AI and the Future of Work: Will Jobs Be Replaced or Transformed?

AI is moving faster than most people expected, and with it comes a growing sense of uncertainty about jobs, careers, and the future of work.

In this episode of Ella Podcasts, we explore the reality of artificial intelligence and whether it will replace jobs or fundamentally change them. While some see AI as a powerful tool for efficiency and growth, others are increasingly concerned about automation, layoffs, and long-term job security.

This conversation breaks down what AI actually is and why it is often misunderstood. Rather than true intelligence, AI is built on pattern recognition at scale, which makes it powerful, but also limited. We also explore the psychological impact, from displacement anxiety to the way people are beginning to form relationships with AI systems.

In this episode I’m joined by:

• Dr Jonathan Marshall, Clinical Psychologist and former professor
 • Duncan Reed, C-suite leader in technology transformation with experience at Microsoft, AWS, and Workday
 • Jeff Paine - Managing Director of PS Engage Global Government Relations

Quotes:

• “AI is not intelligent in the way we understand it, it recognises patterns.”
 • “Nobody is very sure what their job will look like in a few years.”
 • “It’s more likely jobs will change than completely disappear.”

Takeaways:

• AI is built on data and pattern recognition, not true intelligence
 • Job displacement is happening, but full replacement is unlikely in the short term
 • Many companies are still experimenting and not fully adopting AI at scale
 • Human judgement and context remain essential
 • AI is more likely to reshape roles than eliminate them entirely
 • Rapid change is driving uncertainty and anxiety

Timestamps:

0:00:07 Introduction to AI and job concerns
 0:01:03 What AI actually is
 0:03:01 Psychological impact and displacement anxiety
 0:04:17 Why companies are slow to adopt AI
 0:07:23 Why many AI projects fail beyond pilot stage
 0:10:52 Real-world use of AI in business
 0:16:38 Timeline of job automation and layoffs
 0:20:13 Will AI replace jobs or reshape them
 0:21:42 Impact on professions and human roles
 0:23:21 Risks and misuse of AI
 0:27:03 Automation, robotics, and future jobs
 0:31:06 Government response and future support

Conclusion:

AI is not just a technology shift, it is a human one. While there is disruption, there is also opportunity. The challenge is not simply whether jobs will disappear, but how we adapt to what comes next.

🎧 The latest episode of Ella Podcasts is now live. Watch on YouTube, or Listen on Spotify, Apple Podcasts, Amazon Music or Buzzsprout.

🔔 Don’t forget to subscribe for more honest conversations:
 https://www.youtube.com/@EllaPodcasts

🔗 Stay Connected With Us

Facebook Group: https://www.facebook.com/groups/816892014485089
LinkedIn: https://www.linkedin.com/in/ella-sherman/

#AI #FutureOfWork #ArtificialIntelligence #Careers #Technology #Automation

Transcript

SPEAKER_01 0:07

Hello, I'm Ella, and this is Ella Podcast. Everyone's talking nonstop about AI. Some people rave about it and the new efficiencies it's bringing, while others are concerned about their job security and the lack of opportunities for fresh graduates. Is there a risk of global mass unemployment? Here to discuss this hot topic is Dr. Jonathan Marshall. He's a Stanford and Harvard educated graduate who is a leading clinical psychologist and a former professor. We also have Duncan Reed, a C-suite leader driving technology-enabled transformations. He's worked in leadership positions for Microsoft, Amazon Web Services, and Workday. And we have Jeff Payne, the founder of PS Engage Global Government Relations, consulting on public policy for the world's biggest tech companies. So let's start with the very basics. What is AI?

SPEAKER_03 1:03

Oh gosh. So I I think what people see as AI is technology seeming to appear to automate things as if by magic. What's behind that actually is really enabled by data at a really huge scale and some automation that's actually typically been programmed by humans. So one of the things I do want to make clear pretty pretty immediately is that the intelligence part of AI, it's not really intelligent in the way that you or I might see or understand. What it is, is huge amounts of data with patterns. And what computers are really good at is spotting patterns that look like other patterns that they recognize.

SPEAKER_01 1:52

Okay. Jeff, you've got anything to add to that?

