Welcome and the week's AI sessions in context

Rob: Welcome to the National Virtual Expo. I'm your host for this session, Rob Malicki, joining you from Garigal land in Sydney. I'm excited to be welcoming back to the National Virtual Expo Aaron Dodd from the University of New England. Welcome, mate.

Guest: Thank you. Good to be here for our final session. This is our final session, and we've had some great conversations throughout the week. So if folks who are watching along have missed the earlier sessions, we've done three other sessions throughout the week. We did a great session on using AI and studying with AI to get better results, and we've talked about AI fundamentals and AI safety as well. All the details for those are down in the session description below.

Rob: And Aaron, as people are watching, they can ask questions in the chat as well. So if you're streaming along live, and if you're watching on the replay, join the community Discord or the Facebook group for parents. All of your questions will get answered, so make sure you get them in. But Aaron, I'm really excited about this conversation, because it's one of the topics I get asked most about when I'm talking to students about uni. We're talking about AI and the future of work.

Guest: We get to nerd out for half an hour, because this is such an interesting subject.

Is any degree or career truly safe from AI?

Rob: Maybe by way of just opening us up, the question I get asked probably the most is: what degrees and what careers are safe from AI? I might just put that to you, Aaron, as the Chief AI Officer at the University of New England.

Guest: I hate to be the bearer of bad tidings, but not a degree or career is entirely insulated from AI. If you consider robotics to also be part of the AI picture, and I think you have to, one of the big things frontier labs are doing right now is training AI to essentially drop into robots as brains. The brains are almost ready, and the bodies are only a few years away from being ready, maybe even less than a few years. So I think it's pretty safe to say that, whatever career you're planning and whatever degree you're thinking about, you need to start factoring AI into your considerations.

Rob: That's its nature, isn't it? I mean, we can now go to one of these tools, ChatGPT, Claude, and in seconds you can generate images, audio, video, music, text, of course. This is the obvious stuff. But then just as you said, they're 3D printing houses these days.

Guest: And also, you know, the idea that humans have empathy, that's important, and we can talk about that. But even in randomised controlled trials of online therapy, text chats, people can't tell the difference between whether they're talking to an AI or a human therapist. And there are studies showing that the AI consistently gets higher marks from patients than humans do. So even something that you would think of as a very human task is not necessarily safe from AI.

Why soft skills are becoming the most important workplace asset

Rob: I'm really glad you mentioned soft skills. Maybe before we talk about the impact on particular industries, careers, and degrees, can we look at what viewers should be thinking about in terms of the skills they're going to need for the future, regardless of what field of study or career they might go into?

Guest: This is one of the things that's changing most with this technology. Hard skills are technical skills and subject matter expertise: learning the composition of a cell, learning how chemical reactions occur, learning different laws, knowing how to edit a video. These are all technical, subject matter expert skills. Soft skills are the human side of the equation: empathy, listening, the ability to build consensus, motivating people to act, the ability to deal with difficult emotions in yourself and in others. You might have heard the term emotional quotient, or emotional intelligence. All of these sorts of skills are going to become increasingly important in the workplace.

Rob: Let's maybe break the discussion up into knowledge work, which is white-collar work, and then more traditional blue-collar work, and game out the implications of each.

Guest: Let's do that.

How AI is reshaping knowledge work and white-collar careers

Rob: So let's look at white-collar work and start to look at how AI might impact some of those careers.

Guest: What AI does is it creates artefacts. That could be a PowerPoint slide, an analysis of data, a visualisation of data, a business plan, a marketing plan, images or video to go in your marketing plan, an analysis of case law for a lawyer, or a reading of a patient's scan in a hospital that gives a diagnosis. These are all artefacts. The AI is rapidly developing the ability to generate any artefact, any cognitive output. And pretty much any job done on a computer is fair game for AI at some point.

Guest: Right now you can see how agents with the ability to move a mouse and keyboard have full control of computers. So where are we positioned when we're in a world where AI can do a lot of what we consider to be the work of knowledge workers? You could easily spend a day or two writing up a business plan for a project, and now it's done in an hour. You could easily spend a week analysing half a dozen academic papers and writing up a literature review. Now that's done in an hour and a half. So if what we've traditionally been paid to do in the workforce is create outputs, and the AI can do that, where does that lead us?

Guest: This is where the soft skills come in. I can whip up that business plan in an hour now instead of a day, which is awesome, because it's tedious work. But I still have to go to my colleagues and get their buy-in, get their feedback, incorporate their feedback, and then go to the other business units in the enterprise. Maybe I've done the business plan in marketing, but I have to go to design, manufacturing, and then finance to get the funding approved. I still have to be a storyteller, connect with people, enthuse people, and motivate people. So the AI takes away a lot of the cognitive work, but it doesn't take away the emotional cognitive work: the storytelling, the connection, the human side of the job.

