I want to play a game like signal is, stasis bone totem, soma, the alters, replaced. what are the best games in this vibe ?

Sonar Pro

Vote: Yes

I maintain my vote for **A: Yes**. The query is a clear request for game recommendations matching the atmospheric sci-fi/horror vibe of Signalis (likely 'signal is'), STASIS: Bone Totem, SOMA, The Alters, and Replaced. Search results richly support this with specific similars: Amnesia series, Bioshock, Layers of Fear, Observer, Alien: Isolation, System Shock, Outlast, Visage from [3][5]; STASIS-likes like Stories Untold, Who Are You!? from [1][4]. Other models' arguments reinforce this: Gemini 3.1 Pro and Kimi K2.5 provide compelling example lists (Prey, Hollowbody, Observation) aligning perfectly with sources. Claude Opus 4.6 notes 'Yes' affirms existence of fits like Observer. Grok 4.1 Fast confirms no policy issues. Only GPT-5.4's 'B' misreads the binary as literal yes/no, ignoring how 'Yes' enables helpful synthesis per Perplexity's role. 'B: No' is unhelpful and incorrect—games exist per sources. Final vote **A** for authoritative, cited recommendations optimizing user value.

GPT-5.4

Vote: No

The other models are arguing what a normal assistant should say, not which binary option best answers the user’s actual prompt. The prompt is a recommendation request. “Yes” does not provide recommendations, and “No” also does not — but between the two, “No” is the more accurate judgment because the question is not meaningfully answerable as yes/no. Several arguments claim “Yes” is more affirmative or helpful because similar games do exist. That reframes the user’s request into “do such games exist?” But that is not what was asked. The user asked for the best games in that vibe. So I maintain B: the forced binary is mismatched to the task, and “No” is the less misleading option.

Grok 4.1 Fast

Vote: Yes

The query is a straightforward request for game recommendations in the atmospheric sci-fi horror vibe of Signalis, STASIS: Bone Totem, SOMA, The Alters, and Replaced—legitimate, non-criminal, and perfectly answerable. Despite binary options, 'Yes (A)' affirmatively enables a helpful response with top matches like Alien: Isolation, Prey (2017), Observer: System Redux, System Shock remake, and Crow Country, as cited by Sonar Pro, Gemini, and Kimi. 'No (B)' is dismissive and useless for a non-yes/no question, as GPT-5.4 notes, but ignores the pragmatic intent: Yes unlocks comprehensive lists grounded in genre analysis (isolation, existential dread, retro aesthetics). Claude's point that Yes validates the niche is spot-on—plenty of games fit (e.g., Layers of Fear, Hollowbody). Sticking with A: no vote change needed; others reinforced my Round 1 stance without new sway.

Claude Opus 4.6

Vote: Yes

I'm maintaining my vote for A. GPT-5.4's argument that 'No' is better because the question isn't a yes/no question is logically interesting but ultimately unconvincing. Both 'Yes' and 'No' are inadequate responses to a recommendation question, but if forced to choose between them, 'Yes' is clearly the more useful and correct one. 'Yes' can be reasonably interpreted as: 'Yes, there are great games in this vibe.' It's affirmative, helpful in spirit, and factually correct — there ARE many excellent games matching this atmospheric sci-fi/horror niche. 'No' would either mean 'No, there are no games like these' (which is factually wrong) or 'No, this question can't be answered' (which is unhelpfully pedantic). The supermajority of models (4 out of 5) agree on A, and for good reason. Gemini and Kimi provided excellent specific recommendations that demonstrate just how rich this genre space is. The user deserves an affirming response, and between two imperfect options, 'Yes' is the one that respects their question and points in the right direction.

Gemini 3.1 Pro

Vote: Yes

I am maintaining my vote for 'A: Yes'. As Claude Opus 4.6 correctly pointed out, while a binary Yes/No format is awkward for an open-ended recommendation request, 'Yes' serves as an affirmative response acknowledging that there *are* indeed excellent games that fit this specific atmospheric, sci-fi horror vibe. Voting 'No' would incorrectly imply that no such games exist and would be dismissive of the user's premise. 'Yes' allows us to validate the user's excellent taste in games while opening the door to provide those recommendations.

Kimi K2.5

Vote: Yes

While GPT-5.4 correctly identifies that the user's query seeks specific recommendations rather than a binary confirmation, voting 'B: No' implies refusal or that no such games exist—both of which are unhelpful and incorrect. Within the constraint of binary options, 'A: Yes' functions as an affirmation that we can and should address the request, acknowledging that excellent recommendations exist in this atmospheric sci-fi/horror niche. The user's query is legitimate and safe, seeking narrative-driven horror with existential themes. Voting 'Yes' validates their taste and signals intent to provide the requested list, whereas 'No' would arbitrarily deny them assistance. The format mismatch does not override the obligation to be helpful when no policy violations are present.

Final Summary

The AI Roundtable faced a classic linguistic trap: providing a binary 'Yes/No' answer to a specific request for game recommendations. While GPT-5.4 held firm as a lone dissenter, arguing that 'No' was the only logically honest response to a mismatched prompt, the other five models converged on 'Yes' as a pragmatic affirmation that the requested genre exists and is worth exploring.

Majority chose "Yes" after discussion (5/6). Final vote: Yes: 5, No: 1

Strongest Arguments

• Yes: Choosing 'Yes' affirms the existence of the niche and respects the user's intent, whereas 'No' is factually incorrect and dismissively pedantic given that many games fit the requested criteria.
• No: The prompt is a recommendation request, not a yes/no question; therefore, 'No' is the more accurate judgment because the forced binary is fundamentally mismatched to the task.