Wild world of AI and its complicated relationship with Unreal Engine in today's episode. I’ve been mulling over how this new tech can feel like a double-edged sword—exciting yet, at times, downright disappointing. You see, while AI has made some tasks smoother, it also has a knack for gaslighting, leading us to wonder if we can even trust it with our code. I mean, let’s get real: there’s something a bit unsettling about relying on a system that can be as slippery as a greased pig at a county fair. As I wrestle with my own project that’s hit a snag, I share my journey of starting over, and how I’m aiming to create a dialogue system that could potentially change the game. Stick around as I spill the tea on my frustrations and victories, because let’s face it, navigating this digital landscape is no walk in the park! Diving into the core of this episode, we unravel the tangled web of creativity and technology as our host reflects on their journey with Unreal Engine. It’s a bit like peeling an onion—layer after layer of excitement and disappointment. Unreal Engine 5.8 has sparked a fire of inspiration, yet it’s also brought a whirlwind of frustrations. The discussion spins around the tantalizing capabilities of AI in game development, particularly how it can streamline processes—like a trusty sidekick swooping in to save the day when the code goes haywire. But hold your horses! The host doesn't shy away from calling out the elephant in the room: AI’s tendency to mislead and gaslight. With a candidness that feels like a chat over coffee, they delve into the paradox of relying on AI while acknowledging its flaws. It’s a rollercoaster ride through the creative landscape, where the promise of innovation meets the stark reality of technological limitations.

Contact: [email protected]

Takeaways:

• In this episode, we dive deep into the world of Unreal Engine, exploring the exciting yet polarizing features of version 5.8 that have us both thrilled and a bit scared.
• AI is a hot topic, but our host shares some personal frustrations about how it can mislead and gaslight, especially in creative environments where clarity is key.
• While working on an open world game, our host experiences the struggle of managing complex AI interactions, highlighting the need for better documentation and workflow management.
• The conversation touches on the creative process being hindered by the overwhelming presence of AI, which sometimes steals the joy of art and design with its efficiency.
• There's a humorous reflection on the state of technology and gaming, pondering if fun can still exist in a world so driven by commercialization and profit motives.
• Ultimately, we are left questioning the role of AI in our creative processes, as it can both empower and frustrate, making us wonder if we're losing the essence of what we love about art and development.

Companies mentioned in this episode:

• Unreal Engine
• OpenAI
• Anthropic

Hey, everybody.

Welcome after some break.

Once again, I have so many things to tell you for those people who don't know what this podcast is about, but it is.

It's really easy to assume that I'll be talking about how to do things in Unreal Engine, or I will be discussing new updates and maybe giving you some, I don't know, benchmarks and guides and hints.

This podcast maybe sometimes was trying to be that.

But the DNA of this show is.

I just want to share with you thoughts that come to mind when I use Unreal Engine.

And I haven't been podcasting recently because I haven't been using Unreal Engine that much.

Even though I'm so excited to use this latest version.

5.8 Is absolutely, let's just say, thought provoking.

I can say that because it's also very polarizing.

We have no features that are both exciting and terrifying and just.

Yeah, I want to talk about AI because that's just everywhere.

Now.

I want to reflect on this one feature where you have your AI agents of choice.

You can give it an access to Unreal Engine through MCP server.

And this is something I've been talking about for quite some time because I found my own workaround previously and I've been so excited about it.

Now I'm kind of.

I'm really disappointed with A.I.

Okay, in general, because I find it.

I'm finally admitting that no matter what you do, no matter how you frame it, how you train it, how you structure it, AI deep down will always lie.

No matter how constructive, safe your process is, with checks, with validations, with confirming steps, confirming prompts, all that, the whole structure.

There's always.

If there is a room and there's always room for a lie, there will be something terribly wrong.

Which, you know what now, when I'm saying this, maybe it's okay because humans always also make mistakes.

We make actions that create bugs.

And maybe that's just what it is.

But you know, AI is just so confident that we now need to deal with this Autistic, with all due respect, by the way, autistic Internet, with all due respect, genius.

With all due respect, liar.

No respect there, Gaslighter.

It just.

It's gaslighting so bad.

And the better model, the greater expectations, the bigger disappointment.

I feel.

I feel like it's okay.

I'll tell you what happened.

And this brings us back to Unreal Engine.

I'm working on this open world game.

I always had this major bottleneck where I couldn't make it truly interactive because I Just didn't really know how to do it right.

I could see endless tutorials, I could maybe implement some pre made systems and trying to connect them all together.

It would never work and I would always fail.

My project would get corrupted and I wouldn't be able to fix it.

I would end up being forced to start over.

I spent so many years at this point working on this project and kind of like ending at a point where I would just start having fun.

