https://semanticmachines.fyiPrompts, model comparisons and examples of AI analysis.https://semanticmachines.fyi/explainershttps://semanticmachines.fyi/explainersIn-depth articles that unpack complex AI concepts for researchers, educators, and curious minds. No jargon required.https://semanticmachines.fyi/workshopshttps://semanticmachines.fyi/workshopsStructured learning paths that take you from principles to practical application. Each workshop combines explainers and hands-on activities.https://semanticmachines.fyi/activitieshttps://semanticmachines.fyi/activitiesHands-on exercises to explore AI tools, test capabilities, and build practical skills. Filter by type or keyword to find what you need.https://semanticmachines.fyi/evalshttps://semanticmachines.fyi/evalsPrompts for testing AI models, with notes on what to look for in the answer. Copy any of them.https://semanticmachines.fyi/benchmarkshttps://semanticmachines.fyi/benchmarksPublished results from running the same prompts on different AI models. Each one opens its own reader.https://semanticmachines.fyi/benchmarks/bicyclehttps://semanticmachines.fyi/benchmarks/bicycleEach model was asked to make an SVG of a bicycle. Three runs, newest first.https://semanticmachines.fyi/benchmarks/vibe-run-july-2026https://semanticmachines.fyi/benchmarks/vibe-run-july-202671 prompt results from one run on one model, Gemini 3 Flash Preview, on 16 July 2026. Responses are shown as generated; no scores were assigned.https://semanticmachines.fyi/benchmarks/gemma-videohttps://semanticmachines.fyi/benchmarks/gemma-videoTwo videos were analysed by Gemma 4 12B running locally: a screencast and a meeting clip. Only the written results are shown here, not the videos or stills from them.https://semanticmachines.fyi/benchmarks/czech-transcriptionhttps://semanticmachines.fyi/benchmarks/czech-transcriptionFive speech-to-text services transcribed the same Czech recording. Date not recorded.https://semanticmachines.fyi/benchmarks/writinghttps://semanticmachines.fyi/benchmarks/writingCompare Czech poetry, Wodehouse-style prose, an academic parody about a strategic icecream stockpile, a nursery rhyme rewritten in the style of ten English poets and an English translation of a Czech poem by Jaroslav Seifert.https://semanticmachines.fyi/abouthttps://semanticmachines.fyi/aboutSemanticMachines is a repository of knowledge about AI for academic research, a source of updates on the latest developments, and a place to create interactive learning experiences.https://semanticmachines.fyi/explainers/prose-to-structurehttps://semanticmachines.fyi/explainers/prose-to-structureWhy structured outputs like tables and lists are often more powerful than prose for learning.https://semanticmachines.fyi/explainers/training-model-meaningshttps://semanticmachines.fyi/explainers/training-model-meaningsA non-technical guide to the 5 layers of AI model "training" and why the distinctions matter for decision makers.https://semanticmachines.fyi/explainers/rag-and-contexthttps://semanticmachines.fyi/explainers/rag-and-contextUnderstand the difference between the Context Window and RAG (Retrieval Augmented Generation), and why it matters for reliability.https://semanticmachines.fyi/explainers/academic-writing-reflectionhttps://semanticmachines.fyi/explainers/academic-writing-reflectionA collection of questions to guide a reflection on the impact of Large Language Models on academic writing.https://semanticmachines.fyi/workshops/ai-for-researchers-workshophttps://semanticmachines.fyi/workshops/ai-for-researchers-workshopA hands-on workshop designed to move researchers from basic principles to practical application of Large Language Models.https://semanticmachines.fyi/activities/act-semantic-baselinehttps://semanticmachines.fyi/activities/act-semantic-baselineThis activity explores how Large Language Models (LLMs) predict text based on meaning (semantics) rather than just statistical frequency (n-grams).https://semanticmachines.fyi/activities/act-model-knowledgehttps://semanticmachines.fyi/activities/act-model-knowledgeUnderstand the difference between model sizes, versions, and providers regarding factual accuracy. You will compare different Google models and then contrast them with other providers.https://semanticmachines.fyi/activities/act-small-modelshttps://semanticmachines.fyi/activities/act-small-modelsExplore the capabilities of 'open' or small models that could run on local devices, comparing their performance to massive frontier models.https://semanticmachines.fyi/activities/act-reasoninghttps://semanticmachines.fyi/activities/act-reasoningExperiment with 'System 2' thinking (slow, deliberate reasoning) versus 'System 1' (fast, intuitive prediction) using thinking models.https://semanticmachines.fyi/activities/act-tool-use-mathhttps://semanticmachines.fyi/activities/act-tool-use-mathLLMs are historically bad at math and counting because they predict tokens. In this activity, you