Stories Behind 10 Years of AI and Games
Stories Behind 10 Years of AI and Games
Sometimes videos come to be in weird and mysterious ways...
AI and Games is a YouTube series made possible thanks to crowdfunding on Patreon as well as right here with paid subscriptions on Substack.
Support the show to have your name in video credits, contribute to future episode topics, watch content in early access and receive exclusive supporters-only content
If you'd like to work with us on your own games projects, please check out our professional services provided at AI and Games.
Looking back on over 10 years of AI and Games, over 100 videos have been published on the YouTube channel, with a variety of topics covered. Some of those videos have proven very popular, some not so much. But which ones do I personally have some attachment to?
For this episode I took on a question submitted by the community of what are my favourite episodes. But I realise this risks me being a tad self-aggrandising, which very much ruffles against my British sensibilities. So I decided rather than focus on my opinions on which of my videos are great, I want to focus on videos that mean a lot to me personally, that have special stories behind their creation, and otherwise just hold a special place in my deep-fried heart.
It's worth saying now that none of these are in any particular order, and I love them all equally. And so with that, let's get started!
Some Personal Favourites
Before I get into specific episodes which have particular stories behind them, I will mention specific episodes that I am quite fond of.
These episodes are about games that I really enjoy playing. Contrary to popular belief, I don't actually enjoy all the games I cover on the series. But hey, I'm a professional: I can recognise the technical contributions and educational value of a game even if I don't enjoy it. To quote Anita Sarkeesian: "we must be critical of the art we love". But I'd like to add, "we must also recognise the contributions of art we dislike". Sure, it's not as pithy a quote, but it serves my point...
Also it's worth giving a special shout-out to Derek Fagan, an AI programmer at Creative Assembly, who helped out with the Halo Wars 2. That episode really came together thanks to a little bit of extra support from the studio (and on that note thank you to Lucy Boxall the communications manager who has tolerated many an email of mine over the years).
Sadly, with exception of DOOM and Halo Wars 2, almost all of them never did well in terms of views. But hey, who cares, I like them, I'm proud of them and I think they're interesting topics. Go watch'em if you haven't already.
The Troublesome Ones
Plus, a special mention to episodes that I am particularly pleased with how they turned out considering the challenges I faced in putting them together.
Each of these were a mountain of work to put together, with most of them taking weeks to research and write. Writing an episode of the AI and Games case studies is often mixture of two challenges: finding the information I need, and then digesting it sufficiently such that it turns into a cohesive story or narrative. But critically, I need to be able to back up anything I say. If I'm not confident that what I have written is the actual truth, then I am not willing to proceed with the topic until I find sources to back it up.
On that note, there are many episodes that are half-written and never completed, because they lack that one thing that really brings it all together. In fact the video on DOOM 2016 is a great example of this, given there were some GDC talks on parts of its inner workings in areas of combat design and animation, but never the guts of the main AI systems themselves. It sat half-written for about a year. It relied on a developer of the game coming forward and explaining it to me in detail. And in fact the reason that developer came forward was because of the Design Dive episode on DOOM I had released beforehand. That developer shall remain anonymous, given they asked for that to be the case, but of course I then had to do my due diligence in ratifying this person was who they said they were.
Conversely, I had a different situation with Hitman: I was working on this episode and - after announcing on Twitter that I was recording it - I received a DM from the communications team at IO Interactive offering to help. They actually watched the original version of that episode, and actually got the AI team to watch it. They then sent me some feedback to clarify some parts that I didn't know about, and also point out some mistakes that I should fix. All of that feedback was then taken on board, and made that episode one of my more popular ones. Thanks once again to the team at IO for helping me out.
AI and Games #29: Spec Ops: The Line
Spec Ops: The Line is a fantastic third person shooter, that is also one of the few games out there that truly tries to tackle with the themes that surround war and conflict. Episode 29 of AI and Games, which digs into its inner workings, is one of my favourite episodes not just because we got to cover what I think is an important game of its era, but it's also the first episode that really came to life courtesy of a guest interview, in this instance with Jörg Reisig, who was an AI programmer at Yager during the games development.
I met Jörg in person at event called Game AI North which ran in Copenhagen in October of 2017. I was actually invited to come and speak on a panel about designing great AI for games. It was in a bit of a weird time for myself professionally, where I was trying to get out of my academic career as I was deeply unhappy with it all. I had dropped to working one day a week at the university, and was doing contract work where I could to try and start building a portfolio for myself in the industry. In fact, I could barely afford the trip to Copenhagen for the event, and stayed in an AirBnB of a local family who lived in an apartment literally across the street from the venue.