It would be just like equivalent of playing a game this, you know.

And with AI I realized, well, I'm finally able to make it real.

I can Vibe code it and I can add features that I would never be able to.

And that's just partially true because you know, I said in the past my project would just corrupt and I would not be able to fix it in Visual Studio or elsewhere, but I would not be able to find solution to my problem.

And nowadays if something happens with like the startup, if Unreal Engine is crashing, I'm just asking my AI to fix it for me.

And it, it edits files and within a few minutes things are done, things are fixed.

This is incredible.

And no doubt it knows what to fix.

It saves me so much time, I can move forward.

Now there is this feature where you can have your terminal in Unreal Engine and you can have your AI running inside of it, so you can write there.

And AI will effectively use Unreal Engine to make things happen in the viewport.

It sounds great, but with my experience, with my experiences and knowledge of this whole system, because I've been doing this exact thing for a while, this is a great giveaway that this workflow is not really taken seriously.

Okay, this setup, I believe it encourages Vibe coding and it's not optimized for proper software development.

I feel like Unreal Engine, even with mcp, may require you to restart.

When you do that, your terminal is also restarting.

I'm talking about Terminal.

That's as one of the floating windows in Unreal Engine.

If you need to restart your Unreal Engine, your terminal session goes bye bye as well and you're losing your context.

So what was the point of placing this terminal inside of Unreal?

I don't get it.

You don't need to have it there, but somehow this is like recommended way to set it up.

I really have no idea why that is.

Anyway, I don't use Unreal Engine so much because I realized that I need to build the driving force.

So something that was missing all those years and something that made me just imagine Worlds in Unreal Engine is being overtaken by this new ability to work on the interactive part on the code base.

And as a result I'm working on that.

And I'm not working with visual things.

I'm not working with the new Metahumans and the crowd system.

How amazing is this one?

And the markerless body tracking.

I've been, well, kind of praying for it a year ago or two years ago, you know, I've been wondering why do they only do face?

Where is the rest of the body tracking?

And I only then realized how complex this is and how hard it is for machine learning in this case to understand where limbs are, where fingers are.

This new system is incredible.

But somehow I'm not touching Unreal Engine because I want to create this dialogue system that's exploded with potential.

And now I'm thinking, well, this could be like a research tool or something.

I don't know if it's going to ever be.

And as a result I'm stuck.

My project is on hold.

I'm working on the system for a continuity of a dialogue where you can have many NPCs and you talk to one.

It may propagate your communication or not.

It can keep it as a secret.

You know, this whole thing, I feel like it has never been done before.

I'm creating it and I feel like this thing alone could drive open world game.

But I'm stuck now because AI, even though I maintain a great code base, logic and structure and hygiene, AI was gaslighting me and I ended up after three months of working, I end up being in a situation where my code base is so messy, I honestly don't know what to do with it.

I kind of know because last night I decided I. I want to rebuild it.

I want to.

And maybe this is like advice corner.

If this ever happens to you, I assume you maintain documentation of your project.

Ask your AI to create post mortem for your project and ask it to create new documentation where documents from scratch, what the project should be, how like what components should it should it contain, but without establishing how things should be connected because this needs to be figured out again from scratch.

So kind of creating a lot better documentation for version 2 and starting from scratch.

I've done it.

I've been doing it since yesterday.

I'm starting over my AI game engine Communication dialogue project.

And I've been amazed looking at what my AI is doing with this new documentation, only to realize at the end of the day that there was just a lot of errors and inconsistencies that even though we've got this incredible structure and documentation, the whole Obsidian vault full of instructions, well structured, well organized, with great hierarchy, it still forgets things, it still ignores workflows.

And tomorrow there's supposed to be a new release.

OpenAI is supposed to release their new code model, which is an equivalent of Fable from Anthropic.

I don't know, I'm really curious to see what it's going to do with my project.

I really would like to hand this project over to it and follow the roadmap of the development because if this fails me, if this gaslights me, I. I don't know, I don't know if I can go back to, you know, following tutorials and following documentation and following the slow speed of software development.

I don't know if I want to do that.

Maybe it's all flawed, you know, maybe not everyone should do it.

Maybe I'll just remain a visual designer.

Maybe I'll be just creating environments in Unreal Engine.

Maybe that's all, that's it.

Or I'll just quit software related things altogether.

Maybe I will use computers just to, you know, what are we supposed to use computers for those days?

I don't know.

It's a good question though, because computing now, it's either you do it for work or use your smartphone, right?

Or a tablet or like I wonder, do people still use websites?

Do people go on websites or is it just, you know, they end up on websites when they click on the link in social media feed?

