During the “Speed, Customization, Innovation: AI in Gaming” panel during CES this week, game publishers and developers shared their latest insights regarding how they use generative AI tools. A prevailing question involved the impact of AI’s ability to generate pixels and video frames efficiently — especially in light of Nvidia’s keynote the prior evening announcing its new Blackwell RTX 50 Series GPUs’ enormous ability to do so. Other opinions shared during the panel included thoughts on whether AI is overhyped for gaming and wish lists for fixing the limitations of AI tools.

The panel’s moderator Dean Takahashi, longtime Silicon Valley tech journalist who is now lead writer at GamesBeat (from VentureBeat), set the stage with a highlight from Jensen Huang’s Nvidia keynote. Huang touted GeForce RTX’s incredible efficiency of generating three additional video frames via AI for every frame rendered through computations.

In response to that, HP Senior Director of Experience Engineering Haval Othman said the benefit is the game will feel more realistic. Whereas Burcu Hakguder, co-founder and CRO at Layer AI, believes the extra pixels generated can particularly help simulate shaking motions. And in general, she has noticed generative AI has tremendously shortened development of mobile games from the order of three weeks to three days.

However, the panelists acknowledged there are limitations with AI in gaming. Devin Reimer, founder and CEO of AstroBeam, observed that while AI can rapidly deliver 90 percent of gaming development, the remaining 10 percent of honing and perfecting work is not performed with AI tools.

Hakguder shared that artists often prefer a slower paced development to ensure quality and AI tools also have steep learning curves preventing adaptation by creative professionals. She also doubts agentic AI can create games that monetize, to which Gamesight Head of Influencers Emilee Helm agreed because that is achieved by building a user community around a game.