01909nM2.01200024      h001 3507017\x1e002a20241209000000.0\x1e003 20250210000000.0\x1e02
5o1492115428\x1e030 zz5dr|rz|||3|\x1e037beng\x1e050 ||||||||g|||||\x1e051 m||||||\x1e060 \x1faText
\x1fbtxt\x1e061 \x1faComputermedien\x1fbc\x1e062 \x1faOnline-Ressource\x1fbcr\x1e070 BVB\x1e076 RDA-Aufnahm
e\x1e077 \x1fc0\x1e331 Putting AI to the test\x1e335 How does the performance of GPT and 15-
year-old students in PISA compare?\x1e359 Organisation for Economic Co-operation an
d Development\x1e410 Paris\x1e412 OECD Publishing\x1e419 \x1faParis\x1fbOECD Publishing\x1fc2023\x1e4
25a2023\x1e433 1 Online-Ressource (8 Seiten)\x1e435 21 x 28cm\x1e451 OECD Education Spotl
ights\x1e655e\x1fxVerlag\x1fzkostenfrei\x1f3Volltext\x1fuhttps://doi.org/10.1787/2c297e0b-en\x1e75
0 Advancements in artificial intelligence (AI) are laying the groundwork for ext
ensive and rapid transformations in society. Understanding the relationship betw
een AI capabilities and human skills is essential to ensure policy responsivenes
s to ongoing and incoming changes. The OECD has tracked how well AI systems fare
 on tasks from the Programme for International Student Assessment (PISA), compar
ing AI performance to that of 15-year-old students in the test's core domains of
 reading, mathematics and science. Tests were conducted using the Generative Pre
-Trained Transformer (GPT) family of large language models (LLMs), the AI behind
 ChatGPT, which took the world by storm after its public release in late 2022. R
esults show that both GPT versions outperform average student performance in rea
ding and science. In addition, we observe rapid advances in mathematics where AI
 capabilities are quickly catching up with those of students. In November 2022, 
GPT-3.5 could answer 35% of a set of PISA mathematics tasks, a level of performa
nce significantly below that of humans, who answer 51% of the tasks successfully
 on average. However, by March 2023, GPT-4 answered 40% of the tasks successfull
y. Policy implications of these results are discussed in this paper\x1e\x1d