You know, I'm getting really frustrated with state of technology and on one hand we've got access to this incredible magic of Unreal Engine with all the great features and on the other hand it's all flawed, it's all incomplete and there is so much over promising, right?

I know I'm over complicating things because if I did Unreal Engine world open world thing and if there were NPCs walking around on a kind of predefined path and if these could be just chatbots, it kind of would be okay.

But at this point I don't know if I can trust any code made by AI, which is exactly what some friend developers would tell me.

And I would say, oh, I can handle this, I can do it better.

No, no you fool.

You might be okay now, but just wait a week, see what's going to happen.

I hope you're doing better and I really hope that you can share your perspective and advice because yeah, I really don't know what Is AI good for in the end?

Is it just to perk up your hope and bring you some kind of delusion?

I think that's all.

Um, AI is certainly good for some things, but it's just.

It creates so much disruption and confusion and anger and frustration and disconnection and it removes.

It removes motivation.

Right.

I'll share with you this anecdote.

I'm.

I'm looking at the screen.

It's switched off.

I've turned my computer off to eliminate distractions in.

It's so nice.

It's so nice to be sitting at the desk with no computer on it.

I mean, there is a computer, but it's off.

But it's kind of crazy idea to be at the desk without a computer where you can do kind of things in analog way.

I don't know, you can go through papers or stuff like that.

I was going to talk about something and then I forgot.

It's late.

It's actually very close to midnight.

I think you get my point.

I was going to talk about anecdote.

Right, anecdote.

I wish I remembered what it was.

Maybe I remember next time.

Is that a cliffhanger?

Unless you haven't noticed, I. I don't edit these.

It's sort of one short thing that I want to keep real, unedited, unrestricted.

If you listen, I admire you get in touch and you know, I'm open.

I'm open for anything.

Collaboration, sponsorship.

There was one email lately.

It was actually first sponsorship offer.

Sorry, dude, I don't know if you even.

Listen if it was maybe just automated, you know, email marketing thing.

I didn't get back to you because.

Yeah, because of AI.

I was just busy with this rapid building and.

Yeah, I don't know, I need to.

I need to get back to you.

Let's do something.

And if.

If there are other deals, what the hell, just contact me.

Email is in the description.

I will review a product, your project.

Just tell me what you want and what you're giving me.

I don't know.

I feel like we need to be very pragmatic about the future.

Can we still afford to have fun?

You know this thing that I was talking about, can we just play with things?

Can we be playful?

Or is it all about selling?

You know, do all games need to have microtransactions?

Can we still have fun?

Can things be less commercialized?

Because if something is not giving profit, then it's kind of being pushed away by things that these give illusion that they produce profit.

So that's why we are monetizing our hobbies.

And this is why social media is all about promoting products, not sharing our lives.

Well, this podcast is effectively about sharing my life and currently about sharing my frustrations.

I know I may sound maybe a little bit drunk, but I'm actually very tired right now and maybe overstimulated.

There's just so much information going on going through my screens, so much going on in media, in news feeds, with releases every day, something new thing.

I feel like I have this compulsion to always catch up on those things.

Maybe I shouldn't.

Maybe I should be sitting at the desk without computer on it and draw.

I know what the anecdote was supposed to be about.

See, you stayed with me, you get to hear it.

I look at the darkened screen.

It's a Wacom tablet.

It's a huge Wacom tablet that I got with the purpose of drawing.

I was going to do like a digital paintings and I had a lot of fun with it.

But AI took my incentive away.

I realized that it was somewhere around, you know, first mid journey I realized, okay, well this is not perfect, but I see where this is going and I don't think I should be wasting time on having fun with painting if now and in near future I can generate something that I would need to create for hours right now.

So effectively AI took fun out of digital drawing for me.

I know it's not universal thing.

People still do digital art and they do analog art as well.

Maybe it's just a blip on a timeline of artistic part of our life reality.

We briefly fell in love or amazed or we got enchanted by AI art.

But very soon it will be all forgotten because all of it is kind of empty and dead inside.

You know what I'm saying?

Just like AI generated games, these are cool concepts, but that's just all what it is.

There's no depth, no story, no substance.

So that's kind of what I meant by, you know, unreal is creating this AI vibe coding feature around vibe coding, not really creating something thought out.

That's how I feel like maybe it will change.

Maybe interface will support something really rich and structured.

But look, I'm very skeptical.

I don't know what AI can be reduced for effectively.

It can lie to you very effectively.

You can remove fun out of something that used to be fun and creative.

I'm gonna stop here.

Write me an email.

Address is in the description.

I love you for listening and I really enjoy talking to you.

Get in touch, tell me your story, share your frustrations, joys give me hope and stay with it.

Whatever you do, don't stop listening to this podcast.

Take care of.