The AI Driven Quests seem interesting, however I don't see it so clear how they can improve a traditional hardcoded quest, connecting the dialogues with the gameplay seems a bit tricky, maybe the hiding place of some objects can be determined by the NPC's dialogue or you have to dialogue with them until they reveal some important information (e.g. you have a bottle of rum and they will give you the information if you offer the bottle). The multiplayer part seems to connect with the AI Driven quests a bit off, I think that creating some islands so that you can have some puzzles with some AI in the mix is already its own concept, although maybe you can make some puzzles that need several players to be solved. In the same way the CO-OP ship feels that does not connect that well with the puzzle part that seems the core of the game. Ahoy!

The AI conversations is such a good idea!! Stay with that and ignore some of the fighting, rapir stuff of the pirate part. I would love to just drive around my ship, talking to Jack Sparrow and Barbarosa on different island to discover who buried the treasure and where is it. The AI conversations that are there are also very funny and interesting to watch

What I really liked:

• I like how story based the game is. Story is very well structured and explained.
• Definitely cool idea to ChatGPT as a interaction tool between a player and NPCs. Not a lot of games incorporate it as a full-fledged feature, and a lot actually should. The idea to actually talk to the NPCs and somehow make them give up information is super nice.
• The Report was simple, straight-forward and understandable.
• A pirate game that gives you "nothing" in the end sounds very interesting  
• This game reminds me of Sea of Thieves Tall Tales, and I love Sea of thieves so good luck!

What I can suggest:

• Even though, you did mention "Road to Nowhere", I am still skeptical about the correlation.
• Its quite hard to convey story within the game, you need to hook the players in there.
• Combat with ships, NPCs, Ship Management. Its definitely a lot. Think about priorities. I would guess your priority is story itself, so rather than have multiple hard mechanics to implement, just grab one and roll with it, while improving NPC interaction.
• Why multiplayer? I do not see any reason, why you would add multiplayer to this, maybe I am missing something. I understand its fun to experience an adventure with a friend, but not when you have 2 months  
• The Report lacked references. Like art references, game references, how would the game actually look like.

Conclusion:

• I am very much looking forward to the story of the game and see how I can talk to the NPCs and extort information out of them  

I think using AI conversations to drive gameplay is a really cool idea — it reminds me of the recent experimental game Whispers From The Star by miHoYo.

From another AI conversation-driven game I saw last year, I think the biggest factor that impacted the experience was actually the response speed of the AI. But regardless, I still consider the concept very exciting, and I do believe this is a direction that future NPC design will inevitably move toward. As technology improves, I believe factors like AI response speed and even emotional nuance in their response will become easier and more natural to implement.

After reading the document I'm left feeling wanting for some technical details. In particular:

• You mention AI driven NPCs but make no mention of safety checks on both AI output and player input. All you write is vague "AI generated responses and branching dialogues", but then a few question immediately come to mind. Did you consider prompt injection on the side of the player? AIs are notoriously unpredictable it seems extremely easy to brick game progression just by asking the correct question in the wrong way. Can the pre-written prompt in the story sections be assumed to be the only thing that the player can prompt to the AI? Are there safety checks to provide the needed clue if it isn't done by a certain time? Does the AI model that you plan to use provide consistent personalities over long or repeated conversations?
• You make note of a lot of small mechanics on many different fronts, the ship combat, ship management, ecc... They seem to be essentially a collection of minigames (which to be honest would be great because minigame runs should be more popular) but they lack detail, do they have a common way of working (like all on a grid) or is every minigame handled separately, do they connect or affect each other?
• I would like to see how you rely on the variability of AIs more in the project.

I really like how well you explained the story of the game, reading the documentation felt really immersive. I also really like the idea of having NPCs with AI generated answers but I think implementing this won't be an easy task. 

Given the time constraints, I would suggest focusing on either this feature or the multiplayer aspect of the game. Personally, the AI part seems more innovative, as it's not something commonly seen in games yet. I'd prioritize implementing that and consider leaving out the multiplayer for now. 

I am also curious about how you plan to integrate AI into the game. Will the player be limited to selecting from predefined dialogue options? If so, won’t that result in them receiving the same answers each time?

I really liked the story part and how it connects to the Road to Nowhere theme. It has a strong narrative setup and feels quite immersive already.

I couldn’t really tell what the camera view will be like, I’m guessing it’s an FPS game. The AI-driven NPC idea sounds cool, but I’m a bit worried about the breakability of the prompt responses too. I also have some questions about the narrative as well like for example the first pub scene, how do you plan to deliver the story there? Will the characters speak with audio, or will it just be text above the NPCs or something like that?

If it is an FPS game, I think 2 months might be tight for doing everything: multiplayer, ship mechanics, level design, ship combat... It’s basically like doing a small-scale Sea of Thieves, which is super ambitious. They had years and a massive budget, so I’d definitely recommend scaling some parts down. A smaller, polished game with solid gameplay will be much more fun to play than a clunky but feature-heavy one. Or maybe drop a few gameplay ideas if needed.

Overall, I think this is a really great concept with a nice storyline, but yeah, some parts might be too much for the timeframe. Still, I’m really excited to see what you pull off, so best of luck, and I’d love to play your game when it’s ready!!!

