Upcoming games may be based more on AI interactions within their game worlds rather than just having upgraded graphics, sound and new storylines presented and planned ahead of time by dev teams. Events within games are already becoming a lot less scripted than ever before and becoming more dynamic or changing based on player choice and play style.
Games like Disco Elysium are also showing the power of choice within all aspects of gameplay and how choice is shaping the storyline of a game more than scripted events have always been doing. The player is having an ever greater role in shaping the game’s future events within the game world and influencing the outcome of events as they occur in real time. This will continue to be so, but also with AI being implemented to add even more dynamic behaviour into the equation and more freedom for gamers and non-playable characters (NPCs) they encounter.
I previously wrote an article how the next Elder Scrolls should focus on AI interaction and not just upgraded visuals and great storytelling. It seems that some of my ideas were ahead of their time or simply what Bethesda Game Studio was planning on doing anyway.
I just happened to be browsing YouTube the other day and an interesting video popped up on my feed talking about this exact topic I wrote about in the Medium article as a suggestion of the direction Bethesda’s should take with the next game in the series.
Below is the video I ran into that shows how Bethesda is planning on doing just that — or what I recommended with the next Elder Scroll :
(If you cannot see it embedded for some reason, here is the direct YouTube link.)
Dynamic Conversations Between the Player and NPCs
According to the video above, the next Elder Scrolls game may allow for complex AI conversations between the playable character and NPCs within games. This means dynamic conversations with walking townsfolk in the game for instance. What I mean when using the phrase “dynamic” is ever changing and based on encounters or events occurring periodically within a single game experience. Interactions could be shaped based on how far a player is within the game or based on the surroundings of the town or location the NPC is found in or at.
An event can trigger change like a king being captured or a mountain falling down on top of a small village, so new conversations will occur based on such events taking place within game worlds. This already occurs in some open-world games today, but I imagine better algorithms for NPCs could make games incredibly immersive. I imagine an entire house can be changed with NPCs reacting to such an event that some of them would longer be occupying it and travel into some far-off location to work in a mine as an example.
More Possibilities for NPCs and AI within Games To Have Complex Interactions and More Random Events
The interactions between the NPCs themselves could greatly be altered or changed over the course of a game as well, thus making a single playthrough feel fresh each time. We already have been seeing this with games like Stalker 2, but only recently due to patches finally making the developers’ promise become closer to the reality or vision of the game’s years-in-the-making development. Stalker 2 continues to be patched all the time with its A-Life system of AI behavior of NPCs being more interesting, unpredictable and life-like as work continues.
I have not played the latest Stalker yet, but after seeing some of the footage of how it now plays with the latest updates to its A-Life system, I am very keen on doing so. The original Stalker games were already incredibly immersive to me and offered some of the best AI I have played in games as a whole. You could be walking in The Zone and all of a sudden run into ambushes or NPCs fighting one another as different factions roamed the terrain.
AI interactions between NPCs is actually what I really like about botht he Stalker series and the Elder Scrolls series of games (mostly TES V: Skyrim). Being able to hear conversations and not have to click on every character to get info that way is a great immersive way to get a player into the game and its lore or story. There was a lot of improvement with this system in Skyrim vs how it was done in TES IV: Oblivion.
Potential for Game Longevity and Replayability To Be Improved Due to Improved AI Implementation
Going back to the upcoming and sixth title, dynamic AI conversations could really open the game up to longevity and also to replayability. after completing game or mission objectives, for instance, the player can walk around talking to different characters with new experiences and responses related to the completed event that the player was responsible for.
This may be one of the reason for the upcoming game’s development time taking so long; such a gameplay change could open the game up to potential game-breaking bugs or make it too easy to complete. Thus, a lot of fine tuning and play testing is in order to get it right. But then again, it is a known meme in gaming circles that Bethesda games on release come packed with many bugs and it takes time to iron them out with patches.
Just imagine asking an NPC for directions and to show you where to go everywhere or even to fight for you or with you on quests. Having too many options could make the game become different than it was meant to be for the player. Unless, of course, play testing and fine tuning is done to make sure the experience will still be pleasant regardless of the player’s play style and myriad of options available with such new forms of interactivity.
Also this open the game up to characters tricking the player or having more ulterior motives than the developers necessarily envisioned or designed them for. It could be based on alignment or their personality types. This could open games up to more personalised and unique experiences every time they replay them.