will see how using 'tools' (code execution) solves this problem.https://semanticmachines.fyi/activities/act-data-analysishttps://semanticmachines.fyi/activities/act-data-analysisUse the LLM as a data analyst by having it generate mock data and then write code to process and analyze that data.https://semanticmachines.fyi/activities/act-multimodalhttps://semanticmachines.fyi/activities/act-multimodalInteract with the model using inputs other than text, such as video files or screenshots, to test its visual understanding.https://semanticmachines.fyi/activities/act-mpi-explorerhttps://semanticmachines.fyi/activities/act-mpi-explorerExplore how to understand the contribution of different aspects of a model's capabilities using the MPI Explorer tool.https://semanticmachines.fyi/activities/act-long-contexthttps://semanticmachines.fyi/activities/act-long-contextTest the model's ability to 'hold' massive amounts of information in memory at once. If you need sample data, use Semantic Machines prompts .https://semanticmachines.fyi/activities/act-search-groundinghttps://semanticmachines.fyi/activities/act-search-groundingConnect the model to the internet to fetch up-to-date information and compare results with and without grounding.https://semanticmachines.fyi/activities/act-visualizing-contexthttps://semanticmachines.fyi/activities/act-visualizing-contextUnderstand how the model 'sees' conversation history and data using the Model Context Explorer tool.https://semanticmachines.fyi/activities/act-lit-search-comparisonhttps://semanticmachines.fyi/activities/act-lit-search-comparisonCompare 'Raw' models versus 'Reasoning' models for academic research tasks.https://semanticmachines.fyi/activities/act-specialized-toolshttps://semanticmachines.fyi/activities/act-specialized-toolsUse tools built specifically for research (Orchestration layers) rather than raw models to conduct a literature review.https://semanticmachines.fyi/activities/act-building-apphttps://semanticmachines.fyi/activities/act-building-appUse the LLM as a 'Tool Maker' to create software for you. For examples and guides, refer to https://edutools.fyi .https://semanticmachines.fyi/activities/cognitive-baselinehttps://semanticmachines.fyi/activities/cognitive-baselineBefore using AI for complex tasks, it is crucial to establish a "cognitive baseline"—an understanding of what the raw model can and cannot do without tools (like search or code execution) or specific prompting strategies.https://semanticmachines.fyi/activities/context-retrieval-activityhttps://semanticmachines.fyi/activities/context-retrieval-activityModern models have massive context windows (up to 2 million tokens). This activity investigates how models handle large amounts of data and external information, and where the limits of "retrieval" lie.https://semanticmachines.fyi/activities/deep-researchhttps://semanticmachines.fyi/activities/deep-researchDeep Research is a type of AI agent that takes the interaction with an LLM beyond what can be achieved by a single prompt.https://semanticmachines.fyi/activities/vocab-thinkinghttps://semanticmachines.fyi/activities/vocab-thinkingA short reflection activity that sets out basic terms for thinking about AI, such as large language model and context window.https://semanticmachines.fyi/activities/image-generationhttps://semanticmachines.fyi/activities/image-generationMost LLMs do not generate images directly; they call other models. These models have specific limitations and control issues.https://semanticmachines.fyi/activities/academic-writinghttps://semanticmachines.fyi/activities/academic-writingTo stop assuming answers are obvious and investigate gaps in knowledge regarding AI in academia.https://semanticmachines.fyi/evals/introhttps://semanticmachines.fyi/evals/introPrompts to evaluate frontier capabilities of Large Language Models. I use a selection of these prompts to evaluate a new model on the limits of performance.https://semanticmachines.fyi/evals/spatial-recognitionhttps://semanticmachines.fyi/evals/spatial-recognitionAll models fail at these tasks most of the time.https://semanticmachines.fyi/evals/spatial-relationshipshttps://semanticmachines.fyi/evals/spatial-relationshipsThis prompt asks the model to come up with contexts. Come up with contexts in which the sentence The cathedral is in front of the statue makes sense.https://semanticmachines.fyi/evals/visual-reasoninghttps://semanticmachines.fyi/evals/visual-reasoningA prompt for testing AI language models: Visual Reasoning.https://semanticmachines.fyi/evals/coding-capabilityhttps://semanticmachines.fyi/evals/coding-capabilityThis single prompt tests the models knowledge of readability formulas, its ability to implement them and output enough code to achieve this.https://semanticmachines.fyi/evals/metalanguage-anaphorahttps://semanticmachines.fyi/evals/metalanguage-anaphoraThese two prompts test the ability to disambiguate anaphora. The first one is a single