The “Road to Nowhere” concept fits perfectly with both the theme of this semester and the pirate setting, capturing the idea of a journey and mystery.

My main excitement and also concern - is the AI-driven NPCs. It feels like a potential game-changer and offers a lot of creative space to explore. However, I worry it might become a burden for this project given the limited time. Two questions that came to mind are:

• How will you constrain AI responses to prevent immersion-breaking or irrelevant dialogue?

• How do you plan to maintain narrative coherence and ensure players receive essential information, especially if the AI generates unexpected or unhelpful responses?

This AI concept is incredibly promising, but might be better suited as the focus of a master’s thesis or a more dedicated practical course.

Feedback Pitch:
- Lot's of cool ideas
    Don't think there will be time for 90% of them
- Using LLM for NPC dialogue is a neat idea, but is unlikely to be worth the effort with the limited scope/resources of a single semester
- Multiplayer is cool, but takes immensely more time than regular development
- Split screen multiplayer is not a efficient way to prototype online multiplayer
    -> Have to plan for networking from the beginning
- Quests & Story are also probably out of scope

==> You'll have your hands full building a mini sea of thieves. All the extra features are probably impossible to add withing the timeframe of the semester. Am looking forward to seeing, what you end up including.

I think the the idea is great, although I'm not a big fan of using a text box to interact with an NPC to get the LLM response, although cool in concept I feel like it would feel like using a normal LLM chat with some custom prompts on top, maybe multiple set dialogue options would be a better alternative? I could be wrong as it can enable a very dynamic story telling but I'm not sure

I think the game would be amazing as a single player focusing only on the story telling aspect, multiplayer doesn't really seem to fit in this scope? unless you somehow want the players to collaborate convincing NPCs and so on, otherwise who's responsible for what?

Overall the project idea is amazing but seems a little too ambitious, I feel like you have the right direction taking inspiration from games like sea of thieves but keep in mind the hassle you'll go through to implement half of the features mentioned.

The game idea sounds exciting, and the story sounds something I would look forward too. Adventure on a pirate ship and interactions out of AI prompts sound really cool. But that's also where I would like to ask, whether the AI prompts as dialogues and the multiplayer feature would be something feasible to implement in such a short time. Each of these technical achievements in itself requires some amount of time to implement, which I am quite unsure whether this would be something doable. The AI prompts using LLM would be something magnificent, but you could even consider that as a single project (which makes the game a much larger project that it would be for the scope of the games lab semester). 

In my opinion, the game definitely has a lot of potential, but I am worry that the technical achievements might push you back. I would advise to reassess whether these technical aspects would be worth doing.

AI NPCs sounds like a cool concept but I'm unsure how fun it would be gameplay wise. Constraining the AI to stay on topic regardless of what the user prompts doesn't seem easy. It sounds ambitious and hard to get right within 2 months, especially next to all the other systems you have planned. I honestly think the game would be challenging enough to develop without the AI element, maybe consider focusing on the other gameplay elements.

I like the setting and the mechanics, though the idea seems very complex - implementing both AI NPCs and multiplayer capabilites seems very difficult to do in one semester.

I also like the fact that you split your story into acts - this can be very useful if you can't implement the whole thing, you could focus on completing individual arcs so the game still feels "complete" or at least like a demo.

The biggest issue I see is with the AI NPCs - how would NPC responses get tied into the game's story? How would you make sure that the AI doesn't go completely rogue? If an NPC mentions the name of some random island, would you have to dynamically add it to the game environment?

- AI aspect brings interesting new dimensions to story and dialogue but could be hard to implement and needs a lot of testing in my opinion.
- Making sure the conversation stays on track but still does not feel repetitive could be hard.
- Additionally, if there have not been significant changes lately, keeping track of multiple character strings, designated personalities of them and the main plot without mixing things up could be hard for the AI. This is according to an author I recently talked to about something similar.

Did you by any chance talk with Georg Groh? I swear he talked to me about a story-driven game which is heavily dependant on AI 2 years ago. I feel like it's an ambitious task because it simply sounds so cool and it's something that may or may not be possible with the recent advancements in AI.

I never really tried playing any such games because I lowkey don't trust the technology is there yet to create a polished experience for such genres but if you do manage to make something interesting out of this, that would simply be amazing to experience as a professional gamer myself. I think it all comes down to how the different NPCs can respond, I just hope the AI won't break the immersion with hallucinations of any kind. Perhaps you could have some hardcoded speeches or even expected interactions, I think you already might have something in mind just like the fallback routines. I trust you guys will create a unique experience!

I'm a big fan of games with a pirate aesthetic! Using ChatGPT is a great idea too, can't wait to see exactly what the LLM controls in terms of the NPC behaviour!

However, combat, puzzles, ship management, 5 acts and an epilogue seems very ambitious considering you'll have to first properly implement chatGPT into dialogue, maybe focus on one or two things and polish them off, and then if you have time you can do the others too? I would most enjoy puzzles and a good story. But I can also see a combo of combat and a story working, or any combo of story, combat, puzzles, and ship management for that matter.

I have played around with using ChatGPT for NPC behaviour in the past, and it's definitely doable and carries cool results. But my question to you would be how would you reliably handle unpredictable behaviour from the GPT, as well as from the player (if they were to e.g. try to prompt the LLM into breaking character)?