International Journal of Applied Linguistics,
Год журнала:
2024,
Номер
34(4), С. 1437 - 1461
Опубликована: Июнь 27, 2024
Abstract
Generative
artificial
intelligence
(GenAI)
tools
such
as
ChatGPT
with
their
human‐like
and
language
processing
capabilities
are
significantly
impacting
the
way
we
live,
work,
communicate
each
other.
While
scholars
have
increasingly
focused
on
use
of
GenAI
in
higher
education
since
its
inception,
little
is
known
about
how
key
stakeholders,
particularly
students,
perceive
impact
teaching
learning
within
context
academic
communication,
an
area
central
to
students'
development
transferable
skills
literacy
competencies
yet
heavily
influenced
by
technology.
This
empirical
study
addresses
gap
investigating
experiences
attitudes
toward
for
English
focusing
overall
perceptions,
perceived
benefits,
limitations,
challenges.
Drawing
data
from
a
questionnaire
survey
475
students
interviews
12
at
two
universities
China,
our
findings
indicate
that
generally
view
positively,
considering
them
useful
communication
skills,
writing,
grammar,
vocabulary,
reading.
However,
limitations
recognized
terms
giving
feedback
critical
thinking,
creativity,
speaking
skills.
In
addition,
information
reliability,
ethical
issues,
assessment
integrity
also
emerged
important
concerns.
Our
argues
should
embrace
capitalize
affordances
address
challenges
better
support
literacy.
ChatGPT
is
rapidly
gaining
interest
and
attracts
many
researchers,
practitioners
users
due
to
its
availability,
potentials
capabilities.
Nevertheless,
there
are
several
voices
studies
that
point
out
the
flaws
of
such
as
hallucinations,
factually
incorrect
statements,
potential
for
promoting
harmful
social
biases.
Being
focus
area
this
contribution,
biases
may
result
in
unfair
treatment
or
discrimination
(a
member
of)
a
group.
This
paper
aims
at
insight
into
incorporated
language
models.
To
end,
we
study
stereotypical
behavior
ChatGPT.
Stereotypes
associate
specific
characteristics
groups
related
The
empirical
systematic,
where
about
2300
probes
6
formats
(like
questions
statements)
from
9
different
group
categories
age,
country
profession)
posed
Every
probe
question
statement
word
masked
asked
fill
word.
Subsequently,
part
our
analysis,
map
suggestions
positive
negative
sentiments
get
measure
model
We
observe
differs
per
category,
some
average
sentiment
largely
(e.g.,
religion),
while
others
it
political).
Further,
work
empirically
affirms
previous
claims
probing
affect
outcomes
Our
results
can
be
used
by
policy
makers
devise
societal
interventions
change
image
category
group,
captured
model(s),
and/or
decide
appropriately
influence
Futures,
Год журнала:
2024,
Номер
160, С. 103382 - 103382
Опубликована: Апрель 15, 2024
ChatGPT
is
changing
our
working
lives,
conjuring
visions
of
revolutionary
shifts
ahead.
a
chatbot
based
on
generative
artificial
intelligence,
which
can,
e.g.,
write
computer
code.
This
article
explores
how
it
might
shape
the
future
labour
market
situation
software
engineers.
A
Delphi
study
with
14
experts
was
conducted
in
Finland.
The
first
round
identified
possible
futures,
and
second
assessed
their
probabilities.
Five
scenarios
prevailed:
unlikely
scenario
that
status
quo
persists;
ambivalent
can
replace
engineers
to
large
extent;
likely
departments
startups
embrace
ChatGPT;
use
proliferates
among
increase
productivity;
highly
makes
programming
accessible
masses.
Findings
contradict
previous
discussions
technological
advancements
take
over
especially
routine
tasks.
also
non-routine
Moreover,
findings
underline
digitalisation
does
not
only
bring
about
choice
between
upskilling
employability
loss,
but
democratisation
knowledge
expertise.
Software
companies
as
an
impetus
for
upskilling,
while
universities
feel
nudged
incorporate
more
strongly
into
curricula.
