Blog

The Oldest AI Blog in the Czech Republic

We have been writing about artificial intelligence since 2017. 1000+ articles, thousands of pages of thoughts, experiments, and reflections. No sensationalism, no ads.

Year:All2026 212025 1012024 832023 1222022 1102021 1362020 1242019 1652018 1052017 55
Source:All 1022Blog 544Facebook 445LinkedIn 23
Category:Education 3Computer Vision and Image 3Media and Interviews 1Lectures and Events 3Ethics and Philosophy 3Practical Applications and Business 2Society and the Future of AI 1Reinforcement Learning and Robotics 3Tools and Frameworks 1

Tag Filter: Claude Opus 4.7 × cancel

▸Browse by Topics

Topics

aiFacebookfacebookEducationNewsAILanguage ModelsComputer Vision and ImageLectures and EventsMedia and Interviewsartificial intelligenceExperimentsEthics and PhilosophyEventsMachine LearningPractical Applications and BusinessAI IntelligenceSociety and the Future of AIStudyText and SequencesAlpha IndustriesLecturesInterviewsJan TylArtificial IntelligenceTools and FrameworksApplicationsReinforcement LearningOpenAIReinforcement Learning and RoboticsGooglePythonComputer VisionDigital WriterReinforced Learning and RoboticsNeural NetworkDatasetsUnsupervised LearningDita MalečkováAI Art of Painting and Creation

Showing 2 of 2 articles

One Question vs. Ten Major Benchmarks: How Well Did the Mini IQ Test Predict Reality?
4 min

One Question vs. Ten Major Benchmarks: How Well Did the Mini IQ Test Predict Reality?

A few days ago I gave the latest AI models just one single question — no benchmark, no scoreboard, just one intellectual trap. Now it's time to compare the result with the major public benchmarks that have appeared since. GPT-5.5 leads, Claude Opus 4.7 follows closely, Gemini 3.1 Pro was the biggest outlier in my test, and DeepSeek V4 Pro came lowest. A question worth asking every model — before you trust it.

Read
One Question Instead of Ten Benchmarks: A Mini IQ Test for the Latest AI Models
4 min

One Question Instead of Ten Benchmarks: A Mini IQ Test for the Latest AI Models

I took the latest top AI models and instead of endless benchmarks, posed them a single 'genius' question: uncover a hidden rule, calculate a new case, acknowledge ambiguity, propose a falsification test, and critique their own solution. The result? The top performers today score roughly in the 120–135+ IQ impression range — but the difference lies not in what the models know, but in how clearly they can think under pressure.

Read