Guest: And another thing mixed in there is that when the AI spits out its outputs, we have to evaluate them. As we talked about in a previous session, you've always got to verify and you can't trust everything it does. So a lot of your work becomes evaluating the AI's outputs, iterating, improving it, sending it back to the AI for another pass. And then you go: okay, so if my job is evaluation, iteration, and then the human work of deploying and executing, this applies in any industry for knowledge work.

Guest: Then this gets into something called taste. How do you know that's a good song and that other song has been done a thousand times before and isn't original? That's taste. How do you know that a graphic design for your marketing campaign is really on point and well-targeted towards your audience, or whether it's cheesy and inappropriate? That's taste. So all of a sudden things like taste become important. You've got taste, you've got evaluation and iteration.

Guest: And the last thing, I've seen this already where we've rolled AI agents out to university staff: you spend a lot of your time sitting around with a group of people, all telling an agent to go and do something, and then you're watching it come back and you say, oh, it's not quite right. And then you turn to someone and say, what do you think of this, and they say, oh, I'd ask it to do that instead and see what happens. It's a looser, more creative, more playful way of working, because you're not stuck in the trenches writing the business report, which everyone finds boring. I think the promise is a more playful and more enjoyable work environment.

Guest: And when you've got multiple agents, all of a sudden you're a manager. We're all sort of managing teams of agents to do work. That's a broad overview of where we see knowledge work heading. And I should say: no one really knows for sure, but this is our best educated guess.

Entrepreneurial skills and the shrinking advantage of large organisations

Rob: I completely agree with that. Some of the skills you're talking about there are entrepreneurial skills. And one of the things I've been saying a lot recently is: in a world where everything can be generated, images, text, video, music, the whole lot, ideas are a commodity now. You can ask Claude for 500 ideas and it'll give you 500 ideas in about 30 seconds flat. So what becomes the core skill of executing well? It's the ability to take those artefacts, as you say, and persuade another human being to try things, to fail, to be resilient, to realise you just fell on your face, get back up, dust yourself off, and try again. These are the core skills of being an entrepreneur. And an entrepreneur isn't necessarily Elon Musk or Mark Zuckerberg or Melanie Perkins. Statistically speaking, most entrepreneurs are small business owners running their own business, or someone working at a big corporation who launches a new project or product. People who have that entrepreneurial mindset are the ones who are going to thrive the most.

Guest: One hundred per cent. We all have to be leaders now. If you're planning on spending your life as a cubicle warrior, just being told what to do, you're not going to thrive. You've got to learn how to lead, because leaders know how to prioritise. You can get 500 ideas from Claude, but you've got to pick the good one, which comes back to taste. And then you've got to have the gumption to say, let's go this way, let's execute on this idea.

Guest: Entrepreneurship is really interesting right now, because up until recently in the history of capitalism, if you wanted to do something significant, it required a lot of people and a lot of resources. Designing a product, building a product, shipping a product, marketing a product. All of those roles took a lot of resources, and size was the biggest advantage. Size was its own moat, which, if people aren't familiar with that term, refers to what you've got around your product that prevents competitors from just copying you. But now, teams of one or two people can do what a team of 100 could do five years ago. All of a sudden, size loses its advantage. It still has all the downsides: committees, slow-moving, conservative, risk-averse. So size has flipped from being a moat to being a detriment, an inhibitor.

Guest: There's a lot of concern about AI taking jobs. I think it's also going to set up a whole realm of human flourishing, because people will be much freer to pursue things they're passionate about without needing the permission of some big company. As long as you embrace this technology and try to learn it and work with it, you can really reap the rewards.

How UNE is teaching students to build with AI, not just consume it

Rob: Maybe just a little side note there, Aaron. A couple of months ago I visited UNE and I was really impressed with what you were doing in terms of putting this technology into the hands of all kinds of students. Obviously a lot of students are using AI to help with studying, but what I love is how UNE is proactively trying to teach all kinds of students how to build AI agents, even if it doesn't intuitively fit into their degree area. Can you talk a little to that?

Guest: Great question. We don't just turn on a chat window and point students to it, because we don't want our students to be passive consumers of AI. We want them to be makers and builders. For that reason, we've given them a fully fledged online suite. We give students a choice of models from all the labs: you can use an Anthropic model, an OpenAI model, a Gemini model from Google, or one of the open-source models. Different models are good at different things. So we're teaching our students: this is a data analysis task with a huge data set, I should use Gemini for that because of its massive context window. This is a more creative task that involves something closer to emotional intelligence, so I'm going to use Claude for this. We're teaching our students model literacy. That's the first big thing.

Guest: Then we're giving them access to a whole suite of MCP tools that connect agents to the outside world, so they can escape the computer, get on the internet, collect data. When one of these agents goes to search, it hits 100 to 150 different links, rates them all, and returns the results. That's better than a human doing it manually. Plus it has memory, so it knows what you're interested in and what you're looking for. When you combine model choice with tool choice, students can go wild creating these agents. They can create agents to help them study, do research, and build apps and tools. The point is: get your hands on the tools to experiment, learn, and grow.