SPEAKER_00 1:55

Yeah, sure. So I look at AI from a more of a policy and regulatory space. I focus a lot on Asia, but if you look in Asia, there's three countries that have formal AI policies. We have Japan, China, and South Korea. The rest of them are either looking at exploring AI, what can it do for your economy, what can it do for your people? They're also putting in draft legislation that will be reviewed by industry players to ensure that they're putting up, you know, guardrails, not gates. They want to be able to leverage AI to grow their economies. I think everyone would have the same opinion that even if you don't think AI is good for the world, you have to at least embrace it and be involved in it for your country because each country is unique. And I think we'll talk a bit about some of the impacts it's having, uh, especially recently. It's a very dynamic space, that's for sure. You could do the news that we talk about today might be completely different in a week.

SPEAKER_01 2:49

So yeah, it's certainly moving so fast. And Jonathan, why is AI seeing rising unhealthy behaviors, declining psychological safety, and increasing displacement anxiety?

SPEAKER_04 3:01

Yeah, it's having a massive impact. Um, I think people are establishing relationships with AI. It's a source of intimacy. Um, a huge number of teenagers describe their closest friend as AI. And so as people become more and more attached to these things which seem personal and human but absolutely aren't, uh, that there's a disconnect there, and they also become more intolerant and incapable of relating to other people because my AI, for example, I was talking to some psychologist colleagues who were saying how when they go through transcripts of their sessions, and they they will say with confidence, oh, there were five topics or six topics, an AI will find eight topics. And you're like, you're a professional, and an AI does a better job at listening. Like, of course we're going to find a sense of intimacy. And as for anxiety displacement, nobody's very sure about what's going to be happening to their job in four years. And in some cases, much less than that. Yeah. So I think we're all a little bit dizzied by what is going on right now.

SPEAKER_01 3:59

Yeah. And I was reading um Deloitte's latest survey, and they found AI access in organizations has increased from under 40% to over 60% in just one year. So why haven't 80% of companies redesigned jobs around AI capabilities? What do you think, Duncan?

SPEAKER_03 4:17

Well, we're back into the human aspect, really, there. I would say that most companies are giving their employees targets to use AI in the workforce because they want to be out there talking about wherein AI enabled business, because that's a hot thing in the way that the the money markets view the world, right? So a lot of froth in that conversation at the moment. I think most importantly though, when when there are humans involved, there is inertia and there's caution and there's that kind of culture within organizations. Um and there's regulation, as Jeff will say, you know, just about every business in the world actually has to answer to a regulator at some level or another, even if the the regulator is simply are you paying people on time? So the idea that you could just immediately kind of swipe things out of the way and say we can redesign this to be replaced by AI is a bit of a fallacy. I think the other thing, and and to Jonathan's point before, um what people perceive as AI now is is a very subjective thing. Um, you know, technically these things are called large language models. They are large language models?

SPEAKER_04 5:31

They could be other models, couldn't they, for an AI?

SPEAKER_03 5:33

Um what people primarily work with now is a large language model, and they're they are technically they're probabilistic, right? What they do is match patterns, and they will, the machine will give you a pattern that looks like the pattern that you gave it. It doesn't have any understanding of that. So the context that you operate in in your business environment, the thing that controls that, the AI won't have any understanding of what that could be. So when things are probabilistic, um it can be a very useful term. Many business processes, however, are deterministic, where you need to have the same answer, the same mechanism, the same process running every time. I think most of us have seen if you if you ask Chat GPT the same question multiple times, you'll get a different answer. Now you can't bear that in a business. You can't have your call center operatives being automated and giving variable advice to people. It's why you've spent tens of years and millions of dollars making sure people are trained correctly to give the best advice. So you have to be choosing the right tool for the right job, and this is where appropriately companies are being cautious. You know, they want the thing being used and they want to see what efficiency dividend there might be. But to go into your business and say, well, we're gonna we're gonna replace all of our call center people, for example, that that's a really big ask because your regulator is gonna be all over you, and and all of that time and money that you've spent over the years building the capability for your teams to be able to give customers good advice and something that will stand up, you have to be able to do that. If you're gonna do it with technology, it has to be to the same standard.

SPEAKER_04 7:10

So you're saying a lot of companies are at kind of as a fad are going to AI but inappropriately, and actually it's not financially successful because they're coming with probabilistic solutions for deterministic.

