People keep talking about "AI alignment" when they haven't solved the "CEO alignment" problem, or the question "alignment with what, exactly?"
> The goal of Midjourney is to bring creative power to as many people as possible.
AI is unarguably powerful, even if we're still struggling to quantify that in terms of how "intelligent" it is, to what extent it's plagiarism, or reliability, etc. The consequence of power is necessarily interacting with real-world political power structures. There is no "neutral" option; AI will empower some and disempower others, who are going to object to that. As we're seeing in this thread, if you empower or disempower one side of a political dispute (making Xi art or not), there's going to be one side that gets mad at you.
Make cultural products, get culture war.
I do wonder if this is going to be regarded as the "Thomas Midgely" era of AI, where scientists from the 2050s struggle with the legacy of widespread AI "contamination" of everything in the data record by people who didn't acknowledge downsides.
> The goal of Midjourney is to bring creative power to as many people as possible.
AI is unarguably powerful, even if we're still struggling to quantify that in terms of how "intelligent" it is, to what extent it's plagiarism, or reliability, etc. The consequence of power is necessarily interacting with real-world political power structures. There is no "neutral" option; AI will empower some and disempower others, who are going to object to that. As we're seeing in this thread, if you empower or disempower one side of a political dispute (making Xi art or not), there's going to be one side that gets mad at you.
Make cultural products, get culture war.
I do wonder if this is going to be regarded as the "Thomas Midgely" era of AI, where scientists from the 2050s struggle with the legacy of widespread AI "contamination" of everything in the data record by people who didn't acknowledge downsides.