cases like for search and rescue asset maintenance, going around looking at building sites, doing scans and taking photography as well, or just even carrying loads. I think that data collection piece, I think that was quite universal across all of the companies that we talked to.

Everyone basically said, you know, like they're they're hungry for you know, real world the data environments that are you know, not just labs, and there's a big data collection I guess thing going on actually across the world. You've got I think recently, I saw the iron z article where in India they've been paying people to put these you know, I think, recording devices while they're doing task household tasks. I think in US as well as a company Shift I think it was called, actually spun up from a German company and they are offering free household cleaning services in exchange for these you know, cleaners being able to you know, strap on like a like a camera and record other data to basically collect that training data for these robotics applications.

So that's really kind of the next phase is you know, having enough data for to generalize and not just having control environments, you know, like industrial warehouses, et cetera. Yeah, it's sort of the equivalent of what the autonomous vehicle makers sort of went through. And it was a bit easier for them once they got permission to go on the roads of California. They could drive every inch of those roads and eventually, you know, way Mow and Tesla managed to have incredible knowledge of those roads and how cars perform on those.

Going into someone's factory or someone's house is a different ballgame. And I guess to what extent did you get the sense that there's real sort of integrated AI in these robots as opposed to just sort of prescripted motions or you know, it's all basically. On a script. Did you get the sense that the embodied AI got to the point, whether it's in a robotic dog or a humanoid or even a factory robot, that it's making decisions autonomously.

For specific use cases. Definitely. One of the examples is in Family Mart. I think they've deployed over two hundred robot workers there that it's there.

You can speak to it in Chinese and ask it, hey, can you go grab me a hot dog? Can you grab me a drink? It will walk over there and actually grab that for you, serve it up, and you can pay for it there and have a conversation back and forth. And so that's not really that kind of scripted one.

But I think it really depends on those use cases. You have to have all that training data to get it really really reliable in that environment. And I kind of call it a semi semi autonomous at the moment, because you know, it's not you know the old old robotics, you know, like the robotic arms, where it's literally a planned pre plan path, right, but like it's hard coat off anything shifted, then you know it's it can't it can't perform the action anymore. So they are we're talking about they're using think Vision Language Action Models VILAS.

That's one of the core technologies that's driving this. And then also they're looking at world action models as well. I think someone that I'm moving towards that to get a bit more of this understanding of the world. And I think how they talked about how they for example, pair and new autonomous pharmacy where they would go and collect you know, the videos of I think the products, et cetera, and feed those videos into their training loops.

So they do need to collect a bit of data feed in the training loop, but they don't have to like prescript the actions, if that makes sense. Right, based on your impressions, you know, we hear a lot about China. You know, once they have a million robots in place, you know, I think this year or next, So that's a huge scaling up. Can you see that emerging that that possibility in the US.

Obviously, Elon Musk has bet the future of Tesla on Optimus robot and having hundreds of thousands and millions off those deployed across all sorts of labor, from surgery, he's saying, through to care robots. From the glimpse you got of what's going on in China, can you see world like that in the near future. Yeah. I mean, to me, China is the country of scale, and I think and rapid adoption as well, you know, with their rapid adoption of evs.

I think another thing we noticed is that most of the cars on the road are now at EV's and they actually cleaned that air right up Blue Skies, which I think was a you know, rare in the past. But you also, I think the just the general let's see mindset of the consumer and also you know, of the employees is that if it's bringing value, they're quite quick to adopt it, you know, if they see value if as solving problem for them. Blake, you know, in terms of I think Ming said that, or it might have been new.

The digital sort of avatar stuff has sort of been accepted. It might be the visage of celebrity and Chinese consumers sort of know it's not real, but you know, they've embraced it. I think it'd be a massive back. I had to be honest, if you saw, you know, a Boden Barrett or someone doing that here in New Zealand, they just they just laughed off the internet.

So there's obviously an acceptance there. But I wonder, Blake, what since you got in terms of how people are receiving the AI revolution in terms of the workforce. Is there fear there about automation and what it's going to. Mean for employment the opinions of the kind of employee level, because a lot of the people we're speaking to are at that senior decision making level.