On that note, a shout out to Richard Kogelnig who invited to the panel, and to Emil Johansen and the rest of the crew who ran Game AI North, given it had a huge impact on AI and Games. My videos on Halo Wars 2, Ghost Recon: Wildlands and Horizon Zero Dawn came about courtesy of the talks delivered at that event. Plus of course Spec Ops, and Sea of Thieves - which I'll come back to in a minute.
My time at Game AI North is special not just because of the videos it created, but also it was one of the first times I was involved in one of these events because people were starting to take my YouTube work seriously. I met attendees who knew who I was and watched my videos, and were very friendly and keen to chat with me. But then of course, there was the panel. Despite my recent positive interactions with developers, I still felt like a bit of a fraud at that panel, given I was sat next to Jörg on one side, and then Alessia Laidacker, who was previously on the AI Summit advisory panel at GDC, and had been one of the AI leads in the early days of Assassin's Creed. Nonetheless, both were super nice and I had a great time on the panel.
It was only afterwards, as many of the attendees hung out at a nearby bar, the Jörg and I got to chatting more, during which we talked about Spec Ops quite a bit, and he offered to be interviewed for a future video. The final episode emerged around 9 months later, and it's still one I'm very proud of.
AI and Games #44: Goldeneye 007
So the AI of Goldeneye is one of my favourite episodes: it's a popular game from my childhood - all of my friends played it in high school. But I wanted to talk about this episode because it's not just fun to talk about how it came together, but it's an video that subsequently led to several other projects I worked on outside of YouTube.
So yeah, first of all, the episode. This actually came about courtesy of a friend, Dan Scales - who is not only a former student of mine, but I subsequently worked with him for a time as a contract programmer. Dan was, at the time, an organiser for the Norwich Games Festival, an event where I used to run educational workshops for school kids with Shaun Spaulding (who you might know on YouTube from their work making GameMaker tutorials). Plus I used to showcase my own stuff like my game Sure Footing, and I got to hang out with the likes of Mel who was recently on Branching Factor, plus that's where I got to know George and Quang - my podcast co-hosts - a lot better. One of the big speakers that Dan managed to get in to talk at the event, was none other than Dr David Doak who was a developer on Goldeneye.
In fact David had already presented a really nice post-mortem on the game at NGF, and I thought wouldn't it be great to chat to him about the AI stuffs. But I was conscious about cold-calling him. So, I asked Dan to introduce us, and David was really friendly and was kind enough to give a good couple of hours of his time. Plus it was a really fun conversation that led to my making the video. A video that subsequently did really well, and I love being able to share it around.
But it's worth highlighting the extra stuff that kicked off as a result of it. So the following year - the episode came out in 2019, so this is early 2020 right before the COVID lockdowns - I am invited to be interviewed for a documentary about the making of Goldeneye 007, which would later be called GoldenEra. So I actually have a single credit on IMDB just for being part of this documentary which is kind of hilarious, and it's also a really good doc and worth watching. They get a lot of the dev team back, as well as notable journalists from across the industry. Strangely I was actually rather excited to be in the same video as Giant Bomb's Dan Rykert. I'm sure Jeff Bakalar will immediately think less of me as a result, but anyway...
And this led to another project... so the documentary was directed by Drew Roller, who is based out in Australia. In fact, when he interviewed me for GoldenEra he'd just flown into the country, and was doing a whistlestop tour of the UK to grab some final interviews with a few developers. Fast forward a year or so later, Drew is doing some work over at Digital Lode, a games studio that specialises in VR games. And it was Drew who recommended me to Lode’s studio director Mike Bell, and this led to me working for around a year on Espire 2 which is available on the Quest 2. So yeah, one YouTube episode let to a documentary and working on my first VR game. Not bad! On that note, my love to Mike, Sheddy, Josh, Fewey, Anniemay and the rest of the Espire 2 team. Was great working with you despite never getting the chance to meet about 99% of the studio in person.
One closing point about GoldenEye, when I recently announced my 10 Years Later episode. where I reflected on my experience, David wrote some very kind words about my work on Twitter. And I'm not gonna lie, I usually try to separate church and state around here: my love for games is separated from my work on exploring them, but having the Dr Doak write that did bring a tear to my eye. It was very kind of him.
AI and Games #40-43: Sea of Thieves
Given Spec Ops was achieved courtesy of bumping into a dev at Game AI North, I couldn't possibly forget to talk about episodes 40-43, my deep dive on the AI of Sea of Thieves.