SPEAKER_03 7:23

Yeah, we're definitely seeing that in the industry. In fact, there have been several reports recently, and I believe um certainly there was a report about Microsoft quite recently about uh companies experimenting with their co-pilot tooling. Uh, and something like you know, 70% of those experiments never go past the experimental stage. Most of the the AI programs that you hear announced in businesses at the moment aren't making it past the pilot stage because the business case doesn't fully stack up, or the the problem isn't fully solved. And and this is the you know, this is the challenge, is that it's we're super early in the technology here. And and um, you know, as I used the term before, there's a lot of froth in in this part of the industry right now. Um I met interestingly honestly. We need to come off, right? We we need to get this to a position where it's a bit more mature.

SPEAKER_01 8:18

I met somebody recently who has a tech consulting firm that works with all the major banks, basically analysing and auditing their software, hardware. And he said the banks are 20 years behind from a hardware point of view of even being able to embrace a lot of the new AI coming out. So, you know, you've you've you've got a situation where everybody wants it, but are they really ready for it?

SPEAKER_03 8:42

Yeah, and and look, you know, the the the company I currently work for, we we do human capital management software. Um and when we put teams into customers to help them, we're very often solving what you could say are quite basic problems, like I want to run the same payroll cycle across eight countries. That's still a really big challenge for lots of for lots of companies. Um oftentimes they don't even know how many people they have on the payroll. So so kind of getting to a point where they can you know use this magic technology to solve all of their problems, or that you know, we're still coming back to to basics. And one of the things we do find is is the mature question that gets asked is okay, so we've we've got all of our human people, right? We spend a lot of time and money making sure we know who they are and where they live and that they're qualified and certified and all those things. Can we do that with with AI agents? Can we treat those like digital, digital, proper digital workers? And that's the mature question.

SPEAKER_01 9:49

Wow.

unknown 9:49

Yeah. Okay.

SPEAKER_00 9:50

It seems to have accelerated though recently, and I think your point with the financial services, it's a heavily regulated area. So I think that's where regulation really comes into play. Uh, I think a company like Microsoft has done a great job at leading the way on things around ethical use of AI, how to you know deploy AI. So copilot, I have seen articles where it's underutilized. You know, Microsoft has a model where they sell a number of seats and uh you know they sell more copilots than are actually being used because people have yet to adopt it on a fuller scale. But if I had to look at you know some of the companies that are doing a good job, it's people that embed AI in their day-to-day real workflows. So it's not like a uh it's not like a show, it's not a demo or a proof of concept. It's so Microsoft, you know, they have copilot built into everything that they do, their application software, copilot as a standalone app and that. So it it is being used some places, and I think it will grow, but the regulated areas are harder to do healthcare, uh, financial services, education.

SPEAKER_01 10:52

I've loved using Copilot. It's really revolutionized the interview process and talking to candidates instead of having to write everything down. It's all recorded and it'll give you the key points. It's quite amazing.

SPEAKER_00 11:06

The other thing that's interesting is you mentioned there's a bunch of different kinds of AI large language models, and they're competing on a very fast timeline. So if you remember when when Google launched their first AI models, they were not very good. And then they all of a sudden leapfrogged many of the other companies by relief a new release that turned out to be really impressive. Um they do get better over time. Is that Gemini? Yeah. So they they get better over time. And some people will use some for different things. Gemini is great for doing infographics in a blog post, for example.

SPEAKER_04 11:39

Why why are we so why now? Like four years ago, I don't think anybody was talking about AI. Like what happened?

SPEAKER_03 11:45

Yeah, look, I I think um certainly the the hyperscale uh cloud service providers, yeah, uh and they and you know, this is one of a number of things that have been revolutionized through the the availability of of fully scalable compute power, right? The fact that companies don't need to go and buy, you know, oh, we've got a new project and and a new thing to do, so we need to buy 50 uh, you know, 50 servers, 50 big computers to do A, B, and C. They don't need to do that anymore. They can they can use cloud services to give that kind of scale.

SPEAKER_04 12:19

So so the accessibility of powerful machines. Yes, absolutely.

