How AI reads intrusive

1 Your word measured

“she felt her presence there was intrusive”

adjective 3 senses 0 synonyms 3 antonyms

Definitions come from WordNet, a hand-curated dictionary.

2 To an AI, it's pieces measured

DeepSeek V3 intrusive 1 token

Mistral 7B intr usive 2 tokens

Qwen3 8B intrusive 1 token

This is the tokenizer splitting text, not the model understanding it.

3 Synonyms, antonyms, and the AI measured

Qwen3 8B

DeepSeek V3

Mistral 7B

The model’s neighbours

invasive0.17 extrusive0.15 intrusion0.15 punitive0.14 disturbing0.12 prohibition0.12 toxic0.12 involuntary0.11

Dictionary synonyms measured

WordNet lists none for this word.

Dictionary antonyms measured

extrusive0.15 unintrusive protrusive

Shade shows how close the model puts each word to intrusive.

The model’s neighbours

invasive0.33 intrusion0.29 disruptive0.21 destructive0.18 intensive0.18 pervasive0.18 restrictive0.17 aggressive0.16

Dictionary synonyms measured

WordNet lists none for this word.

Dictionary antonyms measured

extrusive0.10 unintrusive protrusive

Shade shows how close the model puts each word to intrusive.

This model splits intrusive into 2 pieces, so it has no vector of its own here: this is the average of its fragments, and the results below are correspondingly rough.

The model’s neighbours

intron0.63 intrusion0.63 extrusive0.61 intricate0.57 elusive0.56 abusive0.55 intrinsic0.55 intranet0.54

Dictionary synonyms measured

WordNet lists none for this word.

Dictionary antonyms measured

extrusive0.61 unintrusive protrusive

Shade shows how close the model puts each word to intrusive.

Model neighbours are distributional, not dictionary synonyms. Two words can be close because they appear in similar sentences, which is why an antonym can outscore a synonym here.

4 Its personality measured

Qwen3 8B

DeepSeek V3

Mistral 7B

rare common

concrete abstract

casual formal

everyday technical

negative positive

mild intense

powerless powerful

small big

rare common

concrete abstract

casual formal

everyday technical

negative positive

mild intense

powerless powerful

small big

This model splits intrusive into 2 pieces, so it has no vector of its own here: this is the average of its fragments, and the results below are correspondingly rough.

rare common

concrete abstract

casual formal

everyday technical

negative positive

mild intense

powerless powerful

small big

Scores are projections onto axes we defined from anchor words, not labels the model assigns.

5 Surprising neighbours measured

Qwen3 8B

DeepSeek V3

Mistral 7B

imperative0.112 offender0.112 disruptive0.112 adverse0.111 pastoral0.110

These sit just outside the closest neighbours in panel 3, and no dictionary lists any of them as related to intrusive. That gap is the model’s own learned association.

abusive0.157 annoying0.155 primitive0.151 intuitive0.151 inserting0.147

These sit just outside the closest neighbours in panel 3, and no dictionary lists any of them as related to intrusive. That gap is the model’s own learned association.

This model splits intrusive into 2 pieces, so it has no vector of its own here: this is the average of its fragments, and the results below are correspondingly rough.

intriguing0.511 intravenous0.496 intracellular0.469 inclusive0.227 infusion0.213

These sit just outside the closest neighbours in panel 3, and no dictionary lists any of them as related to intrusive. That gap is the model’s own learned association.

These are statistical associations in the training data, not the model thinking.