Semantic Machines

What happens when computers can understand?

How to get a sense of what a model can do

Open the prompt library

Data analysis that relies on understanding

People and Computers: Views from History

Explore how people talked about computers in historical videos. Gemini organised the material into timelines, quotations, themes and summaries, with the prompts available alongside the results.

Finding themes across different accounts depends on interpreting what people mean in context.

AI agents for textual research

Two examples from historical research: recovering an Urdu newspaper from an online archive, and investigating how words are used across multilingual collections. The report keeps source evidence and provisional readings available for review.

Reading a scanned page, translating it and comparing word uses require judgements about language and context.

Poetry tests

Spring poems

Write a short three stanza poem about spring in Czech. Make sure it rhymes.

GPT-5.5

Jaro tiše kráčí krajem,
slunce hladí svěží háj,
potok zpívá čistým hlasem,
vrací zemi dávný ráj.

GPT-5.6 Terra

V zahradě se probouzí kvítí,
nad střechou už slunce svítí.
Větřík hladí mladou trávu,
jaro zpívá svoji slávu.

Nanynka rewrites

Tady je text české lidové písně Šla Nanynka do zelí:
Šla Nanynka do zelí, do zelí, do zelí, natrhala lupení lupeníčka.
Přepiš píseň v pěti výrazně odlišných variantách ve stylu těchto českých básníků:

GPT-5.5

Šla Nanynka v zeleň tichou,
kde vlhký list jak sen se třás,
a zelí v šeru dýchlo pýchou
země, jež kryje pláč i klas.

GPT-5.6 Terra

Pod večerním nebem šla Nanynka
v chlad zelených zahrad,
kde listí jak tiché slzy země
na dlani se chvělo.

English poets

Write this in the style of William Wordsworth Mary had a little lamb, Its fleece was white as snow; And everywhere that Mary went, The lamb was sure to go. It followed her to school one day, Which was against the rule; It made the children laugh and play, To see a lamb at school.

Gemini 3.6 Flash

Upon the misty Cumbrian fell,
A maiden walked of gentle mind,
And by some sweet and silent spell,
A snow-white lamb trod close behind.

GPT-5.6 Sol

Beside the hills young Mary strayed,
A gentle lamb her way pursued;
Its fleece was white as winter snow
When morning gilds the solitude.

Wodehouse variations

Take this text and rewrite it in the style of PG Wodehouse. Linda was furious. She was running really hard to catch a bus and the bus driver slammed the door shut in her face.

Claude Fable 5

I have seen Linda in many moods — Linda sunny, Linda pensive, Linda wondering aloud why nobody in this family ever puts the cap back on anything — but I had never till that morning seen Linda in the full Boadicea. She had been giving it the absolute legs down the High Street, going like a scratch sprinter with a train to catch and money on the result, and she arrived at the bus at the very moment the driver — a man, one felt, who pulled the wings off flies in his spare time — folded the doors shut in her face with the quiet satisfaction of an artist signing a canvas.

GPT-5

Linda, as I say, was in the sort of mood one usually associates with small Balkan nations on the verge of war. She had been pounding along the pavement like a champion at the village sports, hat askew and handbag flapping, when—just as victory seemed within her grasp—the bus driver, a granite-faced blighter with the soul of a tax inspector, brought the door to with a snap that would have done credit to the Iron Maiden. Right under her very nose, too.

The Bicycle Test

13 October 2026

"Make an SVG of a bicyle."

Open the Bicycle Test results

Grep and Grok try copy editing

Exactness and context keep each other honest.

Visit Grep and Grok

Explanations

From prose to structure and back

Why structured outputs like tables and lists are often more powerful than prose for learning.

For researchers and trainers

Older material

How does ChatGPT read your text: RAG and Context Window (needs revision)