SPEAKER_03 12:23

That's part of it. And that'll be, you know, that's why companies like NVIDIA who make the powerful chips and machines that that just by chance happen to be good at this stuff, have done so well. Nvidia before the this AI thing came along, NVIDIA were actually a uh a card manufacturer to go into um uh computers that that kids like to do um high power gaming, right? And it's only only by chance that this that the technology requirements are quite similar. That's amazing. Yeah, yeah. Yeah, absolutely. Well, absolutely. If you if if you if you had a thousand dollars worth of shares in NVIDIA, you know, six or seven years ago, you'd be you'd be quite happy with that investment decision, that's for sure.

SPEAKER_00 13:01

But uh but I think also there's a race for AI. And I think you know the US is doing a job to try and race against China to be the leader in AI because the technology is advancing so quickly. And that's why you have issues around like um does the US want to limit exports of NVIDIA chips to China? Uh so there there is a race ongoing, uh, not only between each of the different AI providers, but it's a national.

SPEAKER_01 13:25

And who's winning the race at the moment?

SPEAKER_00 13:28

I th I think the US is. I mean, um, but it it's because it's I do think from kind of um a perspective of adoption with a lot of the corporations and things that I've been around. So there's there's always been that, even when you go back to cloud service providers, people were kind of, you know, they want to make sure that they don't have a vendor lock-in. So they'll get uh Google Cloud, they'll get AWS, and they'll get Azure from Microsoft. So I think there's a little bit of that can happen here as well.

SPEAKER_04 13:55

What's the next big thing that's gonna happen in AI? Is it agentic AI? And actually, what is agentic AI?

SPEAKER_03 14:00

Yeah, look, I I think there's a couple of things. I think to Jeff's point, uh things are moving quickly and things are getting more efficient more quickly. So that so you know there are a lot of AI companies out there today that have huge valuations and um they're viewed as being the leader, and in six months' time that may have changed significantly because um what technology continues to do and has proven out over the years is that things that seem to be special last year become commoditized the year after. And and certainly with the base infrastructure for for AI, if you if you're if you're producing a large language model and that that is your value proposition, in two years' time that won't be a value proposition anymore. But the large language model will just be something that you can run on any cloud service provider, and then where the value comes is from somewhere else. Now, agentix are a different thing, and that and that's what we touched on before, Jonathan, about the the use of the AI technology um combined with a level of of kind of programmed automation to help uh things that look like human task work occur. Um, you know, so if you think about a business process where um pretty standard information comes in, like um doing your expenses, for example, your pretty standard information comes in, you know, you have to sit and you take a photo of the receipt and then you have to key something into a spreadsheet or into a into some sort of system, and then it disappears to somebody in the expenses department who then has to review things and check for things that are invalid. Those sorts of tasks and activities can be largely automated through the use of agentic AI that will provide parameters on how correct does something need to be based on your risk profile and what steps can be automated.

SPEAKER_04 15:48

That doesn't sound so far from now. It doesn't sound so like why all the fuss about agentic?

SPEAKER_03 15:54

Well, I mean, my take is that people are getting ahead of themselves quite a bit with the agentic thing, and back to my point before, which is um again, if if you're if you want your business to run in a controlled and auditable way that meets your regulatory requirements, you are going to need humans in the loop of your processes. You you can't just turn it over to the agents, um, certainly not at this stage.

SPEAKER_01 16:20

Um, if we look at the humans, yeah, because I'd like to bring it back to humans and and jobs. I mean, Jeff, I know you've been looking into this. What's the timeline for AI replacing jobs on a wide scale? According to Deloitte, by the end of this year, 10% of jobs will be going automated.

SPEAKER_00 16:38

So that's the timeline. It's almost like technology growth in Asia Pacific. Um, it's always like uh better than it was expected to be. So I think the timeline is faster than people expect. Uh, if you look at the US big cloud service providers, they're investing so much in uh IT infrastructure for AI that it's for the first time ever they're having to borrow money to do it. And that's very unique for the big companies that are very profitable: Meta's, Google's, Microsoft's, uh, Amazon as well. So I think that it's happening a lot faster. So we've seen big tech layoffs in the end of 2025, the end in the beginning of 2026, um, big companies were investing heavily in AI infrastructure. Amazon, in since late 25, 26, it's been over 30,000 reductions in staff. Microsoft has got around 15,000. Uh Meta just announced last week about a 16,000 staff layoff. So I think when it goes back to the human element, people are skeptical and worried because if you have a job, and then when you use AI, you see how powerful it is. And I think that that's something that regulators keep an eye on. Um, but the companies are trying to do good things. If you look at the cloud service providers in the US, they're now going to build their own power stations to avoid increasing the electricity rates for normal citizens in America. It's obviously a big challenge they have. So they're looking at doing things like that. Um and even actually, there's a lot of people protesting uh AI infrastructure builds out in the US. So there's a you know still a regulatory hurdle that they have to get over to be able to build those data centers. So it's a very dynamic space.