So while my first evening in Copenhagen I had sat chatting with Jörg, that same week - I think it was the same evening actually - I had found myself at a table with Andy Bastable and Rob Masella, two developers who at that time were based at Rare, working on Sea of Thieves, which was only a five months away from release at that time.
We'd been chatting quite happily, but a couple of times our conversation was put on pause as a couple of developers had came over to say hi to me, and thanked me for my work on YouTube - which is of course, super lovely and very humbling. Bearing in mind, this is only about 3.5 years after the channel has started. So I still felt like a horrendous fraud (well, not much has changed there). It was after this happened twice that Andy asked me what that was all about, and I explained that people know me as a YouTuber (believe me this doesn't get any easier to explain the older I get). And then we got back to our regular chat.
The following day, Andy tells me that he had sat and watched some of my videos that night back at his hotel, and enjoyed them. He'd gave me his business card the night before and said 'I'd love to chat to about the stuff we're doing in Sea of Thieves'. He was particularly fond of the work his team had done the sharks as I recall. Of course, there was no point reaching out now, the game wasn't even out yet, but the idea sat in the back of my head for a time.
And then one morning, I was going through my business cards after a recent event, and I stumbled upon Andy's card again. Sea of Thieves had been out for around six months at this point, and I thought, why not. So I dropped him an email, asking if he'd be up for it. And within around two weeks of emails back and forth, we'd agreed on date where I was to be invited to the studio to interview some of the team.
So myself and my buddy Matt Syrett went over to Rare for the day - we booked the day off work, and Matt was both our driver and my cameraman given at that time I'd never done anything like this before. So we romp on up to the small village of Twycross, where Rare is based. It wasn't my first time visiting the studio - I'd previously had students do internships there and visited them. So I knew the location, but it was the first time I really go to see a lot of the day-to-day as the team were working on the game. Though we were only allowed on one side of the building, given early phase work on what I presume was Everwild was happening in the other half.
We spent the whole day at the studio, as we interviewed several members of team. Had lunch at the canteen, and generally just went for a wander to have a look at how the game actually is being built. I've never had that level of access to a studios work since, given the subsequent collaborations with Ubisoft all happened remotely during COVID. So hey, if you're a studio who wants to invite me down to meet your team and dig into how your game works, drop me a line!
The funniest part of this whole thing was also at the time the most frustrating. In order to get the videos out, we needed the communications team at Rare to approve them. This is of course a standard thing: they'd invited me to the studio, I'd interviewed the developers, and I was under an NDA (non-disclosure agreement). They had to agree that what was shared was an acceptable use of the NDA. Making sure that the videos were on point, accurate and not defaming or slamming the studio. Like I say, this is quite a standard process, and not an issue. It's part of making these videos a reality.
But the problem was, we'd never really briefed the communications team on what we would talk about, and what was or wasn't acceptable content to put in the video. What I mean by that is, we could perhaps have a conversation about a core system in the game, but it would be agreed that it wouldn't appear in the video. This is sometimes useful given it helps me understand some production realities about the game, even if I don't share them with you directly - but it might influence how I script the episode or explain a key aspect of the material. So the team at Rare finally get to see the very first draft and they were mortified! They realised just how much the devs told had us so about the underlying systems at work in Sea of Thieves, and they were worred that if it got out, the player community would then exploit that knowledge for their own benefit - and I both agree with some aspects of that decision (which I had already cut out) while also disagreeing with other stuff that I felt was harmless. As a result, there was a good 20 minutes worth of material from those interviews we had to cut out of the episodes..
Funnily enough, if you read the comments of the Sea of Thieves videos, people state that they believe certain systems or behaviours only trigger in certain contexts. Now I know which of those predictions and assertions is actually true - or was true at that time, who knows, the game has changed a lot in the years since. But I'd argue we cut around 15-20 minutes worth of content that the studio didn't want acknowledged.
Nonetheless, I'm still super happy with those videos. I'm very grateful to Andy, Rob and the rest of the team for taking the time to chat with us. Plus of course Sarah Noonan, Stuart Holland, Tristan Bell and Chantelle Porritt. Grateful that Matt was happy to help out. Plus I learned a lot from that experience about how to conduct these interviews when working with a games studio. It had a huge influence on my subsequent projects, especially the Division 2 and Watch Dogs: Legion episodes. They went really smoothly by comparison and I'm still very proud of how they turned out.
AI and Games #15: Alien: Isolation
Now, I know what you're thinking, of course he's going to put one of the Alien: Isolation videos in there. Those are the videos that have the most views, they're the videos everyone knows the channel for.
