TE
TechEcho
Home24h TopNewestBestAskShowJobs
GitHubTwitter
Home

TechEcho

A tech news platform built with Next.js, providing global tech news and discussions.

GitHubTwitter

Home

HomeNewestBestAskShowJobs

Resources

HackerNews APIOriginal HackerNewsNext.js

© 2025 TechEcho. All rights reserved.

Microsoft Says New AI Shows Signs of Human Reasoning

47 pointsby danielmorozoffabout 2 years ago

11 comments

gumballindieabout 2 years ago
&gt; “When we see a complicated system or machine, we anthropomorphize it; everybody does that — people who are working in the field and people who aren’t,”<p>That and marketing. Microsoft wants all of them sweet £ flowing in and thus over selling what their ai does. Sad part is that many will fall for it.
评论 #35961074 未加载
评论 #35961437 未加载
herskoabout 2 years ago
Not surprising at all. I throw a wall of code at GPT-4, write a paragraph of what i want it to do (I couldn&#x27;t spend the time to figure out a good solution), and within a minute I had a great functioning solution perfectly integrated into my existing code.<p>The surprising thing is not that it was able to code a good solution, it&#x27;s that it was able to understand the specifications of how i wanted it to work and how it should be integrated into the existing code.
评论 #35965370 未加载
ActorNightlyabout 2 years ago
I mean, ability to parse information, compress it into the latent space, and then regenerate it is what humans do, but I wouldn&#x27;t call that reasoning.
评论 #35962108 未加载
ilakshabout 2 years ago
The problem is that people aren&#x27;t able to use language consistently and precisely, and are conflating all kinds of human&#x2F;animal characteristics together.<p>GPT-4 really does do human-like reasoning. And it&#x27;s clearly quite general purpose within its limitations.<p>But it doesn&#x27;t have many other aspects of humans&#x2F;animals such as self-direction, high bandwidth senses, a stream of subjective experience, emotions, certain types of adaptivity, etc. It&#x27;s not alive and it&#x27;s not a digital person.<p>But people aren&#x27;t able to separate all of those different things, so they can&#x27;t admit that it has any kind of useful intelligence, because for them that means it&#x27;s a digital person.<p>Actually it&#x27;s even dumber, many people jump to assuming that it goes straight to god-like superintelligence if it has any intelligence at all. Which is another reason they can&#x27;t admit there is any reasoning, because that would mean the end of the world or something.
评论 #35974325 未加载
seydorabout 2 years ago
Paper was released and discussed in april 13. NYtimes dismissed AI for way too long and now trying to catch up. This is old news
alienalpabout 2 years ago
Last thing I want from an AI is to reason like humans. My one of the biggest expectation from AI is to decrease dependence on human reasoning.
skilledabout 2 years ago
<a href="https:&#x2F;&#x2F;archive.is&#x2F;pSRth" rel="nofollow">https:&#x2F;&#x2F;archive.is&#x2F;pSRth</a>
neomabout 2 years ago
Highly highly recommend listen to this lecture by Sebastien Bubeck (quoted in the nyt article), it&#x27;s extremely interesting:<p><a href="https:&#x2F;&#x2F;youtu.be&#x2F;qbIk7-JPB2c" rel="nofollow">https:&#x2F;&#x2F;youtu.be&#x2F;qbIk7-JPB2c</a>
catocabout 2 years ago
Any sufficiently advanced technology is indistinguishable from mAGIc
评论 #35966604 未加载
kapperchinoabout 2 years ago
Wasn’t there a paper on how gpt4 showed signs of early agi? Is this referencing that?
评论 #35966375 未加载
评论 #35963868 未加载
coolhand2120about 2 years ago