Did those sorts of people talk at all about what they're planning to do to either upskill or reskill workers or redeploy them if their job is being automated. Was there any discussion of. That, Yes, I think a lot of the discussion was more around their amplification. When we were talking about is there going to be job lost?

What do you see there? It was a bit of hesitance to answer the question, I think fully, But a big point that they kind of talked about was the fact that there are an aging population. They already don't have enough force, especially in the healthcare sector, to deliver the results they need for such a huge population, so potentially it is unlikely to see as much job lost because of that, And like as you mentioned, it's more on the amplification and then also kind of potentially switching into other roles and reskilling there.

They really they really did emphasize amplification part, which I think is the case for most most countries. But I think there's also an interesting case where I think it was last year that one of the Chinese courts ruled that replacing a worker with AI was not valid grounds for dismissal. So that kind of set a precedent and kind of interesting precedent as well. You can't just go replace someone with AI, which is I think I don't know if that's happened here in other countries, but yeah, and I mean Train has also been struggling a bit with I think you've unemployment as well.

I think is one of the one of the highest in the world. So they're kind of battling two things at once. I suppose, you know, I really want to prove the efficiency and productivity, especially as you mentioned before, peaching population, and they mentioned the same thing aging population as one of the big challenges they're trying to solve. But you know also also yeah, they employment side of things.

They've clearly got this strategic approach to the amplification the penetration of AI across Chinese society. Did you get a sense that they're also thinking about the things that frankly have slowed down the uptake of AI here, which is around governance and maybe ming with your AI forum had on you know, this is something that the forum thinks about a lot. How you deploy AI safely, how you make sure it's explainable if people have decisions go against them, that it's explainable why that decision was made.

It's not just a black box. There's bias issues, there's hallucinations, things like that. Any sense about how the Chinese are dealing with these issues. I guess definitely there is I think in China, and I haven't looked at the specific details, but I think there is a lot of guidance around you know, ligne of the models that they're training, and you know there will be things obvious things that they're you know, making sure they're they're aligned on.

But also I think public safety has been you know, a theme and I remember in I think it was by Do they were talking a little bit about their automnus vehicles and despite you know them aiming for I think it's level four, they've got level four driver list taxis now, they did say that, you know, there's it's it's still quite strict in terms of the safety regulations and you know, they're they're they're deploying them in a very kind of lessly cautious and conservative approach.

So, you know, despite having all this you know, cutting edge technology, there still seems to be a lot of you know, guidance and safety aspects in general, and that you know would probably propagate to the training of the lms as well, and especially as they get smarter and heading towards you know, artificial superintelligence. Like one of the things I think that again is an advantage of the political system over there is the government basically has control and oversight of everything.

So when it comes to public services, the data may be in different places, but it can be all brought together to get a view off a particular citizen or a company or an organization. We're probably not quite ready for that level of concentration of data in New Zealand. We're putting social welfare data, tax data and health data together probably legally probably not possible to some extent here, but amazing what you can do when when you have that richness of data and that centralization of data.

What and then you apply that to artificial intelligence hugely powerful. Yeah, So one of the ones that was really impressive to me was the Ali Baba cloud. And obviously they have so much information out there. One of the product services that they're kind of offering is for the new market research.

If I want to get a new product out there, how do I figure out who my target audience is? How do if it's a clothing brand, which was the example, what is the style that I'm going for their outfit? Because they have so much data around consumer trends on these mega apps, they have all that information that they can use and so companies can pretty much use AI to analyze all the consumer trends and data based on their brand information. They're targeting everything and design all their new products kind of using AI and getting really really good accuracy and the results.

Hey, just finally, as you sort of came home from China, we're mulling over what you saw in comparison. Obviously, where a fraction of the size of one city of China, so very different scale, centralized command system versus a democracratic sort of distributed system to some extent, but anything you thought we could really apply there And do you think there is real scope to collaborate with Chinese robotics and AI companies? Yeah, I think for me it was just mind blowing seeing the scale, just how big their population is and their deployments on scale.

