TE
科技回声
首页24小时热榜最新最佳问答展示工作
GitHubTwitter
首页

科技回声

基于 Next.js 构建的科技新闻平台,提供全球科技新闻和讨论内容。

GitHubTwitter

首页

首页最新最佳问答展示工作

资源链接

HackerNews API原版 HackerNewsNext.js

© 2025 科技回声. 版权所有。

The Fundamental Distinction Between Brains and Turing Machines (2002) [pdf]

4 点作者 maximgsaini超过 10 年前

1 comment

tgflynn超过 10 年前
This is a long verbose article that contains some confused ideas, however I think it also contains some very interesting ones which could serve as a basis for further consideration.<p>The two ideas I find most interesting here are:<p>A) True random numbers cannot be computed by a Turing machine whereas human minds may have access to them. The author does not address the fact that it is relatively easy to augment a TM with a physical RNG. I&#x27;m not aware of any evidence that such an augmented TM can solve problems a simple TM cannot.<p>B) Human minds seem to be able to always abstract to a higher level meta-language whereas programs&#x2F;TM&#x27;s seem to lack this ability. I have a feeling this issue may be key to the difficulty in producing machines that seem &quot;truly intelligent&quot; but I don&#x27;t currently have a clear grasp of this problem.<p>The author makes a great deal of use of the term &quot;consciousness&quot; without being very clear about what he means by it. He seems to be using it in the sense of what I would call &quot;true intelligence&quot; or strong AI rather than in the sense of &quot;capable of having subjective experiences&quot;.<p>I have generally tended to hold the following &quot;beliefs&quot;:<p><pre><code> 1) Algorithms cannot have subjective experiences (strongly held). 2) A strong AI algorithm could exist (somewhat weakly held). </code></pre> Issue (B) above leads me to question belief (2). Perhaps it will turn out that &quot;subjective experience&quot; is necessary for strong AI and hence that no strong AI algorithm exists ?
评论 #8370210 未加载