If anyone has read Blindsight by Peter Watts the hero encounters an alien that is pure reflex, no cognition. The alien is compared to the Chinese room thought experiment (<a href="https:&#x2F;&#x2F;en.wikipedia.org&#x2F;wiki&#x2F;Chinese_room" rel="nofollow">https:&#x2F;&#x2F;en.wikipedia.org&#x2F;wiki&#x2F;Chinese_room</a>). ChatGTP is the Chinese room. As you peel the onion you discover that some of the arguments around thought are really just semantics.<p>My personal experience shows me that ChatGTP 4 is something of a fool with a big fast reference book. I typically use it for programming, but I also study music theory. When I was exploring one of the more complex (to me) ideas I asked chat GTP 4 to explain it to me and provide examples.<p>After previously asking about Cmaj to Emaj, I asked &quot;can you give me an example of the H transformation?&quot;<p>ChatGPT: &quot;Sure! To demonstrate the Hexatonic pole &#x27;H&#x27; transformation using chords similar to C major and E major, let&#x27;s first introduce the concept of the hexatonic system. A hexatonic system consists of six notes, and two triads can be extracted from it. In the context of the &#x27;H&#x27; transformation, these two triads are related by a common tone, and the other two notes in each triad are a tritone apart.<p>Let&#x27;s use the C major triad (C, E, G) as a starting point. We&#x27;ll build a hexatonic collection containing this triad. One possible hexatonic system containing C major is the one with the notes C, Db, E, F#, G, and A. From this collection, we can extract two triads:<p>C major (C, E, G) F# major (F#, A, Db) Notice that the C major and F# major triads share a common tone (C&#x2F;Db enharmonically), and the other notes in each triad (E and G in C major, and F# and A in F# major) are a tritone apart. In this case, the &#x27;H&#x27; transformation links the C major and F# major triads.<p>So, the Hexatonic pole &#x27;H&#x27; transformation example using chords similar to C major would be transitioning from C major to F# major.&quot;<p>Me: &quot;I&#x27;m pretty sure F# major isn&#x27;t F#, A, Db&quot;<p>ChatGPT: &quot;You are correct, and I apologize for the mistake. The F# major triad should be F#, A#, and C#. Let&#x27;s re-examine the &#x27;H&#x27; transformation with the correct F# major triad...&quot;<p>After about 10 more prompts trying to extract information from it I ended up needing to be very literal to get results out of it:<p>Me: &quot;the definition is &quot;The &#x27;H&#x27; transformation takes a major chord of root X to the minor chord whose root is eight semitones above X.&quot; can you give me a VALID example of this&quot;<p>ChatGTP: &quot;I apologize for the confusion in my previous responses. Based on the definition you provided, the &#x27;H&#x27; transformation takes a major chord of root X to the minor chord whose root is eight semitones above X. Here&#x27;s a valid example using this definition:<p>Let&#x27;s start with a C major triad (C, E, G). To find the target chord, we need to count eight semitones above the root note (C):<p>C# &#x2F; D♭ D D# &#x2F; E♭ E F F# &#x2F; G♭ G G# &#x2F; A♭ So, the root of the target chord is A♭. Now, we form an A♭ minor triad (A♭, C♭, E♭).<p>The &#x27;H&#x27; transformation in this example would be transitioning from C major (C, E, G) to A♭ minor (A♭, C♭, E♭).&quot;<p>Nice! This time it was correct (I think!). But it required that I feed it the exact definition and only when I saw it doing the work did I have any confidence that it was correct. In the end I did learn how it these concepts work, I think I did at least. Overall it&#x27;s not a great teacher, half of the time I had to teach it and correct it when it seemed to be very confident in its answers, and even more confident that my corrections were correct, which I&#x27;m not 100% sure of.<p>One observation is that ChatGTP4 didn&#x27;t have any idea what it was talking about the entire time. It was just putting related information together. That observation was based on the notes it kept claiming were related to each other via interval degrees. The intervals it was using were always wrong, not even close, just pure nonsense. Only when told exactly how to count did it get it right and only when the counting was part of the response. Now mind you, the things it was getting wrong were fundamental music theory 101 stuff, but it was making these fundamental mistakes in the middle of a explanation of a very complex topic. I don&#x27;t know what it all means, but I wouldn&#x27;t trust it to fly or build an airplane, or even boil water now that I think about it. How would you know when it goes dumb?
评论 #35967496 未加载
评论 #35967308 未加载