Stories Gamers Tell of their Experiences Unique to Their Playthroughs Count
Some of the best experiences I’ve had with the Elder Scrolls games happened due to random events not planned by the developer, as I’ve previously described in my article on this topic (linked above). For instance, how one NPCs reacted during a fight I was randomly thrown into with a dragon. These anecdotes and events that occur, within say Skyrim, is what made the game so fun and replayable for me.
If Bethesda manages to make the next title even more like this and use AI tools to make each play session entirely unique and memorable, the game will truly be special. It also will open up the industry to AI being let free vs confined to specific patterns and developer input in game design. Maybe in the near future games will have games within the games run by the AI let loose to engage with players within the games in unique ways.
This makes me think of the movie Tron by the way (there is also a good PC game based on the title by the way) and how gaming and the real world could become closer to one another than ever before.
AI Also Used for Game Development Today
Another thing that is happening in gaming and that could show huge potential to improve gaming besides just interactions with NPCs or AI characters within games is how AI can be used to help create and even code a game. In fact, AI can already be used to code game worlds all by itself and without the need of an entire game studio or coders behind the scenes although this is still limited and a nascent tech.
A person with a great idea for a game can focus on his idea and how he wants the mechanics of a title to work and let an AI system actually code it for him or her. This can open up game development to more individual individual creators or Indy developers than ever before.
Of course, AI still cannot do everything, at least as well as teams of paid professionals do together… at least not yet. However, we are living in interesting times when a new revolution within gaming may occur that is more based on internal mechanics of games than outward appearance, graphics, or hardware-based potential of what a game system offers its user base.
One example I found of a no-code game generator that may be worth exploring for those indy devs, or simple people with a good idea in mind, for a game is Base44. I entered “AI being used to code games” into Google Search and found it along with the Google Adwords text description:
Game idea in your head? See it come to life with Base44. No coding skills required. Finally, a gamecreator that gets you.
It seems to me like this is just one example of what is available on the market today at low cost to potential game devs. The future is very bright in this aspect of utilizing AI to augment or even be a main factor in cretaing a game.
Automation with the Help of AI Makes Overall Development Faster & Cheaper in the Long Run
What AI is often used for today is to automate tasks, simpify things and also create game objects that can be replicated on the fly and without the need for an artist to individually add every object or structure into an open-world city, as an example. Grand Theft Auto Definite Edition, especially now with the 2024 patch that fixed its previous shortcomings, really showcases the power of AI and how it can fix a game series in a relatively short period of time. This is because the title improved previous entities in the franchise and without rleying on a team of artists ot individually fix everty asset by hand, and AI was used to enhance the visuals within the games.
“AI upscaling is used frequently, which is a technique that uses machine learning to smartly enhance and clean up detail in the low resolution source artwork, according to Digital Foundry. “The second technique sees the developer redraw the assets, mimicking the original art but at far superior resolution. Finally, other assets are simply replaced with new art that can look quite different – often lacking the quality and style that went into the original game.”
Generative AI is the Buzz Word and Tech Right Now
Generative AI is huge in games today. Game objects that previously required development teams to be adde dby hand or randomly generated within the game world can be done on a massive scale using AI. The developer can still focus on crafting unique assets or objects within the games in iportant sections that they want done in a certain way while levaing room for AI to take care of the rest.
These game assets can be repeated countless times while the AI does variations of them and still feel fresh, unqiue or different from one another. Fauna or flowers is a good example of somehting that can be AI generated while developers focus on other aspects of the visuals within games.
AI and farms of servers leveraging features within games may be more important than the piece of hardware you see in front of your TV. However, RAM and some aspects of hardware should be greatly improved as AI can benefit from this.
Certain Hardware Specs More Important than Others for Upcoming Systems?
The system has to store data based on a lot of dynamic changes occurring periodically and this was an issue on the PlayStation 3 version of Skyrim as it did not have as much RAM as the Xbox. the game had to keep track of items being left all over the ground or inside dungeons in a huge open world and it taxed the PlayStation 3 to a point crashes were occurring and quests to this day are broken. Although a patch mitigated some of these issues.
Gaming systems should continue to evolve and be iproved, but also not be afraid to draw form the cloud and server farms to add to games. Sure, many gamers will complain they have to always be online to get into such a game, but complaints will be mitigated by the implementation and successful innovation within the games themselves.
