5 ms·
I really think that intelligence augmentation (maybe with GitHub Copilot as a notable exception) is an undersold aspect of recent advances in generative AI. Am
by catlifeonmars 5y ago
I really think that intelligence augmentation (maybe with GitHub Copilot as a notable exception) is an undersold aspect of recent advances in generative AI. Am I wrong in thinking that IA (as opposed to autonomous AI) has huge potential today to change the way we interact with and create things?
- smoldesu 5y ago> Am I wrong in thinking that IA (as opposed to autonomous AI) has huge potential today to change the way we interact with and create things? There is potential, but there's also roadblocks. Just to name a few: - The cost of training and interacting with these models is ludicrous - AI is ultimately constrained by it's training data - Designing models that people can get good results from is hard. It requires intense cherry-picking that's ultimately opinionated, and therefore flawed. Maybe the "IA" you're looking for it "intelligence automation" rather than "intelligence augmentation". Too often do we forget that AI only understands that which it has already seen; there really is nothing new under the sun.