SPEAKER_03 18:23

Yeah, I mean I I see it a bit differently. I think the compression in the technology industry is is a that that's a lagging indicator post-COVID, having having been in the industry for such a long time and seeing how I was part of this, how we overhired, genuinely overhired in the run-up to and then through the COVID thing. In fact, when I worked at one of the big technology companies I worked at, I remember saying, you know, what is my headcamp budget? What do you, you know, what do you want me to do? And they said, Oh, don't worry about that, just go and hire the very best people you can find and keep going. Yeah. And and and I was I did gosh, face-to-face interviews. I did 360 in about an 18-month period, face-to-face. So I was, you know, I was doing two a day, every day. Wow. Um, and and um so we're seeing some of that come out. I think, I think my my take, certainly in the tech industry, that some of the the compression we're seeing, AI is a convenient excuse. I think we'll also see, again, a bit of a lagging indicator, genuinely in the tech industry, using AI to help you in the early stage of writing your code, if you're a software engineer, really very, very helpful. Yeah, but your work then changes from spending, let's say, 70% of the time writing the code and 30% checking, flips in the other direction because you have if you're gonna roll some software out into any kind of production environment that a machine has written, you need to review that correctly in a way that you wouldn't necessarily have done had you written it yourself from the start. So it's more change in the dynamic of the role. Yeah, and I think we will see that initially. We'll we'll go through a peak, I think, where we'll see companies say, Oh, well, we you know, we're we're we're gonna reduce all of these roles. And then when reality sets in, it'll be like maybe it's more that we have to change the roles and change the way we think about things rather than a full a full replacement.

SPEAKER_01 20:13

Do you think it'll be more of a balancing act rather than mass global unemployment?

SPEAKER_03 20:17

Yeah, for sure. I mean you know, the the the part of my brain goes back to um so so so when when the telegraph was put coast to coast across the US, that was a major technology innovation, and and within one week the Pony Express went out of business, right? Which had been the only way to get your information across from coast to coast in the US beforehand. But there wasn't mass unemployment because of the telegraph, there was mass opportunity became because of the telegraph. And I think on the one hand, we we hear you know that the kind of the the doom side of AI could replace everything. What the the thinking hasn't caught up really clearly about well, what opportunity does it present and what other new things are gonna occur? Because helping always does that.

SPEAKER_04 21:07

Because even in my profession as a clinical psychologist, I was talking to colleagues just this week who are describing that they they're not filling their their slots, people who've always been oversubscribed. And uh unanimously across the table they're all saying you think it's AI, because a lot of clients are coming and saying, Don't need to see you this week. Uh I I've been chatting to Claude. And uh and I'm like, wow, already. And and and literally a year ago, the same group would have said, We, we we're not gonna be touched by AI anytime soon. Yeah. So like using clients, perhaps through this.

SPEAKER_00 21:42

Even in New York, they've come out with new regulations to block uh New York based businesses from using AI in health or in financial services. So it hasn't finalized the concerns of maybe the kind of thing you've got. But uh but I uh going back to your point, Duncan, I do I don't uh I'm not saying these companies with the big layoffs are saying, oh, it's because of AI. But they won't say it loudly anyway. I think that there could be other reasons, structural reasons, and things like that. Um, but what's been talked about a lot about in the press lately has been these huge inflated costs that even very profitable companies have to seek debt financing and that to pay for. So they don't say, oh, it's because of AI, but like some of it will be. And I think that's what regulators will really take a look at. Because no regulator wants to hear, oh, you're gonna scrap 30,000 jobs in my market because these people are my you know citizens, right? So I think that that's something they have to balance out.

SPEAKER_04 22:38

You sound way more optimistic than I expected. Like I'm I'm like going, uh-uh. Like I expect in a few years a graduate student studying psychology who knows how to use devices I can't even think of right now will do a better job than I will, even though I all have had 40 years experience on them.