I remember when the original Steam launched people complained about its at the time always-online requirement, which Valve went back on, but now other streaming services have this requirement and I do not see any complaints about it anymore. Gamers are fickle and over time change habits depsite early complaints if something is successful or worth a sacrifice such as always being connected to the server or being always online to play a game.
Cloud streaming is now also gaining steam with the likes of Nvidia’s GeForce Now. However, Google failed in this approach to gaming with the Stadia game-streaming platform, so it also is not an easy task and something that requires fast Internet speeds to function well. Despite this, it will continue to make advancements and be used in the future as an alternative form for game connectivity besides owning expensive hardware at home with the latest specs.
Older Games Also Showed Potential of AI That Were Ahead of Their Time
Many games in the past were coded in a way that made them stand out and even stand the test of time in the potential they showed for AI to impact gaming. Whether this was because they had incredible coders behind them that made the game AI stand out for the time or due to ideas they presented within the games that would one day become standard.
One example of such a game was Seaman on the Sega Dreamcast. The game had players raise a sea creature in an aquarium and in concept was similar to Tamagotchi. However, it went beyond just raising a creature and feeding it over time.
Players could talk to it and it would remember your previous answers as you, the player, asked it. It would also hold somewhat dynamic conversations with you. I’ve seen people describe it as a precursor of what ChatGPT does today and allows for. Obviously, it was on a much smaller scale and based on coding skills rather than the algorithms we base AI on today with learned models and databases of information these algorithms have at their disposal today raw from. However in many ways it has the basic idea of how ChatGDP is used today right, and it did it in a visual form: a game.
The Dreamcast as a gaming system was so ahead of its time and I owned one it is insane to think how Sega failed with it to capture the gaming market and went out of business in terms of hardware development of gaming systems following its failure. However, that is a separate discussion for a separate post or article. I briefly go into it in the Shenmue trailer being AI based linked on the bottom of this post that I recently published on Medium. So check it out for more of my thoughts on this subject.
When we talk about great AI in gaming we can also look at chess games and how they evolved over time. There are various Chessmaster games available for those who enjoy playing chess against a computer or an AI system of sorts and they generally have been improving over time, but the latest releases of this franchise occurred in 2007/2008. Also, and this is more relevant today or in the modern day, Chess.com, which can be played via a browser is supposed to have descent game AI, although I have not tried it personally.
In 1996, a famous chess game was played between a chess world champion named Gary Kasparov and an AI system named Big Blue. During the first match, Deep Blue overall lost more than kasparov. However, the software was upgraded and in the second match Deep Blue ended up winning against Kasparov more than Kasparov won against the AI. This was seen as monumental.
IBM Watson also ended up winning Jeopardy in 2011. This was a huge win for AI and showed how powerful it can be in games that are played by millions of people every year.
Getting back to the subject of mainstream computer or video games showing great potential for AI, we should consider how computer-controller opponents, called bots in some games like first-person shooters, are programmed and were programmed in the past. Personally, I consider the real-time strategy series of StarCraft (I played both games) as an example of good AI in terms of fighting comoister-controlled opponents.
These computer opponents had behaviour, in the way, that mimicked players attacking bases and defending them well. What I mean is it mimicked how players acted in multiplayer matches up to a certain level well without feeling cheap or cheering too much like it can feel like when playing against computer-controlled opponents in other strategy games.
I also found some corridor shooters like Quake 3 Arena to have good bots for single player gameplay when online opponents were lacking on harder difficulties. Quake 3 came out all the way back in 1999 if I recall right, although StarCraft 2 is a lot newer and Blizzard still updates it for at least multiplayer gameplay and unit balancing periodically.
Another game that is a classic in its genre and a staple in story-telling and story progression that I want to mention having good AI systems in place for characters within the game world you encounter and how they interact with one another is Baldurs Gate 2. The game allowed you to recruit various characters ur o your party and they would interact with one another and sometimes even start fighting depending on their allegiance or their morality they had (say chaotic evil or lawful good would not get along well).
Every playthrough felt different because you could mix and match different characters and see how they would work together during battles or time traveling the world map and react dynamically to other party members next to them on the journey. This was a good example of developers creating effective AI onto their game years ahead of its time.
