Ask HN: 神经多样性是否蕴藏着构建更优秀 AI 的秘密？
3 分•作者: EdDel•9 个月前
我对目前正在构建人工智能系统的人们的观点很感兴趣。特别是，你们在哪里看到我们对神经多样性和大脑生理学的理解影响了人工智能的开发方法？
作为背景，我是一名生物化学家/生物技术专家，并非深入的人工智能专家。我一直在探索从神经多样性中获得的经验教训如何帮助塑造人工智能的开发。
与其直接丢一个链接，这里有一个简短的TLDR：
* 不同的脑生理结构和由此产生的思维模式可能会暴露盲点并激发新的解决方案
* 人工智能的开发是否正在收敛于一套狭窄的优化策略？
* 人类的认知多样性可能指向互补的方法，从而扩大人工智能的能力范围
我写了一篇更长的文章，详细阐述了这个想法，并欢迎大家讨论这些想法是否、在哪里以及如何被实践。
全文链接：https://www.linkedin.com/pulse/could-neurodiversity-hold-clues-fixing-ais-jagged-edmund-delaney-3zpke
查看原文
I’m interested in views from people who are building AI systems today. In particular, where have you seen our understanding of neurodiversity and underlying brain physiology influence approaches to AI development?<p>For context, I’m a biochemist / biotechnologist by training and not a deep AI specialist. I’ve been exploring how lessons from neurodiversity might help shape AI development.<p>Rather than just drop a link, here’s a short TLDR:<p>• Different brain physiologies and the resulting thinking patterns can surface blind spots and spark novel solutions
• Is AI development converging on a narrow set of optimisation strategies?
• Human cognitive diversity might point to complementary approaches that expand the range of AI capabilities<p>I’ve written a longer piece that sets out the idea in more detail and would welcome discussion on whether, where, and how these ideas are being tried in practice.<p>Full article here: https://www.linkedin.com/pulse/could-neurodiversity-hold-clues-fixing-ais-jagged-edmund-delaney-3zpke