<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>ART: Automatic multi-step reasoning and tool-use for large language models · HackerPapers discussions</title><link>https://hackerpaper.kaipfister.de/p/2303.09014/</link><description>Collected HN comments and submissions linking this paper.</description><language>en</language><lastBuildDate>Mon, 02 Nov 2026 11:16:22 GMT</lastBuildDate><atom:link href="https://hackerpaper.kaipfister.de/p/2303.09014/feed.xml" rel="self" type="application/rss+xml" /><item><title>ART: Automatic multi-step reasoning and tool-use for large language models — HN submission</title><link>https://news.ycombinator.com/item?id=35460526</link><guid isPermaLink="false">urn:hackerpapers:submission:2303.09014:35460526</guid><pubDate>Wed, 05 Apr 2023 21:21:00 GMT</pubDate><description>&lt;p&gt;Show HN: Guesstimate – Generate a spreadsheet-like interface for any question&lt;/p&gt;&lt;p&gt;Hey HN, I built something to estimate things: https://guesstimate.ai Unlike ChatGPT, Guesstimate can do calculations. I asked both to calculate the time it&amp;#x27;d take an object to fall from the ISS to sea level (about 400km, assume no air resistance). ChatGPT, using a lot of words, did everything correctly right until the last step [1]. Guesstimate took a similar approach, but it got the right answer: https://guesstimate.ai/e/est114xrhnj1f6h9cm44 I think it&amp;#x27;s absurd that we&amp;#x27;re trying to teach LLMs basic arithmetic when there&amp;#x27;s a literal arithmetic processing unit right next to the GPU. You…&lt;/p&gt;&lt;p&gt;1 point; 0 discussion replies on HN. &lt;a href="https://hackerpaper.kaipfister.de/p/2303.09014/"&gt;Paper and source history&lt;/a&gt;. A submission is not an endorsement.&lt;/p&gt;</description><category>cs.CL</category></item><item><title>ART: Automatic multi-step reasoning and tool-use for large language models — mentioned by pinkcan</title><link>https://news.ycombinator.com/item?id=35281419</link><guid isPermaLink="false">urn:hackerpapers:mention:2303.09014:35281419</guid><pubDate>Thu, 23 Mar 2023 21:18:22 GMT</pubDate><description>&lt;p&gt;It&amp;#x27;s called ART - Automatic multi-step Reasoning and Tool-use https://arxiv.org/abs/2303.09014&lt;/p&gt;&lt;p&gt;&lt;a href="https://hackerpaper.kaipfister.de/p/2303.09014/"&gt;Paper and discussion history&lt;/a&gt;. A mention is not an endorsement.&lt;/p&gt;</description><category>cs.CL</category></item></channel></rss>