AI in games goes way back as all non-controlled entities within games (whether robots or NPCs) can be described as a rudimentary form of AI in a way designed by the programmer or team of developers to run a certain way and function or react to the player a certain way. Whether this reaction stays the same every time or is altered and dynamic in a way depends on the algorithms and the way the game was programmed (and all aspects of it, including the NPCs etc.). You can go as far back as the 80s to see AI influencing game design.
AI can be something complex, or it can also just a buzzword for all non-human entities and the way they function in games.
The Future of AI in Gaming Has No Bounds
AI is touching all industries and augmenting them in various ways today. Gaming is just one of many that it truly can transform in the way we both look at games and how they are or will be created. I previously wrote an article on Medium about this idea of how AI will transform everything and today some years after writing it here we are… gmaing is just one of the many industries it is altering today.
One thing that I worry about is that adding too much complexity to games and giving too much power to AI algorithms within the games can make them too cumbersome for users or make a game take too long to complete and really get bored of. This will cause gamers to keep one game and just play with its options for long periods of time and not purchase new titles, thus making the industry less profitable for developers.
However, this is how gaming used to be when I was growing up and when the price of cartridges were around $70 for new titles in terms of systems like the Super Nintendo that I owned back in the mid 90s; and keep in mind that adjusted for inflation this often would amount to hundreds of dollars today. Us kids growing up back then couldn’t afford to purchase or ask our parents to purchase games often and would play around with a single title for long periods of time. Although this may not please developers in the long term, it is better for gamers and if a game is good new players will buy it anyway and despite how busy of their backlog is these days with all the Steam/GOG sales and Epic Game giveaways available.
The bottom line is we are living in a tech revolution governed by uncertainty and without limit of future potential. AI is truly allowing things to be done in industries like gaming not possible ever before. In a future Elder Scrolls game we can may be able to have long conversations with NPCs and even maybe find a cheap shrink, of sorts, within a game world living within our monitors or TV sets.
I really do think that AI advances in gaming as well as the way it is implemented for development of upcoming games will be what drives future platforms in gaming. As technology advances, what is possible also will advance and algorithms in the way they are used for say NPC behaviour (as Stalker 2 shows in the videos above of what its developers are trying to do with A-Life, for instance,) will drive the industry forward.
In the future, games will have a lot less scripting, especially outside major plot points the dev teams will want the players all to experience and path finding for NPCs will go beyond what it is today. Simple pathways and waypoints will be replaced with NPCs walking around doing different things all the time. Towns will become like real-world towns where outside having to work in specific locations, NPCs will randomly do different things every day in the in-game clock.
AI and various aspects of it as it is implemented to or within the games industry will be what will keep players engaged and this includes gamers bored of endless sequels without much change to them over previous entries besides the usual improvements in visual fidelity, sound and scope. AI changes will allow them to keep playing and seeing ways to interact with the game worlds and characters they have cone to love by playing games in long series and often across different consoles.
By the way, I would also like to close this post with a link to another recent gaming article I published on Medium that sort of touches on this topic a bit. It relates to a Shenmue 4 AI-generated trailer that people thought may have been a real trailer as they were confused whether it was a reveal of an actual game or fully AI generated. I argued for the latter and believe it was proven or revealed to be AI generated since my article.
The trailer showed how AI can be used to trick viewers into believing something is professionally developed and also to create storylines, and ideas for sequels. The future is very interesting in gaming as it relates to how simple pathfinding for NPCs will be replaced by complex behaviours and AI algorithms. GTA VI and TES: VI are the games to truly watch in this space when they release. Stalker 2 is also worth following as it receives updates over time to its A-Life system.
Below this story link, I also linked the Elder Scrolls story I wrote (which I also have hyperlinked above, but not all readers here may have noticed it) where I gave Bethesda advice on what the next title should focus on, which again was AI interactivity.
- https://maciej-duraj.medium.com/new-shenmue-4-trailer-or-anai-generated-idea-that-shows-potential-4b477e57db29
- https://medium.com/super-jump/the-elder-scrolls-vi-should-focus-on-ai-interactivity-over-visuals-e491ad62a9e2
Also, please check out my much older article on how augmented reality was changing gaming or its potential as well. I published it under my former U.S. name or byline of Mike Lata for IDG Consumer, specifically the blog Tech Hive and Macworld, where I was an associate editor in San Francisco:
Maciej Duraj 
Leave a comment