Berliner Boersenzeitung - AI makes writing easier, but stories sound alike

EUR -
AED 4.225652
AFN 74.781422
ALL 93.172811
AMD 421.052567
AOA 1054.989049
ARS 1719.121524
AUD 1.645162
AWG 2.073739
AZN 1.966712
BAM 1.952625
BBD 2.316801
BDT 142.037287
BHD 0.433844
BIF 3445.684793
BMD 1.150479
BND 1.475205
BOB 14.004252
BRL 5.872623
BSD 1.150259
BTN 109.610603
BWP 15.644674
BYN 3.362513
BYR 22549.38963
BZD 2.313477
CAD 1.616521
CDF 2602.383819
CHF 0.933234
CLF 0.026952
CLP 1064.251123
CNY 7.768725
CNH 7.775841
COP 3730.370837
CRC 522.516208
CUC 1.150479
CUP 30.487695
CVE 110.733442
CZK 24.214474
DJF 204.463134
DKK 7.476256
DOP 67.015623
DZD 153.007976
EGP 57.663627
ERN 17.257186
ETB 184.018741
FJD 2.552112
FKP 0.853363
GBP 0.857308
GEL 3.003154
GGP 0.853363
GHS 13.4721
GIP 0.853363
GMD 85.135506
GNF 10066.691947
GTQ 8.776194
GYD 241.031269
HKD 9.022224
HNL 30.831984
HRK 7.535344
HTG 150.395619
HUF 364.449882
IDR 20694.817381
ILS 3.501638
IMP 0.853363
INR 109.654226
IQD 1507.127572
IRR 1582196.33128
ISK 142.015652
JEP 0.853363
JMD 182.213385
JOD 0.815707
JPY 180.642471
KES 148.791673
KGS 100.609628
KHR 4645.047021
KMF 492.405068
KRW 1645.311841
KWD 0.356071
KYD 0.958587
KZT 545.649452
LAK 26000.827293
LBP 103025.400201
LKR 386.210115
LRD 208.524101
LSL 19.017405
LTL 3.397066
LVL 0.695913
LYD 7.322825
MAD 10.739434
MDL 20.084204
MGA 4899.755071
MKD 61.416653
MMK 2415.641494
MNT 4134.279366
MOP 9.291565
MRU 46.122794
MUR 53.795901
MVR 17.775066
MWK 1998.382168
MXN 19.949934
MYR 4.711789
MZN 73.518466
NAD 19.0174
NGN 1569.161798
NIO 42.327671
NOK 10.991775
NPR 175.378125
NZD 1.962585
OMR 0.442347
PAB 1.150249
PEN 3.905298
PGK 5.076458
PHP 70.072799
PKR 319.631884
PLN 4.314383
PYG 6861.424306
QAR 4.194359
RON 5.247219
RSD 117.37763
RUB 92.498091
RWF 1688.903264
SAR 4.305701
SBD 9.297142
SCR 15.579762
SDG 690.287687
SEK 11.014508
SGD 1.476323
SLE 28.12944
SOS 657.390936
SRD 43.495
STD 23812.593791
STN 24.458736
SVC 10.064801
SZL 19.017571
THB 38.471631
TJS 10.622786
TMT 4.026677
TND 3.382985
TRY 54.678932
TTD 7.799807
TWD 37.364799
TZS 3043.321989
UAH 51.597975
UGX 4304.756289
USD 1.150479
UYU 46.262154
UZS 13821.855768
VES 860.387379
VND 30242.643122
VUV 136.964711
WST 3.145975
XAF 654.855061
XAG 0.01993
XAU 0.000285
XCD 3.109228
XCG 2.073103
XDR 0.813195
XOF 656.350485
XPF 119.331742
YER 274.271757
ZAR 19.037552
ZMK 10355.697109
ZMW 21.617342
ZWL 370.453789
  • BCC

    5.9750

    82.355

    +7.26%

  • JRI

    -0.1310

    12.829

    -1.02%

  • RIO

    -1.1600

    95.69

    -1.21%

  • CMSC

    -0.0400

    21.8

    -0.18%

  • NGG

    -0.3400

    79.63

    -0.43%

  • RBGPF

    -3.2100

    66

    -4.86%

  • CMSD

    -0.1100

    22

    -0.5%

  • GSK

    -0.1750

    51.515

    -0.34%

  • BTI

    -0.9900

    59.66

    -1.66%

  • BCE

    0.0650

    21.745

    +0.3%

  • RYCEF

    0.8000

    20.3

    +3.94%

  • VOD

    -0.1750

    15.605

    -1.12%

  • AZN

    -11.1500

    158.49

    -7.04%

  • BP

    -1.0150

    44.205

    -2.3%

  • RELX

    0.8400

    36.26

    +2.32%

AI makes writing easier, but stories sound alike
AI makes writing easier, but stories sound alike / Photo: JOEL SAGET - AFP/File

AI makes writing easier, but stories sound alike

Books and movies of the future could all start to feel the same if creative industries embrace artificial intelligence to help write stories, a study published on Friday warned.

Text size:

The research, which drew on hundreds of volunteers and was published in Science Advances, comes amid rising fears over the impact of widely available AI tools that turn simple text prompts into relatively sophisticated music, art and writing.

"Our goal was to study to what extent and how generative AI might help humans with creativity," co-author Anil Doshi of the University College London told AFP.

For their experiment, Doshi and co-author Oliver Hauser of the University of Exeter recruited around 300 volunteers as "writers."

These were people who didn't write for a living, and their inherent creative ability was assessed by a standard psychology test that asked them to provide 10 drastically different words.

The scientists then split them randomly into three groups to write an eight-sentence story about one of three topics: an adventure on the open seas, an adventure in the jungle, or an adventure on another planet.

Participants were also randomly placed into three groups that received varying levels of AI assistance.

The first group got no help, the second was provided a three-sentence story idea from ChatGPT, and the third could receive up to five AI-generated story ideas to help them get going.

- Individual benefit, collective loss -

After completing their stories, participants were asked to assess their own work's creativity through measures including how novel it was, how enjoyable, and how much potential the idea had to be turned into a published book.

An additional 600 external human reviewers also judged the story on the same measures.

The authors found that, on average, AI boosted the quality of an individual writer's creativity by up to 10 percent, and the story's enjoyability by 22 percent, helping particularly with elements like structure and plot twists.

These effects were most significant for writers who were judged during the initial task to be the least creative, "so it has this kind of leveling the playing field effect," said Doshi.

But on the collective level, they found AI-assisted stories looked much more similar to each other than those produced without any AI help, as writers "anchored" themselves too heavily to the suggested ideas.

Hauser said this created a "social dilemma." On the one hand, "you make it easier for people to get in; lowering barriers is good." But if the collective novelty of art decreases, "it could be harmful down the line."

Doshi said the research also showed that, just like introducing pocket calculators to children too early could prevent them from learning how to do basic arithmetic, there was a danger that people could rely too much on AI tools before developing underlying skills in writing, music or more.

People need to start thinking about "where in my workflow can I insert this tool to get the most benefit, while still inserting my own voice into the project or outcome."

(O.Joost--BBZ)