by 
If you've been following the MiSTer FPGA scene recently, you'll no doubt be aware of the sharp rise in AI-assisted cores for the platform.
We're currently witnessing a flurry of new cores that are apparently based on MAME rather than the original arcade boards, which, in the past, have been carefully deconstructed to produce the super-accurate FPGA cores that have made MiSTer so popular.
What's happening at the moment is that developers are leveraging AI and existing MAME data to produce MiSTer cores, flooding the web with what could be termed "easy wins" that may contain multiple inaccuracies. Annoyingly, those who wish to avoid these cores are finding it hard to do so, as they're being added to the 'update_all' script without thorough review or clear AI-generated labelling.
While some would argue that more cores are better for the player as they offer more choice, FPGA developer Jotego – who is one of the leading lights in the world of MiSTer FPGA and has even worked on the upcoming Neo Geo AES+ – has penned a short-but-sweet rebuttal of the approach on social media:
Jotego's process for creating an FPGA core involves either going hands-on with the original arcade hardware to create accurate schematics, which then allow him to produce a core that simulates the original board's performance at the hardware level. This is incredibly time-consuming, which is why he and many other FPGA developers have Patreon accounts so people can support their work.
Jotego's post has attracted a lot of support within the community:
However, not everyone agrees AI-made cores are a bad thing – in fact, they see it as the next step in the evolution of FPGA gaming, where AI is used to speed up development and save time:
Where do you stand on this issue? Are AI-generated FPGA cores a bad thing if they give us access to more games, or do they poison the well of FPGA development, forcing those who do super-accurate work out of the community? Vote in the poll below and share your thoughts with a comment.