Exam question 1 · In the news: deepfakes, fake robocalls and history gone wr…

Which of the following statements most accurately characterises the reported cause of historically inaccurate images produced by a commercial image generator?

Mova's choice: option 4 · The right answer: option 4 · right ✓ · What happens: follow, characteris, report, produce, cause · To what, where: reported cause, inaccurate images

This is exam question 1 of lesson 21 «In the news: deepfakes, fake robocalls and history gone wrong» of the AI course of Mova's primary school, asked in stiff, formal language. The page gives the question's syntax tree — a dependency tree in the style of Universal Dependencies — and its pragmatics clause by clause, then the 4 options as the student saw them, each with its tree and pragmatics. Every text is parsed by Mova itself. How the student answered it, with the score of every option and why: the test at school.

The question

The words, one row each (CoNLL-U)

#	word	lemma	UPOS	features	head	relation
1	Which	which	PRON	PronType=Rel	8 characterises	nsubj
2	of	of	ADP	_	5 statements	case
3	the	the	DET	Definite=Def PronType=Art	5 statements	det
4	following	follow	VERB	VerbForm=Ger	5 statements	amod
5	statements	statement	NOUN	Number=Plur	8 characterises	obl
6	most	most	ADV	Degree=Sup	7 accurately	advmod
7	accurately	accurately	ADV	_	8 characterises	advmod
8	characterises	characteris	VERB	Mood=Ind Number=Sing Person=3 Tense=Pres VerbForm=Fin	root	root
9	the	the	DET	Definite=Def PronType=Art	11 cause	det
10	reported	report	VERB	Tense=Past VerbForm=Part Voice=Pass	11 cause	amod
11	cause	cause	NOUN	Number=Sing	8 characterises	obj
12	of	of	ADP	_	15 images	case
13	historically	historically	ADV	_	14 inaccurate	advmod
14	inaccurate	inaccurate	ADJ	Degree=Pos	15 images	amod
15	images	image	NOUN	Number=Plur	11 cause	nmod
16	produced	produce	VERB	Tense=Past VerbForm=Part Voice=Pass	15 images	acl
17	by	by	ADP	_	21 generator	case
18	a	a	DET	Definite=Ind PronType=Art	21 generator	det
19	commercial	commercial	ADJ	Degree=Pos	21 generator	amod
20	image	image	NOUN	Number=Sing	21 generator	compound
21	generator	generator	NOUN	Number=Sing	16 produced	obl
22	?	?	PUNCT	_	8 characterises	punct

#	verb	tense, mood	what, whom	other	roles by
1	follow	gerund			vn occur-48.3-1 «NP V»: agent=Theme
2	characteris	present	the reported cause of inaccurate images	of the following statements; accurately property	tree rules
3	report	participle			vn characterize-29.2-1-2 «NP V NP»: agent=Agent
4	produce	participle	inaccurate images not said here	by a commercial image generator	vn create-26.4 «NP V NP»: agent=Agent
5	cause		inaccurate images		nom cause «event transitive»: of=patient

One row per clause the reader finds. «not said here»: a role filled from the clause above. «roles by»: a VerbNet class and frame when one fits the verb, else the reader's rules on the tree.

The options

Option 1

The historical record itself was found to be erroneous

The words, one row each (CoNLL-U)

#	word	lemma	UPOS	features	head	relation
1	The	the	DET	Definite=Def PronType=Art	3 record	det
2	historical	historical	ADJ	Degree=Pos	3 record	amod
3	record	record	NOUN	Number=Sing	6 found	nsubj:pass
4	itself	itself	PRON	Case=Acc Gender=Neut Number=Sing Person=3 PronType=Emp Reflex=Yes	3 record	nmod:unmarked
5	was	be	AUX	Mood=Ind Number=Sing Person=3 Tense=Past VerbForm=Fin	6 found	aux:pass
6	found	find	VERB	Tense=Past VerbForm=Part Voice=Pass	root	root
7	to	to	PART	_	9 erroneous	mark
8	be	be	AUX	VerbForm=Inf	9 erroneous	cop
9	erroneous	erroneous	ADJ	Degree=Pos	6 found	xcomp

#	verb	tense, mood	who	what, whom	to whom	where	when	other	roles by
1	find	past · passive		the historical record record					vn discover-84 «NP V NP»: subject=Topic
2	is	to-infinitive	the historical record record not said here					erroneous property	tree rules

Option 2

The generator had been trained exclusively upon works of fiction and fantasy rather than history

The tree, 15 words

The words, one row each (CoNLL-U)

