Hidden Forces feed 2024年07月17日
David Weinberger | Complex Systems, Inexplicable Models, and the Future of Prediction
index_new5.html
../../../zaker_core/zaker_tpl_static/wap/tpl_guoji1.html

 

本期节目探讨了预测科学的演变及其未来。主持人与哲学家David Weinberger讨论了从古代哲学家到现代机器学习算法的预测方法,并分析了这些方法的优缺点以及对人类认知的影响。

🤔 从古代哲学家到现代机器学习算法,预测科学经历了漫长的演变。古代哲学家和数学家建立了经典的预测模型,这些模型依赖于对现象的解释,并基于这些解释进行预测。

🤖 现代机器学习算法则不同,它们能够提供比传统模型更准确的预测,但往往无法解释其预测背后的原因。这使得机器学习算法成为“黑箱预言机”。

🤯 这种新的预测方法挑战了传统的科学方法,即通过可证伪的预测来验证理论模型。机器学习算法的出现引发了对科学本质的思考:如果机器能够提供更准确的预测,但我们无法理解其背后的逻辑,我们是否应该接受这些预测?

🤔 随着机器学习算法的应用范围不断扩大,我们是否会进入一个可预测的世界?机器是否会成为新的“德尔菲神谕”?

🧐 预测科学的演变也引发了对人类意识的思考。人类意识是无法通过科学方法完全解释的,而机器学习算法的出现是否会重新开启人类对意识的探索?

In Episode 87 of Hidden Forces, Demetri Kofinas speaks with philosopher David Weinberger about the science of prediction, its evolution, and its future.

The two begin by exploring classical approaches developed by early philosophers and mathematicians in the ancient world and upon which advancements were later made by enlightenment thinkers and experimental scientists.   

The models developed in this tradition have, until now, provided explanations for phenomena, which are used to make predictions about the future states or trajectories of these and other phenomena that adhere the same laws of action or motion.   

What is new today is the evolution of what are known as “machine learning algorithms,” many of which provide superior predictions to those generated by conceptual or working models, but which often times cannot provide explanations for these predictions. They are, in this sense, block-box oracles.   

This represents a fundamental break with the sort of epistemological approach taken by the ancient Athenian philosophers who demanded that beliefs be justified by reasoned arguments or those of empirical scientists who relied upon falsifiability of testable hypotheses. In other words, whereas traditional approaches to science have necessitated the development of theoretical models of the world that can be tested empirically through the act of making falsifiable predictions, these new approaches are capable of generating predictions without a means by which to understand the causes at play.    

What are the implications of this new science? If predictions provided by highly intelligent machines become consistently more accurate across all domains of study, would we prefer to accept these inexplicable solutions over less accurate ones whose methodology we understand? At the limit, if we were to implement every prediction of every MLA, would we arrive at a fated, perfectly knowable world? If machines become the equivalent of Delphi’s Oracle, what will be the value of doing science? The scientific method, after all, is the means by which we have been able to navigate and understand the material world, in material terms. Does this re-open humanity’s door to the preoccupation with the mystery of conscious experience, which cannot be explained through the scientific method of objective, empirical analysis?  

These are the questions we explore in this week’s episode with David Weinberger and Demetri Kofinas.

Producer & Host: Demetri Kofinas

Editor & Engineer: Stylianos Nicolaou

Subscribe & Support the Show at http://patreon.com/hiddenforces

Join the conversation on Facebook, Instagram, and Twitter at @hiddenforcespod

Fish AI Reader

Fish AI Reader

AI辅助创作,多种专业模板,深度分析,高质量内容生成。从观点提取到深度思考,FishAI为您提供全方位的创作支持。新版本引入自定义参数,让您的创作更加个性化和精准。

FishAI

FishAI

鱼阅,AI 时代的下一个智能信息助手,助你摆脱信息焦虑

联系邮箱 441953276@qq.com

相关标签

预测科学 机器学习 科学方法 人类意识
相关文章