We've been experimenting with the Chinese open source models....one interesting thing we've found is that, when installed and run on a local server, none of them are censored. Asked about Tiananmen 1989, they all return info about the student protests and crackdown. This implies that the censorship is happening downstream of the models, like a black mirror of western AI system's safety module to filter hallucinations. It also makes sense if you consider that everyone is essentially using the same training data and all the foundation models, Chinese or US, are inevitably heading towards commoditisation.
unsub'd
Fabulous. Easier than culling my subs list myself
We've been experimenting with the Chinese open source models....one interesting thing we've found is that, when installed and run on a local server, none of them are censored. Asked about Tiananmen 1989, they all return info about the student protests and crackdown. This implies that the censorship is happening downstream of the models, like a black mirror of western AI system's safety module to filter hallucinations. It also makes sense if you consider that everyone is essentially using the same training data and all the foundation models, Chinese or US, are inevitably heading towards commoditisation.