I think a lot of us in New Zealand go oh, I've got one hundred and fifty s. It's quite a complicated thing to roll out AI for when you look at them when they have hundreds of thousand stuff. Some of these huge companies as well. The thing that was most passionate for me is around that young KEYBIS and AI.

So how do we actually get our younger generation into these roles. We have a lot of tech unemployment, just general young people unemployment. At the moment. Are universities as a.

Whole pretty much banning AI if not teaching the very most basic level, and so we're getting a huge mismatch for what employers are needing with these AI skills and what universities are teaching. One of the lunches we had was with some distinguished professors and one of the questions we asked is what what degrees should students be studying at the moment. I would have thought originally would be really focused on the engineering the steam related fields, but they pretty much said the degree doesn't matter as long as you're learning about AI and their process, and that the universities, although they don't have it one hundred percent right, are focusing the shift onto the problem solving.

Here's a problem, here's your assignment. It's not about the end solution because aiking get you there. It's what are the tools that you used, what is your thought process? How did you get there?

How can you communicate that clearly? And why is this the best solution? And so that's the flip. I think a lot of the New Zealand education needs to be looking.

At the moment. Yeah, yeah, I guess couple of observations from me. I mean, like, you know, they've they've gone very much a top down approach with the you know, their National Air Blueprint essentially the AI plus blueprint. You know, we're probably less of a top down approach but more more bottom up, I guess in comparison, but it was encouraging to see, you know, from that previous example at the bank, where you know, within the organization such a big organization, they have a bit of a bottom up approach where you know, the employees were finding their own use cases and you know, I trust you know, order companies and organizations and using startups and established companies be able to find some really good use cases.

And I think where we can play as a nation would definitely be the application integration layer. You know, sure we don't have the manufacturing the scale you know that that China has, for example, but I think we do have some very clear, clever innovators and you know, coupled with a bit of that number eight wire mentality, I think that's where it can really shine. And I think one of the companies we visited, they're actually there were New Zealand grad started a company, Atom Intelligence in Shanghai and actually bringing the company back into New Zealand there, you know, bringing their headquarters back to New Zealand.

One of the things that they cited was the talent, the talent pool in New Zealand and actually they said, you know that the New Zealand team was way more productive than they're they're the local team over there, so yeah, we've got we've got that, I think you teald Vantage. They also highlighted a raw opportunity was the application there as well, So yeah, so that that that that was super interesting. And also the you know, the data sovereignty aspects as well. I think one is you know, when we are ready maybe at some point in the nation to start bringing some of the model that I guess the AI stack in house.

But also I think they've got quite interesting data sovereignty rules as well. They've got very strict export controls on data you know, of their citizens. At anytime a company like well One, all of that has to remained i think in the country, and if it is exported in any way, there's quite a few approval processes they need to go through. So yeah, that's it for this episode of the Business of Tech.

My thanks to Ming Chiuk from element X and AI engineer Blake Harkness for sharing their insights from inside China's AI and robotics ecosystem. It's been about six years since I was last in China. That was pre the COVID era, and that was actually in the midst of rising tensions about Huawei, the Chinese tealercommunications company that was effectively banned from being involved in building five G infrastructure in New Zealand over security concerns. That was a very symbolic example of the fracture in tech development mainly between the US and China, which extends to semiconductors, AI, lots of sensitive sorts of technologies.

I think what Men and Blake saw in China is a reminder that while political and security concerns are reshaping how tech flows between China and the West, the underlying innovation race is very much still alive, and New Zealand has choices to make about where and how it plugs in if it does at all. If you've got thoughts on today's conversation or examples of New Zealand China collaboration in AI and robotics, I'd love to hear from you. You can find me on LinkedIn or via Business Desk.

If you enjoyed this episode, please follow or subscribe to the Business of Tech and leave a review. It really helps other listeners discover the show. I'm Peter Griffin, Thanks for listening, and I'll catch you next time on the Business of Tech.