The era of heavily subsidised AI compute is ending – but the experiences of developers who have worked with it suggest limited productivity gains that will make higher costs hard to justify.
The games industry has been abuzz with talk of Generative AI in recent years, with opinions ranging from enthusiastic support to cautious interest and ethical concerns. However, the actual utility of AI in game development has remained uncertain, with a lack of concrete data on its potential impact. Productivity gains from AI tools have been noted, but they vary depending on the task and require significant human oversight.
Recent experiences across various sectors using AI tools have shed light on the inconsistent nature of productivity gains, leading to a more realistic outlook on the technology. The rising costs of AI implementation, shifting from subsidised to token-based billing, have prompted companies to reassess the long-term value of these tools and the need for tangible returns on investment.
As the industry grapples with the actual benefits of AI in practical use cases, it is becoming evident that while there are incremental productivity improvements, the initial grand promises of cost-effective, transformative AI solutions may not materialise as expected.
