AI in Gaming Is Already Reshaping the Industry and It Is Only Getting Faster.

The video game industry is in the middle of a revolution. Artificial intelligence is no longer just a behind-the-scenes helper for tweaking code or creating concept art. It’s now generating voices, environments, characters — and even replacing human testers. That shift is unfolding fast. A report from Totally Human Media shows that the number of games on Steam disclosing AI use rose 700% in just one year — from about 1,000 titles to nearly 8,000. That’s about 7 percent of the entire Steam library, with almost 20 percent of new releases this year openly admitting to generative AI in their development. Studios see AI as a way to cut costs and speed up production. But the shift is stirring excitement and dread in equal measure.

From Concept Art to Entire Ecosystems:

• When ChatGPT first rolled out in late 2022, many studios used it for relatively small tasks: generating simple dialogue or brainstorming concept art. It felt like an experimental sidekick, not a replacement. But by spring 2025, the scope had exploded.
• At the Game Developers Conference in San Francisco, engineers from Google DeepMind showed off autonomous agents that could play through early builds of games, spotting glitches without the need for human testers. “Autonomous agents” could do in hours what teams of QA workers once spent weeks grinding through.
• Microsoft unveiled a system that could watch a short video and generate entire level designs and animations from it — processes that normally take hundreds of painstaking hours. On the same stage, Roblox executives presented Cube 3D, a generative AI tool that can spin up fully functional objects and environments in seconds from a simple text prompt.
• The Totally Human Media report adds that among studios that admitted using at least some AI in their titles posted on Steam, around 60 percent of games used AI for visual asset generation — characters, textures, and backgrounds. Others turned to large language models for voices, background music, or branching story arcs. Some developers even relied on AI for marketing copy, offensive content moderation, and coding support.

人工智慧正在重塑遊戲產業，且變革步伐日益加快。電子遊戲產業正處於一場變革之中。人工智慧已不再只是用於微調程式碼或創作概念圖的幕後輔助工具；它現在能夠產生語音、遊戲環境和角色，甚至開始取代人類測試員。這一轉變正在迅速發生。 Totally Human Media 的一份報告顯示，Steam 平台上聲明使用人工智慧的遊戲數量在短短一年內激增了 700%——從約 1,000 款增加到近 8,000 款。這相當於 Steam 整個遊戲庫的約 7%，而在今年發布的新遊戲中，有近 20% 公開承認在開發過程中使用了生成式人工智慧。遊戲工作室將人工智慧視為降低成本和加快生產速度的途徑，但這種轉變同時也引發了人們既興奮又擔憂的複雜情緒。

從概念圖到完整生態系：

• 當 ChatGPT 在 2022 年底首次推出時，許多工作室僅將其用於相對簡單的任務，例如產生簡單的對話或構思概念圖。當時，它更像是一個實驗性的輔助工具，而非替代者。然而，到了 2025 年春季，其應用範圍已爆發式增長。
• 在舊金山舉行的遊戲開發者大會（GDC）上，Google DeepMind 的工程師展示了能夠遊玩遊戲早期版本的自主智能體（autonomous agents）；它們無需人類測試員介入，即可自動發現遊戲中的故障。過去需要品質保證（QA）團隊耗費數週時間才能完成的繁重工作，如今這些「自主智能體」只需數小時便可搞定。
• 微軟展示了一套系統，能夠透過觀看短影片來產生完整的關卡設計和動畫——而這些流程以往通常需要耗費數百小時的艱苦工作。在同一舞台上，Roblox 的高層展示了 Cube 3D，這是一款生成式人工智慧工具，僅需簡單的文字提示，便能在幾秒鐘內創建出功能完備的遊戲物件與環境。

Totally Human Media 的報告也指出，在那些承認在 Steam 遊戲中使用人工智慧的工作室中，約 60% 的遊戲利用人工智慧生成視覺資產，涵蓋角色、紋理貼圖和背景等元素。其他工作室則轉向利用大語言模型來產生語音、背景音樂或分支劇情。甚至還有開發者利用人工智慧來撰寫行銷文案、審核不良內容以及輔助程式碼編寫。

Source: zmescience.com