SPEAKER_03 22:56

I think the the context remains important, however. I think there are things you know, I I come from that kind of mindset of augmenting through technology rather than play replacing. And let's not forget that you know we humans are in control of this. It's up to us to decide how how we want this to be used.

SPEAKER_04 23:15

Could you inspire me with hope? Because I'm like, augmenting, great. I'm thinking replacement, I just don't see it.

SPEAKER_03 23:21

Yeah, well, let's look at it sort of negatively for a second, rather. I actually I I have a private bet on with a friend of mine who's who's very senior in one of the technology companies, and it's a when, not an if bet. It's when this year will we see a business catastrophe caused by misuse of AI. Uh-huh. Now I'm calling August, and he's calling October.

SPEAKER_00 23:43

Oh, really? Yeah. So I'm calling last week.

SPEAKER_03 23:45

You're calling last week, yeah. You know what I'm talking about. I do know what you're talking about. Can you say?

SPEAKER_00 23:49

So there was a company, we don't have to mention the name. They were utilizing IT, the um code that was generated by AI, and it took their systems down for a period of hours that they estimate cost up to six billion dollars. Oh, yeah. You can't say the name, or I don't want to say the name because you know it's at the end of the day, you don't always know the insides of what happens at big companies. So, but I do think that those stories are going to be important, and you're absolutely right, Duncan. And you still have to have humans for sure. It's what you can do faster. And some of these agents, they're really interesting what they can do. They can analyze data so quickly from so many different sources that you could actually make yourself a you know professional investment manager uh and have quality reports in a very short period of time that normally you wouldn't even know how to do it, or you'd have to have an extensive education background and access to a lot of things like professional financial services. Uh so that a lot of things like that are happening.

SPEAKER_03 24:48

But um to come back a little bit to your question, Jonathan, and the and you'll be aware of this being a medical professional, you know, the the when when you're when you're training a young person up as they come into the in into your field, you you you have this scenario where you know the the student will go chasing zebras when actually we're just looking at the horses. Yeah. So people won will want to go to the edge cases without actually looking at can we stay within one standard distribution of the mean, please, because that's where it's going to be. Yeah. Now the machine doesn't know any of that, right? The technology can't differentiate the context between the edge cases and uh what's normalized. That's where that's where the human context and the human uh skill set and experience still comes to play. You can't really replace that right now. You can augment it and you can give prompts and things, but you can't really replace that context right now. And the reason why so large language models, which is the state of the art today, they're based on linear algebra. Uh and that is what it says it is, right? It's a mathematical construct that kind of runs in a straight line. And why you see the amount of investment that has it is having to go into the technology is that if we if we use the mathematics that describes the current state of the art continuously, then to make more of that and make it happen more quickly, you need to buy more technology in a linear fashion, which is why you see the big technology companies going, oh, we've we've got to buy all of the Nvidia chips, we've got to spend 50 billion. What linear algebra doesn't do is deal with immediate context and changing information that will then change your probability as things move forward. That's a different set of mathematics, that's Bayesian inference, which isn't the state of the art at the moment. Interestingly, all this mathematics is is is very, very old. You know, linear algebra was um Aristotelian from way back when, and Bayesian inference is from the 1700s. The mathematical models that that describe how all of this stuff works are not new things. And that's also something to be borne in mind, right? This is you know, the mathematical models here are designed by humans.

SPEAKER_01 27:03

But surely businesses, their goal is to eventually radically cut their people costs and replace people with AI and robotics. I mean, you could look at Amazon as an example, you know, with their distributions. At the moment, you've got people manning the distribution centres. That could be done by robotics at some point. You've got driverless delivery vans that eradicates all the drivers around the world. I mean, it's going to have a huge impact. There is going to be a glut of professions that are eradicated, and the companies are just going to become extremely wealthy from not having to pay salaries.

