SEGA Will Use AI In Game Development-Carefully, And Only Where It Adds Real Value
SEGA confirmed in a Q&A following its Q2 financial results that it will use AI in game development, but only where it makes sense. The company emphasized caution, noting industry pushback around AI use in creative roles like character art.
"Rather than fully following the trend toward large-scale development, we will also pursue efficiency improvements, such as leveraging AI... we will proceed by carefully assessing appropriate use cases, such as streamlining development processes."
Translation for engineering teams: SEGA isn't chasing headcount. It's targeting operational efficiency and selective automation. The goal is to reduce production overhead without compromising a franchise's art direction or voice.
What This Means For Dev Teams
- Pipeline automation: build orchestration, asset validation, log triage, flaky test detection, and smoke tests.
- Content ops: asset tagging, deduplication, LOD generation, compression presets, and batch conversions.
- Localization support: first-pass translations and subtitle timing with human QA on final text.
- Design tooling: prototyping scripts, dialogue variants, and placeholder barks-reviewed and approved by writers/designers.
- QA augmentation: synthetic test data, bug clustering, and telemetry summarization to speed triage.
None of this replaces artists or writers. It reduces time spent on repetitive glue work so experts can spend more time on the parts that make games worth playing.
Guardrails You Should Expect
- Human-in-the-loop for character art, narrative, and core IP assets; clear opt-in from creative leads.
- Dataset provenance and licensing checks; no training on unapproved IP.
- Secure model usage: avoid uploading unreleased content to third-party services; isolate models and logs.
- Reproducibility: versioned models, pinned prompts, seed control, and documented evaluation baselines.
- Cost and latency budgets: measure token spend, inference times, failure modes, and rollback plans.
- Policy alignment: union agreements, credits, and disclosure where AI-assisted content ships.
This approach mirrors a wider sentiment in the industry. The director of The Outer Worlds 2 said you "shouldn't outsource the things you're good at to AI," while others at the studio noted they haven't been using it. SEGA's stance lands in the same zone: use AI for speed, not for identity.
SEGA also shared a separate update: it plans to increase the average monthly salary for all staff by 10% starting April 2026.
Source: VGC
If you're evaluating AI for your own pipeline and want structured upskilling, these resources can help: Prompt engineering practices.
Your membership also unlocks:
