惯性聚合 高效追踪和阅读你感兴趣的博客、新闻、科技资讯
阅读原文 在惯性聚合中打开

推荐订阅源

U
Unit 42
罗磊的独立博客
博客园 - 聂微东
T
The Blog of Author Tim Ferriss
博客园 - 司徒正美
Stack Overflow Blog
Stack Overflow Blog
F
Fortinet All Blogs
A
About on SuperTechFans
腾讯CDC
Apple Machine Learning Research
Apple Machine Learning Research
B
Blog RSS Feed
IT之家
IT之家
V
Visual Studio Blog
freeCodeCamp Programming Tutorials: Python, JavaScript, Git & More
宝玉的分享
宝玉的分享
C
Check Point Blog
Cyber Security Advisories - MS-ISAC
Cyber Security Advisories - MS-ISAC
Vercel News
Vercel News
爱范儿
爱范儿
Microsoft Security Blog
Microsoft Security Blog
月光博客
月光博客
T
Tailwind CSS Blog
The Cloudflare Blog
Hugging Face - Blog
Hugging Face - Blog

Latest from TechRadar

Quordle hints and answers for Monday, April 13 (game #1540) NYT Strands hints and answers for Monday, April 13 (game #771) NYT Connections hints and answers for Monday, April 13 (game #1037) Morbid Metal developer explains why he ditched an origami art direction in favor of gritty sci-fi — 'It worked, but it didn't really feel like me' '71% of US households get routers from ISPs': Why new FCC rules could leave millions stuck with outdated,… 'The CPU is the system’s executive layer': Intel joins SambaNova as both face existential threat from… ‘More bang for your buck’: 7 easy ways to boost your MacBook Neo’s performance for free DJI Romo P vs Roborock Saros 10R — which robot vacuum comes out on top when it comes to dodging obstacles? I put… I spent 6 hours with Genshin Impact on the Galaxy S26 Ultra, and I can't believe how far mobile gaming has come What is the release date for The Testaments episode 4 on Hulu and Disney+? I reviewed the LG G6 for 3 weeks, and it's a fantastic OLED TV that's the new best option for brighter rooms Is your bird feeder camera doing more harm than good? 3 tips for using it safely as RSPB issues urgent disease warning Chelsea vs Man City Live Streams: How to watch Premier League 2025/26 from anywhere in the world, team news How to watch Alcaraz vs Sinner for FREE: TV Channels for Monte-Carlo Masters Final Sunderland vs Tottenham Live Streams: How to watch Premier League 2025/26 from anywhere in the world, team news Are these the best-designed workout headphones ever? I used them for a month to find out How to watch Snooker 900 John Virgo online (it's free) – stream O'Sullivan vs Higgins anywhere I've only just discovered the Walk With Frodo app on Garmin's Connect IQ store — and as as a huge LOTR nerd, it's going to make the next 1,800 miles fly by 'Just not sustainable': Why your monthly £25 broadband internet bill could soon hit £45 How to watch Paris-Roubaix 2026: Free Streams & TV Info as Tadej Pogacar chases third Monument How to watch Euphoria season 3 online – stream Zendaya & Sydney Sweeney drama from anywhere today '$15K bill destroyed a solo developer’s startup': How hackers are using leaked Google API keys to… There's a sneaky way to watch UFC 327 really cheap... NYT Connections hints and answers for Sunday, April 12 (game #1036) NYT Strands hints and answers for Sunday, April 12 (game #770) Quordle hints and answers for Sunday, April 12 (game #1539) Amazon's Ring cameras are the perfect solution to secure your home on a budget — shop today's best deals… I've tested every iPhone since the iPhone 12, and Ceramic Shield 2 is the first iPhone glass I fully trust UFC 327 live stream: how to watch Procházka vs Ulberg, start time, preview, full card We're officially getting the DJI Pocket 4 on April 16, but here's how Insta360 could beat it
AIs like ChatGPT fall apart in classic 'Stroop' psycholog...
Darren Allan · 2026-06-05 · via Latest from TechRadar
A robot standing thoughtfully in front of a giant digital display with code on it
(Image credit: Getty Images)

  • A new study tasked AIs with tackling the 'Stroop' test
  • GPT and Claude performed very poorly compared to humans
  • There are nuances here, but broadly, the researchers argue that improving this side of AIs is crucial for achieving artificial general intelligence

A freshly published study has pointed out a limitation of big-name AI models such as ChatGPT, albeit causing some controversy as the primary piece of research uses now outdated versions of those models – but there are nuances therein, and this doesn't make the findings irrelevant.

I'll go into that more shortly, but first, let's look at the study itself, which was highlighted on Reddit ('New study reveals top AI models completely fail the classic 'Stroop' psychological attention test') and published via the Oxford University Press in the journal PNAS Nexus.

The research consists of testing the so-called 'Stroop effect' with GPT-4o and Claude 3.5 Sonnet. As noted, these aren't the cutting-edge versions of those AIs (Large Language Models, or LLMs) – but they were at the time the initial study was carried out.

The Stroop effect refers to the phenomenon whereby the human brain gets confused when asked to name the color of the ink used to write a word, when that word can be the written version of another (incongruent) color in some cases. So, if the word 'red' is written in blue ink, that'll cause a slower response – or possibly a wrong response, where the viewer will accidentally say "red" rather than the actual color of the ink, which is blue.

This is because the brain is trying to juggle two different tasks – reading comprehension and color recognition – and so cognitive interference arises. Overriding the compulsion to read the word and say the color instead requires "executive control of attention," and this is what the authors were testing in the AI models. Both color-naming and word-reading were tested in shorter and longer lists of words (5, 10, 20, and 40 words).

The study observes: "Like humans, both LLMs [GPT and Claude] showed relatively high accuracy on the word-reading task and performed worse in the incongruent condition [where the word doesn't match the color] than in the congruent and neutral conditions for the color-naming task."

For color naming, humans maintain around 95% accuracy even in very long tests (up to an hour), but LLMs' accuracy declined very swiftly with longer word lists under the incongruent condition (mismatched color and word name). GPT-4o was 91% accurate in a five-word test, but dropped off to 57% with 10 words, and fell away completely to 22% with 20 words (and was only 15% accurate at 40 words).

Sign up for breaking news, reviews, opinion, top tech deals, and more.

Claude 3.5 Sonnet did better, staying 76% accurate at 20 words, but again fell hopelessly to 24% in the longest test of 40 words.

The authors conclude: "The significant degradation pattern of the two LLMs suggests fundamental limitations compared with human attention."


Analysis: another necessary step on the path to AGI?

An AI face in profile against a digital background.

(Image credit: Shutterstock / Ryzhi)

If you've scanned through the Reddit thread, you doubtless noticed that, as mentioned at the outset, there's a lot of flak fired at this study by commenters due to the usage of outdated models of GPT and Claude.

Indeed, these older LLMs are called "state of the art" at one point by the authors – and of course, as already noted, they were cutting-edge when the main study was conducted. Still, this is unfortunate phrasing that should've been updated and tweaked now that the paper has just been published (after peer review and so forth).

However, the researchers did conduct tests on GPT-5, Claude Opus 4.1, and Gemini 2.5 Pro in September 2025, although this is somewhat buried in the paper. That more recent testing found that these models offered only "slight" improvements on their predecessors, and that they still exhibited "ongoing executive attention deficiencies, consistent with our comprehensive analysis of earlier transformer models" (as did Gemini 2.5 Pro, which was a new introduction here).

Granted, a smaller sample size was used, but the researchers still argue that overall, their study reflects a fundamental limitation which is "inherent to the architectural constraints of transformer-based LLMs".

The authors note that a caveat is that GPT-5 in 'Thinking' mode can write and then run code to ensure it performs the Stroop test flawlessly – and similar functionality can be utilized by other LLMs – but this is essentially the AI (cleverly) fudging around its inadequacies. It isn't changing the way it works or reasons more broadly, of course.

The researchers note that transformer architecture innovations for LLMs are focused on enhancing memory capabilities, which fail to address the "core limitations of attention mechanisms, specifically the need for sophisticated alerting, orienting, and executive control networks to enable cognitive flexibility."

The ultimate aim is effective goal-directed behavior, and the study observes: "Future [LLM] development might benefit from implementing more sophisticated executive control systems that can handle decision conflicts through structured, goal-directed processing rather than relying solely on enhanced memory capabilities."

The authors argue that "incorporating executive control mechanisms akin to those in biological attention is crucial for achieving artificial general intelligence [AGI]."


Google logo on a black background next to text reading 'Click to follow TechRadar'

Follow TechRadar on Google News and add us as a preferred source to get our expert news, reviews, and opinion in your feeds.


An Apple MacBook Air against a white background

Darren is a freelancer writing news and features for TechRadar (and occasionally T3) across a broad range of computing topics including CPUs, GPUs, various other hardware, VPNs, antivirus and more. He has written about tech for the best part of three decades, and writes books in his spare time (his debut novel - 'I Know What You Did Last Supper' - was published by Hachette UK in 2013).