SPEAKER_00 27:39

But there's a competition element to that because if you look at the example you share with Amazon, Walmart does the same thing. And if Walmart didn't do it, they wouldn't be able to compete with Amazon. And I remember a few years ago when Walmart wanted to build their online business because they were trying to compete with Amazon because nobody wanted to leave their house. They were just ordering things on Amazon. So Walmart then was competing more hard and fighting. So they there's a lot of uh automation that's taking place in both those companies. Now the latest things that I've the news changes every day. So the latest things now is that um, you know, Jeff Basil is trying to raise a hundred billion dollars to start automating all kinds of factories, not just like his e-commerce businesses and that. So I think that that's the kind of thing that raises a bit of fear in people because if you start automating everything with robots, um, then you know you will think if you're especially if you're a manual labor. Um Elon Musk with his Optimus robots and things like that. So there's a lot of things to be fearful of.

SPEAKER_04 28:36

Optimus robots?

SPEAKER_00 28:37

Yeah, so you know, Tesla, they make Optimus robots. What's an Optimus robot? So Optimus robot is a robot made by Tesla that it's going to be like it's kind of a it's actually pretty amazing.

SPEAKER_01 28:47

It's very human-like, yeah.

SPEAKER_00 28:49

Um it's pretty amazing. And and he's expecting that the Tesla business will be bigger in Optimus than it will be making electronic electric vehicles into the future.

SPEAKER_01 28:59

You have to watch eggs. There's loads of videos of these robots kung fuing and all sorts of stuff.

SPEAKER_03 29:05

Yeah, the in I mean the interesting things in there, you know, I talked a little bit before about probabilistic versus deterministic. Now, the robots in those sorts of spaces are genuinely deterministic. That they are programmed to do a very specific set of things. So when you see a robot kung fuing, that's because somebody's programmed it to do that. If you went and hit it with a stick, it would fall over and really struggle with what to do next because it hasn't been programmed for that. Now, in a factory, you want that deterministic, you you want the robot to pick up the package, scan the barcode to know what the package is, and literally move it in a straight line from that box to that box. That that is a factory-level thing.

SPEAKER_02 29:41

Yeah.

SPEAKER_03 29:42

Coming out of the factory, what we've what we've discovered over the years, uh, since and we're we're probably seven or eight years into now the whole driverless vehicles thing. We've discovered actually that that's a really hard human problem to solve, and it's a really hard technology problem to solve. And probably the state and the state of the art in that space is the the US company Waymo, um, who we recently found out that actually they augment their driverless cars with a real-time call center in the Philippines, where they've got hundreds of trained people in the Philippines who can take control of those vehicles if they need to do in real time if the risk profile changes. Tesla still haven't released their fully driverless uh technology because the the regulators keep going to investigate and saying actually that's not that's not good enough to meet the requirement. So those you know, the the the responding to real-time changing context by technology is a really, really hard problem to solve. Uh and and it feels like a simple thing. If driving a car feels like a simple thing that maybe technology can solve, it's not that easy.

SPEAKER_01 30:54

Well, we're running out of time, guys. So my last question is, and we're gonna end on this will governments around the world have to step up and pay people's mortgages and put food on the table as and when jobs are eradicated?

SPEAKER_03 31:07

Uh like they always do, that's what progressive progressive taxation is for.

SPEAKER_00 31:13

Always has been. There's a lot of people that talk about UBI, universal basic income, which is you get paid to not work, um, including Elon Musk. And I think that you know, people realize that there's gonna be a concentration of power in those who own the compute and all the technology behind the AI evolution. Um, and you can't you don't want to have a revolt. I think one of your questions talked about revolt. I think that you don't want to have that, and you don't want to instill fear in people because I think AI can make a lot of things very efficient, you can improve on it, and that's why, you know, even when you look at cloud computing, you remember when it first came out and you know, people were concerned about data and where it's stored and all that kind of stuff. They're still concerned about that, but I think a lot of society has moved up the curve on that. They're now more comfortable using it and deploying it. And companies put a lot of focus and energy into working with regulators, working with the policymakers to ensure that you can leverage the best at a technology without worrying about, you know, cloud computing is not good if everyone has data localization policies as an example.

SPEAKER_01 32:19

So thank you guys. I'm sorry to cut it short, but we've got our part two coming up. So thank you for joining us today on Ella Podcasts. Tune in for part two on AI, where we're going to explore which jobs will survive and which will thrive. If you want to suggest a topic from our next episode, please join our Facebook group at Ella Podcasts and message us. Please subscribe, rate, and share this podcast. Sending you a big human hug.