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。

Maybe ChatGPT has some pre-frontal cortex problems

19 点作者 solresol4 个月前

7 条评论

MisterKent4 个月前
This is a really odd way to test capabilities of an LLM. First, most photos of clocks are 10:10, since the training data for watches are usually set to 10:10 (in order to better sell watches etc).<p>Second, I don&#x27;t think the photo generation aspect of chat gpt is being marketed or presented as a problem solving AI.
chomp4 个月前
I like the part where the AI couldn’t be trusted to draw a clock, so we trusted it to psychoanalyze the incorrect clock
solresol4 个月前
I administered the CDT to ChatGPT and got Claude to diagnose what was wrong with the &quot;patient&quot; based on the results.<p>There are signs of pre-frontal cortex damage or early stage dementia.
评论 #42672602 未加载
pnm456784 个月前
Here&#x27;s the thing (which you probably knew going in).. Generative AI is quite well-known to be terrible at drawing specific times on clock faces.<p>This is down to the training data. It has been trained on a huge amount of images.<p>That includes advertising. For whatever reason, wrist watch manufacturers have a tendency to set watches to 10:10 in ads, almost without exception. Perhaps it&#x27;s just a nice-looking time, or it&#x27;s good for comparison purposes.<p>Simply Google &quot;wrist watch&quot; and you&#x27;ll see.<p>So, these generative models have a huge bias towards 10:10 on clock faces, because that&#x27;s what all the clocks they&#x27;ve been trained on look like.
airstrike4 个月前
FWIW, Claude 3.5 Sonnet got the SVG right on the first try: <a href="https:&#x2F;&#x2F;claude.site&#x2F;artifacts&#x2F;8dedf16e-b861-4497-96e2-872773d71baf" rel="nofollow">https:&#x2F;&#x2F;claude.site&#x2F;artifacts&#x2F;8dedf16e-b861-4497-96e2-872773...</a><p>Prompt was just &quot;create an svg of a clockface with the time being 10 past 11&quot;
pockybum5224 个月前
I love the concept of the article where one LLM can&#x27;t draw a simple clock but the other one can accurately diagnose medical conditions from a hypothetical drawn image.
batch124 个月前
It has sentience problems...