At the just past Game Developers Conference (GDC 2026), the Google DeepMind team brought its latest generative AI model-Genie 3. Although this technology aims to explore the boundaries of interactive content generation, the team admitted during the meeting that its current capabilities are quite limited: the interactive game world generated by Genie 3 will lose coherence after only a few minutes of running, with a large number of glitches and problems.
According to industry insiders who attended the GDC DeepMind special session, the gaming experience generated by Genie 3 is smooth in the first minute, but will "basically collapse" after that. This status quo is a clear indication that the technology is still a long way from replacing the jobs of actual game developers.
Google chose to display Genie 3 at GDC, an industry event, which is an intriguing move in itself. Because generative AI is seen as a technology that may impact employment, the game industry is generally cautious or even opposed to it. Therefore, DeepMind’s candid sharing at the conference also provided key background information for the outside world, especially investors: there is no need to worry too much about the disruptive impact Genie 3 will have on the industry structure.
However, from the perspective of technological development, the speed of progress is also worth noting. It is said that just a few months ago, the interactive world generated by Genie 3 could only last for a few seconds. It remains to be seen how far this technology can develop by 2027 if Google continues to invest and support it.