Well yes, but I have to say episode 15, the first video about Alien: Isolation is really not one of my favourites. The audio is busted, the editing is bad, and I really feel like my revisiting the topic in episode 50 was a much better video. But, it's the story behind the first Alien video that makes it one of my favourite episodes and it's also the cautionary tale that it represents, that makes it so important to me.
The original Alien video is inspired by a talk delivered by Andy Bray, who was an AI lead on Alien: Isolation. The talk was delivered at the nucl.ai conference in 2016 which was organised by Alex and Petra Champandard. I was a track chair for nucl.ai, a volunteer role that meant I helped find speakers to get them to the conference, and then support them in getting their talks ready. This is very similar to what I do now over at the AI Summit for GDC. Now I was super excited for this talk, given I had previously reached out to Creative Assembly to come and give a talk at nucl.ai in 2015. But they never responded. So receiving a pitch for a 2016 talk, even though it wasn't in my track, was super exciting.
I went along, I watched the talk, and I thoroughly enjoyed it. I scribbled down pages of notes during that talk thinking 'this will make a great video'. But I had a few more questions, and so I ended up having a chat with Andy during the coffee break later that day, plus also got to meet a new member of their AI team, Derek Fagan, whose I mentioned already given he helped out with my video on Halo Wars 2. Now the key thing about the coffee discussion is that there are facts in my original Alien video that only exist because they were shared with me over a cup of joe.
I got back home not long after and then quickly put together the video. Excited that not only did I find out how it worked, but also that I had some juicy extra tidbits of knowledge from the developers themselves. I put it all together and launched the video in early August of 2016 and... nothing...
To this day one of the biggest challenges is getting people to watch your content. My views are often at the mercy of the interest in the topic, given if they're not that interested in the game I'm exploring - or if someone else has explored it before I have - then that leads to a lack of interest. Interestingly my videos on first party PlayStation exclusives are almost always a lot less popular than you would think. Plus, I don't really follow the big games people are talking about. A lot of content creators make videos about popular games, or games that are in the conversation. I explore interesting AI applications and challenges, and then tell stories about them. Sometimes it's right after a game comes out, but typically it's years later. So I don't really get the chance to feed on the gaming zeitgeist in a way that a lot of other creators do. So yeah, even now 10 years and more than 200K subscribers later, most of my videos don't get a huge amount of views.
But surely Alien: Isolation was going to catch eyeballs? Nope. In fact the videos before and after it, about Transformers: Fall of Cybertron, and Far Cry 4 respectively performed better in their first 2-3 months than Alien did. I was kind of bummed out about it, given I thought surely people would want to know about how the xenomorph works.
Fast forward to January of 2017, and then something weird happened. I woke up to find my Alien video had received a significant boost of around 1000 more views than the day before. In fact that was around 15% increase in the total views the video had ever had. And then it kept growing, and growing, and growing. Within weeks, the video had received over quarter of a million views, and my subscriber count had pretty much tripled. At the time of writing, that video is now sitting on around 1.6 million views.
My thoughts on it are fairly mixed. As I say, I don't think it's one of my better videos, and if anything I revisited the topic in episode 50 just so I can could actually be happy with the quality of what I'd done. But I can't discount the fact that my original Alien video is the reason my channel still exists, and has since became modestly successful, and people still to this day watch that video every day - on average that video still gets around 300 views a day.
It was rather exciting, and intoxicating, to see the numbers on that video go up, and up, and up. But it's also a cautionary tale, because no other video has ever really had that level of success on my channel. In fact, my second most popular video, is my *other* Alien: Isolation video. And you run risk of damaging your mental health as you wonder whether you can make lightning strike twice, or whether there really is an audience that's interested in hearing you talk about things that *aren't* Alien: Isolation.
It's something any content creator who has had a modicum of success has experienced, as they try to reconcile their creative interests and outputs with an audience that is held at arms length from them courtesy of the algorithm. It's perhaps unsurprising that so many popular creators are winding down now after many years, given the challenges of chasing those number does more harm than good.
And so episode 15 is a favourite of mine because it reminds just how fragile and fickle this entire thing is. Without that video, I wouldn't have benefited from its success, and it enabled for my online audience to grow drastically. But it's also a solemn reminder that success of that magnitude is an outlier, and seldom straight forward - I mean look at the rest of my video view counts. It is not in my best interests to try and chase it, nor is there any guarantee it will ever happen again.
The F.E.A.R. Retrospective
Now this episode only just aired, but this was an incredibly important episode for me to make, and one that I still pinch myself thinking that I got to make it happen.
