"The line must be drawn here! This far, no further!"

Please forgive the Captain Picard First Contact quote (it's been a couple of weeks since Patrick Stewart's death and only now can I mention his most famous character without welling up) - I promise it's relevant as an illustration of an admirable, principled stance.

In the Borg, good ol' Jean-Luc had a well-defined enemy to engage, with the threat of humanity's assimilation providing a pretty clear line between good and bad. There are larger philosophical questions about the hive mind and our aversion to Star Trek's cube-riding automatons, but for the purposes of this article, let's assume we're all in agreement: Borg=baaad.

Back in the real world, though, that line between right and wrong often isn't so easy to define. Today there'll be a lot of commentators in the video game space, players and developers alike, longing for a better understanding of another encroaching, all-consuming entity.

House House's belated disclosure that LLMs were used during the development of indie hit Big Walk pours fuel on an already raging debate and brings questions over permissible AI use in game dev right back into the discourse - specifically the differences between generative AI for player-facing audio-visual assets and writing versus coding and other less obvious uses.

This latest flashpoint followed a statement released by a subset of IGF Award jurors explaining why they won't be voting for games developed using any form of generative AI.

Last year, Clair Obscur: Expedition 33 — which swept The Game Awards with nine wins from 13 nominations — raised similar questions after the Indie Game Awards disqualified Sandfall Interactive's game upon discovering AI had been used in the development pipeline.

I've voiced my distaste for GenAI, and it remains a massive no-no - professionally and personally. I'm fortunate to not have corporate overlords mandating its use around these parts, which certainly isn't the case for some other outlets, nor for many creatives in the gaming space. The gulf in quality between the human-written word and generated bilge is still enormous and easily identifiable, and for now, the same is broadly true of GenAI visual artwork, too.

What this Big Walk situation highlights, however, is how difficult it can be to discern in other places, and the increasing pull it has on even the most principled developers.

The phrase gets bandied around a lot, but it's fair to call Big Walk a proper 'critical darling'. It garnered "Universal Acclaim" back in August, with an impressive 91 Metascore. Alex gave it a "Great" 8/10 after reviewing the Switch 2 version for us - technical hiccups took the shine off it for him, but he still loved the co-op experience. And by September it was already Nile's GOTY.

As someone who didn't play Big Walk at launch, its place in my GOTY-catch-up priorities may have shifted, but wherever you land on the GenAI spectrum, the consensus among critics and players is as unanimous as it gets: Big Walk is an excellent video game.

There are many issues to consider — not least where AI use crosses from tool to theft — but an important one relates to how this game in particular has had a profound effect on many players. Rejecting something flagged as 'suspicious' on principle is easy, especially with so many brilliant games releasing each week. But something you've played already and internalised and loved? From a dev team you respect?

For some, this revelation will feel like when a beloved artist/actor/writer/musician is revealed to be problematic, but we're not talking about illegal behaviour or allegations here.

If House House's statement is to be taken at face value, the team used a chatbot to find a solution to a coding issue with Unity, the third-party engine the game is built on. People online with real chops have already investigated the specifics of House House's single example:

That's the one example the devs offer in their statement, and the language around further usage feels vague. The impression it gives is of a team searching for a solution to a specific problem using common tools, as opposed to vibe-coding huge swathes of the game.

Now, I've never programmed anything in my life, but I can appreciate that efficient code must be an art in itself. Is it any less of an art because that code is invisible to players? Then again, to me, generating great chunks of game code feels very different from using an LLM to squash a bug or find a solution when third-party documentation fails. Is one usage as 'bad' as the other, or is this slippery slope, gateway-drug territory?

(Perhaps if any programmers are reading, they'd like to share thoughts in the comments - as the headline says, this is very much a talking point and I'm keen to explore with as much nuance as the internet will permit.)

My point is that seeing exactly where to draw that line is increasingly difficult when AI tools of all types become ubiquitous. A blanket rejection of anything inevitably now means excluding not just an increasing number of games, but also artistically significant, important ones.

And with executives like former Nintendo of America president Doug Bowser advocating the use of AI for indie teams to improve workflow, it's also easy to understand how a developer might turn to the tech. The whole topic is getting greyer by the day, much to the pleasure of evangelists, I'm sure.

Like it or not, this situation forces AI sceptics to reconcile their positive feelings for the game. House House estimates "well under 0.1% of Big Walk’s code" has any trace of generative AI, but to paraphrase a certain Starfleet captain again, "How much does it take before it becomes wrong, hmm?" That is the question.

For some players who simply don't care, it's a non-issue. Some may reluctantly cut the game from their lives and GOTY lists. I suspect that others, though, disappointed or otherwise, will start reconsidering where they personally draw the AI line.

And if House House hadn't disclosed this, would anyone have ever known? I'd guess not. And if a celebrated indie team turned to AI as a problem-solver, how many other devs aren't disclosing similar usage? And how many sure-as-hell won't do so after this?

Lots of questions, no clear answers, just murk. How unsatisfying. I've heard 'ends-justify-means' sentiments online — 'hey, if the game is good, does it matter?' — which feel unsavoury given the documented, quantifiable damage AI is doing to the games industry and the wider world. It's that damage which has IGF jurors abstaining from voting from Big Walk this year: an understandable, admirable stance.

Which leads us to one final question: What would Captain Picard do?

Naturally, he'd call everyone into the Observation Lounge for a big old meeting in a calm, civilised manner, that's what. Conference.

Let us know below if this revelation changes your relationship with the game or where you draw the line when it comes to generative AI in its various forms.