Rob: And would you say that goes for any type of student? A student going into nursing, or something which might be seen as relatively safe from this technology?

Guest: One hundred per cent, and that's an important point. Not everyone wants to be an entrepreneur, and not everyone's cut out for that life. But there is a clear path forward for people who want to work as doctors, nurses, psychologists, lawyers, and that is the role of subject matter expertise. Right now, as we were talking about earlier, the role of evaluation is critical. The AI gives you an output. How do you know it's great? That depends on whether you're a subject matter expert. If you're looking at a scan and the AI tells you this person has cancer, you need to be a subject matter expert to check the AI's work. If you're a marketer and the AI recommends a channel strategy to promote your new wellness product, you need subject matter expertise to know whether that channel is actually appropriate for your audience. So don't ever think subject matter expertise is no longer required.

Guest: If you're passionate about medicine, nursing, or teaching, an AI can be a great tutor and help students, as we discussed in a previous session. But most learning is still social and contextual. Teachers need subject matter expertise, the ability to teach socially, and the ability to contextualise. There's a whole world of work for the entrepreneurs, but for people who are still passionate about becoming subject matter experts in their chosen field, AI is just a level-up. It gives you superpowers, but you're still there, still required.

Embodied intelligence and relational intelligence as durable human advantages

Rob: I'd like to come back to something you mentioned earlier: embodied intelligence and relational intelligence.

Guest: Embodied intelligence is the intelligence we have from living in a body. Our intelligence is embodied, and we are constantly getting our five senses stimulated. We're hearing things, smelling things, tasting things, touching things, feeling pressure. We live in a physical reality, and that shapes our intelligence and it shapes our taste. AI will feel, sense, create, and design things it can never do, because it is not embodied and it will never fully get what it is to be human until it is embodied. Our embodied intelligence applies in every career. If you're a lawyer dealing with someone who's been falsely arrested, or a nurse dealing with someone in intensive care, you bring your life experience, your humanity, and your embodied intelligence to your work.

Guest: The other form of intelligence I wanted to touch on is relational intelligence: the intelligence we develop by being in relation to each other. A lot of physicists at the quantum level will tell you everything is relational. If you look at what it is to be human in the workplace, it's all about the relationships you have with the people around you. That is the same in the school playground. Our relational intelligence is getting a massive workout all day, every day. The AI can't replicate that. It might give you some advice as a therapist because it's read everything ever written about therapy, but it has never had to live through a relational experience. So embodied intelligence and relational intelligence are both incredibly important in the workplace, applicable to almost every job you could imagine.

Guest: The AI is going to do the artefacts. Don't stress about that. Just embrace it. Run toward the challenge: that's the best piece of advice I could give any student. Don't be afraid to change. As long as you're relying on your strengths and being brave and curious, you'll find a way. We're humans and they are not. The human aspect, the humanising aspect of everything we do, is the most important element of it.

What AI and robotics mean for blue-collar trades

Rob: Last question for you, Aaron, and I'm conscious we're getting close to time. Right back at the top we said we'd talk about white-collar and then blue-collar work. Can we touch very quickly on blue-collar jobs: builders, trades, things like that, and how this technology might impact those fields in future?

Guest: It's a little bit further out, so we're really guessing at this point. But my guess is it'll follow a similar pattern to what we're seeing play out right now with knowledge work. In the same way the AI is doing the artefacts of knowledge work now, robots will come along and do a lot of the actual physical product of blue-collar work. There'll be a multi-purpose humanoid robot that can unblock a toilet, load a dishwasher, and chop carrots. But knowing what the problem is, knowing why that pipe over there is bubbling, or architecting and designing the wiring plan as an electrician, that still requires a human. The human aspects, the planning, the prioritisation, the evaluation, the taste, the judgment, and the embodied intelligence, are really important when you're working with tools. You might know there's no way that wrench is going to fit in that tight little space, and a robot won't work that out.

Guest: I think you'll find that AI will speed things up, make them more efficient, and give tradies superpowers too, just like it does for knowledge workers. But the human aspects of planning, prioritisation, evaluation, and embodied intelligence will remain. I'd guess we'll see a similar pattern to what's happening in knowledge work, but maybe five to seven years later. I could be totally wrong on all of that, but that's my best guess.

Rob: Aaron Dodd from the University of New England, Chief AI Officer, thank you so much for joining me throughout the week at the National Virtual Expo. On a personal note, I have loved having these conversations with you, and I think we've been really fortunate to have your expertise to share with the audience. Thank you very much for joining me.

Guest: Thanks so much, Rob. This is an amazing event and I'm super happy to be part of it. Thanks for having me.