This episode is a deep dive into the making of FEAR, but I focussed it around multiple interviews I had with the games AI programmer Jeff Orkin. When you've spent a lot of time reading and learning about AI for games as much as I have, you begin to recognise names that have importance in the field. Jeff Orkin is one of those names. And I thought wouldn't it be great, if we went back to FEAR one last time. To look back on the very first episode of AI and Games, and do it with the guy who made it. That would just be fantastic.
Now critically, Jeff and I had never met, and never spoken outside of a brief exchange on Twitter some years ago. And so I whipped up this email and asked if he would be interested. I won't lie, I was nervous about this. But Jeff was super excited at the prospect, and was happy to do it. The only snag was we delayed it slightly, given he was going through a lot of work to finalise a round of funding for his new start-up BitPart.ai, which I'm happy to say has all worked out and they're up and running successfully. Hence that's why we launched it in April, rather than in March, and instead I brought up my 10 Years Later reflections episode to fill that all important March 2nd date.
This episode was incredibly satisfying to put together, especially as someone for whom FEAR was highly influential in my own career - a point I make at the end of that video. I was reading Jeff's paper from his GDC talk while I was in grad school, I was working on my own research on planning for games. It was great to not only hear Jeff explain this all from his perspective, but also to discover that we have rather similar perspectives and views on a lot of aspects of AI for games.
We still have our follow-up episode on Jeff's grad school career and work at BitPart coming up, but it has been very rewarding to put this episode together, and it's genuinely one that I never thought I'd ever be able to make. Funny how things change over the course of a decade huh?
AI and Games #75 - ???
Okay, so this last one might sound like a real cop out but the NEXT episode of the main AI and Games series, which isn't out yet, is without a doubt one of my favourite videos I've ever made. But, let me just explain why this is an important one.
First of all, it's an important game, not just to me but the gaming community as a whole. And while it wasn't a game I needed to know the ins and outs of versus some others, there was sufficient mystery behind it, that when the opportunity arose, that I decided to explore it.
The second, is that trying to piece it all together has taken a lot of time. A lot of my episodes are - for me - like puzzles, I find a piece of information here, a little bit over there. A fact thrown out in a press release, a comment made by a developer in an interview, some little bit of code or inner detailing a modder has been able to surface. But quite often the hard part, is being able to see the whole picture those pieces comprise. We discussed this already with my video on DOOM (2016): there are many topics I have pages of information on, but I lack that one critical component that can make it all come together. Even if I can't solve the entire puzzle, having that one piece can help me see the end result, and what parts are missing. It's what makes every episode of AI and Games so much work, because I'm trying to figure out how it all works, and then I need to understand it well enough that I can then explain it to you sufficiently in a video.
Sometimes we have a dev talk or similar presentation that fills in all of those gaps, and then sometimes I also get a chance to chat with the developers themselves - as discussed already. But other times I just don't get that kind of access. Contrary to the stories I've told throughout this video, access to studios is still something I struggle with regularly. For example, in the past year I've sent out dozens of press requests to studios across the industry and they either get rejected or more often than not, simply ignored. And that's fine, I understand the business reality of that. Like I'm just an idiot on the internet, with a modest but not huge fanbase, so why would they respond?
So sometimes I need to find another way to piece it together. And that can be a very long process. Like I say, many episodes are in a half-written state, and they stay like that often for years at a time. Until something comes along and it all clicks into place, and a couple of months ago, it finally took shape.
This is a very roundabout way of saying that AI and Games #75 has been a labour of love for quite some time now. A side project I've been chewing on for over a year, and one that I'm very excited to finally see come to pass. At the time of writing, that episode is largely scripted and I'm now recording it, and I can't wait to share it with you all very soon.
Closing
So there we have it, some stories behind many of the videos I've made thus far. There's actually a lot more stories we don't have time to cover, but we’ll save them for another time perhaps!
But I figured some closing words of wisdom for anyone out there who is keen to get into games and make their way in the industry. You'll notice that every one of the stories I shared today all hinge on my building connections, and more importantly, friendships, with people across the industry. It speaks to the value of networking, be it online or in person and critically approaching each interaction with kindness rather than with an agenda. Being an asshole won't get you anywhere in the long run. Sure, I've met a few in my time, but it's more important to engage with people with a positive attitude and leave a positive impression while you're at it.
Anyway, I hope you've enjoyed my ramblings and found it interesting. Of course, if you have a favourite episode, then do let me know down in the comments. It's always interesting to hear your perspectives, given I'm so close to all of this stuff it's hard to tell.