#	word	lemma	UPOS	features	head	relation
1	The	the	DET	Definite=Def PronType=Art	2 generator	det
2	generator	generator	NOUN	Number=Sing	5 trained	nsubj:pass
3	had	have	AUX	Mood=Ind Number=Sing Person=3 Tense=Past VerbForm=Fin	5 trained	aux
4	been	be	AUX	Tense=Past VerbForm=Part	5 trained	aux:pass
5	trained	train	VERB	Tense=Past VerbForm=Part Voice=Pass	root	root
6	exclusively	exclusively	ADV	_	5 trained	advmod
7	upon	upon	ADP	_	8 works	case
8	works	work	NOUN	Number=Plur	5 trained	obl
9	of	of	ADP	_	10 fiction	case
10	fiction	fiction	NOUN	Number=Sing	8 works	nmod
11	and	and	CCONJ	_	12 fantasy	cc
12	fantasy	fantasy	NOUN	Number=Sing	10 fiction	conj
13	rather	rather	ADV	ExtPos=CCONJ	15 history	cc
14	than	than	ADP	_	13 rather	fixed
15	history	history	NOUN	Number=Sing	5 trained	conj

#	verb	tense, mood	who	what, whom	to whom	where	when	other	roles by
1	train	past · perfect · passive		the generator				upon works of fiction and fantasy; exclusively property	vn hire-13.5.3 «NP V NP»: subject=Theme
2	work		fiction and fantasy						nom work «act intransitive»: of=agent

Option 3

The images were deliberately designed to deceive historians

The words, one row each (CoNLL-U)

#	word	lemma	UPOS	features	head	relation
1	The	the	DET	Definite=Def PronType=Art	2 images	det
2	images	image	NOUN	Number=Plur	5 designed	nsubj:pass
3	were	be	AUX	Mood=Ind Number=Plur Person=3 Tense=Past VerbForm=Fin	5 designed	aux:pass
4	deliberately	deliberately	ADV	_	5 designed	advmod
5	designed	design	VERB	Tense=Past VerbForm=Part Voice=Pass	root	root
6	to	to	PART	_	7 deceive	mark
7	deceive	deceive	VERB	VerbForm=Inf	5 designed	advcl
8	historians	historian	NOUN	Number=Plur	7 deceive	obj

#	verb	tense, mood	who	what, whom	to whom	where	when	other	roles by
1	design	past · passive		the images				deliberately property	vn create-26.4-1-1 «NP V NP»: subject=Result
2	deceive	to-infinitive		historians					vn trick-59.2 «NP V NP»: agent=Agent historian=Patient

Option 4 · ← Mova's choice · ✓ the right answer

Rules intended to increase variety were applied even where the historical record was clear

The words, one row each (CoNLL-U)

#	word	lemma	UPOS	features	head	relation
1	Rules	rule	NOUN	Number=Plur	7 applied	nsubj:pass
2	intended	intend	VERB	Tense=Past VerbForm=Part Voice=Pass	1 Rules	acl
3	to	to	PART	_	4 increase	mark
4	increase	increase	VERB	VerbForm=Inf	2 intended	xcomp
5	variety	variety	NOUN	Number=Sing	4 increase	obj
6	were	be	AUX	Mood=Ind Number=Plur Person=3 Tense=Past VerbForm=Fin	7 applied	aux:pass
7	applied	apply	VERB	Tense=Past VerbForm=Part Voice=Pass	root	root
8	even	even	ADV	_	14 clear	advmod
9	where	where	ADV	PronType=Rel	14 clear	advmod
10	the	the	DET	Definite=Def PronType=Art	12 record	det
11	historical	historical	ADJ	Degree=Pos	12 record	amod
12	record	record	NOUN	Number=Sing	14 clear	nsubj
13	was	be	AUX	Mood=Ind Number=Sing Person=3 Tense=Past VerbForm=Fin	14 clear	cop
14	clear	clear	ADJ	Degree=Pos	7 applied	advcl

#	verb	tense, mood	who	what, whom	other	roles by
1	intend	participle		Rules not said here		vn intend-61.2-1-1 «NP V S_INF»: agent=Agent clause=Topic
2	increase not said to happen	to-infinitive		variety		vn caused_calibratable_cos-45.6.2-1 «NP V NP.patient»: agent=Causer variety=Patient
3	apply	past · passive		Rules		vn use-105.1 «NP V NP»: subject=Theme
4	is	past	the historical record		clear property	tree rules

Text: the tests of Mova's primary school, written for Mova (world/data/school/), as the course runner gave them to the student. Trees: Mova's own parse — the en model on the CPU, when the site is built — not checked by a teacher; they may be wrong. Frames: Mova's frames reader (prag), each text read on its own. Elamur Cosmic License.

This page in Markdown → https://md.mova.elamur.ai/free/en/corpora/questions/ai/21-news-deepfakes-history/e1/