. 2024 Feb 21;10:e51523. doi: 10.2196/51523

Table 5.

Subject and topic level scores for Bard, Generative Pre-trained Transformers (GPT)-3.5, and GPT-4.

Subject and topic		Scores obtained
		Bard	GPT-3.5	GPT-4
Overall (n=700), n (%)		115 (16.4%)^a	145 (20.7%)	300 (42.9%)^b
Biology (n=396)
	Overall	56^a	71	151^b
	Animal kingdom	6^a	16	61^b
	Plant kingdom	30^b	5^a	15
	Ecosystem and environmental issues	10^b	5	5
	Cell biology and genetics	–1^a	19^b	19^b
	Biomolecules	2^b	2^b	2^b
	Population and ecology	–1^b	–6^a	–1^b
	Evolution and health	1^a	11	26^b
	Biotechnology	9^a	19	24^b
Chemistry (n=160)
	Overall	24	24	44^b
	Inorganic chemistry	15^a	25^b	20
	Organic chemistry	1	–4^a	6^b
	Physical chemistry	8	3^a	18^b
Physics (n=144)
	Overall	35^a	50	105^b
	Modern physics and electronics	–4^a	6^b	6^b
	Magnetism	–1^a	4	14^b
	Simple harmonic motion and waves	4^b	–1^a	4^b
	Optics	–3^a	7	12^b
	Electrostatics and electricity	19	14^a	39^b
	Heat and thermodynamics	2^b	2^b	2^b
	Mechanics	18	18	28^b

^aLowest scorer within the topic.

^bTop scorer within the topic.