sentence and the second one expands.https://semanticmachines.fyi/evals/writing-lengthhttps://semanticmachines.fyi/evals/writing-lengthMost models cannot output text longer than about 3,000 words (even if the actual limit is higher).https://semanticmachines.fyi/evals/multilingual-germanhttps://semanticmachines.fyi/evals/multilingual-germanThe correct translation is “You do not have to understand life” even though the literal words say “You must not understand life”.https://semanticmachines.fyi/evals/multilingual-czech-metalanguagehttps://semanticmachines.fyi/evals/multilingual-czech-metalanguageNapiš 10 vět, které ilustrují různá použití slova totiž. Write 10 sentences illustrating the use of the word totiž.https://semanticmachines.fyi/evals/multilingual-czech-morphologyhttps://semanticmachines.fyi/evals/multilingual-czech-morphologyThis requires a high level of metalinguistic awareness and ability to maintain this.https://semanticmachines.fyi/evals/multilingual-culturalhttps://semanticmachines.fyi/evals/multilingual-culturalA prompt for testing AI language models: Multilingual: Cultural Background.https://semanticmachines.fyi/evals/multilingual-idiomshttps://semanticmachines.fyi/evals/multilingual-idiomsA prompt for testing AI language models: Multilingual: Idiom Testing.https://semanticmachines.fyi/evals/small-models-knowledgehttps://semanticmachines.fyi/evals/small-models-knowledgeA sequence of short prompts to test knowledge, coherence and instruction following of small models.https://semanticmachines.fyi/evals/small-models-multilingualhttps://semanticmachines.fyi/evals/small-models-multilingualThis prompt tests multilingual performance linked to knowledge. Even models of 14b will struggle with this - even if they can perform relatively will in Czech.https://semanticmachines.fyi/evals/creative-writing-wodehousehttps://semanticmachines.fyi/evals/creative-writing-wodehouseA prompt for testing AI language models: Creative Writing: Wodehouse.https://semanticmachines.fyi/evals/creative-writing-poetryhttps://semanticmachines.fyi/evals/creative-writing-poetryA prompt for testing AI language models: Creative Writing: Poetry.https://semanticmachines.fyi/evals/image-generationhttps://semanticmachines.fyi/evals/image-generationPicture of a wineglass full to the brim with a clock on the wall behind it showing half past three.https://semanticmachines.fyi/evals/long-context-anachronismshttps://semanticmachines.fyi/evals/long-context-anachronismsA prompt for testing AI language models: Long Context: Anachronisms.https://semanticmachines.fyi/evals/long-context-retrievalhttps://semanticmachines.fyi/evals/long-context-retrievalA prompt for testing AI language models: Long Context: Retrieval.https://semanticmachines.fyi/evals/visionhttps://semanticmachines.fyi/evals/visionA prompt for testing AI language models: Vision.https://semanticmachines.fyi/evals/agentic-taskshttps://semanticmachines.fyi/evals/agentic-tasksMake a list of Oxford museums and galleries that are within 10-15 minute walking distance of Jesus College for conference attendees.https://semanticmachines.fyi/evals/arithmetic-fixed-caseshttps://semanticmachines.fyi/evals/arithmetic-fixed-casesSmall computable baseline for checking arithmetic reliability and output discipline.https://semanticmachines.fyi/evals/czech-case-analysishttps://semanticmachines.fyi/evals/czech-case-analysisTests multilingual grammatical analysis in a language with visible case marking.https://semanticmachines.fyi/evals/heavy-and-light-idiom-translationhttps://semanticmachines.fyi/evals/heavy-and-light-idiom-translationTests whether a model notices metaphor and avoids word-by-word transfer.https://semanticmachines.fyi/evals/image-generation-wineglass-and-clockhttps://semanticmachines.fyi/evals/image-generation-wineglass-and-clockGeneralised image-generation comparison prompt inspired by the `gpt-image-2-capabilities` gallery.https://semanticmachines.fyi/evals/two-turn-constraint-shifthttps://semanticmachines.fyi/evals/two-turn-constraint-shiftTests whether the model can revise a previous answer when the user adds a stricter constraint.https://semanticmachines.fyi/evals/czech-spring-poem-three-turn-revisionhttps://semanticmachines.fyi/evals/czech-spring-poem-three-turn-revisionArchived three-turn Czech spring-poem test.https://semanticmachines.fyi/evals/sla-nanynka-do-zeli-five-poetshttps://semanticmachines.fyi/evals/sla-nanynka-do-zeli-five-poetsFive controlled Czech poet-style transformations.https://semanticmachines.fyi/evals/terra-blind-critique-springhttps://semanticmachines.fyi/evals/terra-blind-critique-springTerra comments on blind-labelled final Spring outputs. Labels: A GPT-4o, B o1, C o3, D GPT-5, E GPT-5.5, F Terra.https://semanticmachines.fyi/evals/terra-blind-critique-nanynkahttps://semanticmachines.fyi/evals/terra-blind-critique-nanynkaTerra comments on blind-labelled Nanynka outputs. Labels: A GPT-4o, B o1, C o3, D GPT-5, E GPT-5.5, F Terra.