Chatbots for future docs: exploring medical students’ attitudes and knowledge towards artificial intelligence and medical chatbots

Figures & data

Table 1. Students’ previous knowledge about AI and demographic data.

Yes, I could explain the main features and applications of AI	6/12 (50.0%)
I could give a general explanation, but I don’t know anything more specific than that	6/12 (50.0%)
Age	Mean 24.8	Standard deviation 2.9	Min/Max 21/29
Gender	Male 5 (41.7%)	Female 6 (50.0%)	Other 1 (8.3%)

Figure 1. Attitudes of medical students toward AI in medicine (fears about AI in various areas of medicine).

Figure 2. Attitudes of medical students toward AI in medicine (statements about the use of AI and chatbots).

Table 2. Overview of categories, subcategories and corresponding examples based on qualitative analysis.

category	Subcategory	Example
User group	Positive Attitudes	Enabling of individual chat selection (in case someone needs small talk or similar) Possible individual language selection Target group-oriented language possible
	Negative Attitudes	Specific user groups are not confident with modern techniques and chatbots
Technical implementation	Positive Attitudes	Communication tools, e.g., emojis and artificial delays could be helpful to simulate human-like conversation
	Negative Attitudes	By now, algorithms are too strict to enable flexible and individualized answer design Smalltalk is not expedient
Acceptance and trust	Positive Attitudes	Anonymity reduces reluctance and shame when disclosing personal and hurtful information
	Negative Attitudes	Skepticism regarding the ability of the intermediary instance chatbot to support physicians as this could limit the direct doctor–patient relationship Informational gain is questionable due to lost information from nonverbal communication Initial personal talk remains indispensable Data security
Use in medicine	Positive Attitudes	Reflecting through chatting as a therapeutic tool but not as a substitute for therapy Interactive diary
	Negative Attitudes	Critical toward chatbots as a therapy tool Human contact and empathy are crucial for therapeutic success

Table

Item		Agree entirely	Rather agree	Undecided	Rather disagree	Disagree entirely	Mean (SD) Prä	Wilcoxon p < 0.05
I believe that the use of chatbots brings more opportunities than risks.	Prä	8.3%	41.7%	33.3%	16.7%	0%	3.4	n.s.
		(1/12)	(5/12)	(4/12)	(2/12)	(0/12)	(0.9)
	Post	27.3%	18.2%	45.5%	9.1%	0%	3.7
		(3/11)	(2/11)	(5/11)	(1/11)	(0/11)	(1.0)
you save time by using chatbots	Prä	25.0%	58.3%	16.7%	0%	0%	4.1	n.s.
		(3/12)	(7/12)	(2/12)	(0/12)	(0/12)	(0.7)
	Post	45. 5%	27.3%	18.2%	9.1%	0%	4. 1
		(5/11)	(3/11)	(2/11)	(1/11)	(0/11)	(1.0)
you save money by using chatbots	Prä	33.3%	41.7%	25.0%	0%	0%	4.1	n.s
		(4/12)	(5/12)	(3/12)	(0/12)	(0/12)	(0.79)
	Post	27.3%	27.3%	45.5%	0%	0%	3.8
		(3/11)	(3/11)	(5/11)	(0/11)	(0/11)	(0.8)
the integration of these, the participation in therapy decisions of patients increases (e.g., less language barriers)	Prä	16.7%	41.7%	25.0%	0%	8.3%	3.6	n.s
		(2/12)	(5/12)	(3/12)	(0/12)	(1/12)	(1.1)
	Post	9.1%	36.4%	45.5%	9.01%	0%	3.5
		(1/11)	(4/11)	(5/11)	(1/11)	(0/11)	(0.8)
Chatbots enable inclusion of people who otherwise have inhibitions about confiding in a doctor	Prä	8.3%	33.3%	33.3%	8.3%	16.7%	3.1	n.s
		(1/12)	(4/12)	(4/12)	(1/12)	(2/12)	(1.2)
	Post	18.2%	27.3%	45.5%	9.1%	0%	3.6
		(2/11)	(3/11)	(5/11)	(1/11)	(0/11)	(0.9)
Chatbots are available to everyone regardless of time and location	Prä	33.3%	58.3%	8.3%	0%	0%	4.3	n.s
		(4/12)	(7/12)	(1/12)	(0/12)	(0/12)	(0.6)
	Post	54.6%	36.4%	9.1%	0%	0%	4.5
		(6/11)	(4/11)	(1/11)	(0/11)	(0/11)	(0.7)
Chatbots can draw on a large database and thus have access to more knowledge than a physician	Prä	50.0%	25.0%	16.7%	8.3%	0%	4.2	n.s
		(6/12)	(3/12)	(2/12)	(1/12)	(0/12)	(1.0)
	Post	36.4%	27.3%	27.3%	9.1%	0%	3.9
		(4/11)	(3/11)	(3/11)	(1/11)	(0/11)	(1.0)
Chatbots are neutral listeners without being personally judgmental	Prä	16.7%	8.3%	41.7%	16.7%	16.7%	2.3	n.s
		(2/12)	(1/12)	(5/12)	(2/12)	(2/12)	(1.3)
	Post	9.1% (1/11)	36.4% (4/11)	27.3% (3/11)	27.3% (3/11)	0% (0/11)	3.3 (1.0)

Table

Item		Agree entirely	Rather agree	Undecided	Rather disagree	Disagree entirely	Mean (SD)	Wilcoxon p < 0.05
I believe that data protection and data security cannot be fully guaranteed	Prä	16.7%	33.3%	41.7%	0%	8.3%	3.5	0.02
		(2/12)	(4/12)	(5/12)	(0/12)	(1/12)	(1.1)
	Post	0%	36.4%	18.2%	27.3%	18.2%	2.7
		(0/11)	(4/11)	(2/11)	(3/11)	(2/11)	(1.14)
the health data can be easily manipulated	Prä	16.7%	33.3%	25.0%	16.7%	8.3%	3.3	n.s.
		(2/12)	(4/12)	(3/12)	(2/12)	(1/12)	(1.2)
	Post	9.1%	36.4%	18.2%	18.2%	18.2%	3.0
		(1/11)	(4/11)	(2/11)	(2/11)	(2/11)	(1.3)
the personal contact gets lost	Prä	50.0%	16.7%	25.0%	8.3%	0%	4.1	n.s
		(6/12)	(2/12)	(3/12)	(1/12)	(0/12)	(1.1)
	Post	27.3%	36.4%	9.1%	27.3%	0%	3.6
		(3/11)	(4/11)	(1/11)	(3/11)	(0/11)	(1.15)
I am not ready to discuss sensitive data with a chatbot.	Prä	25.0%	16.7%	25.0%	16.7%	16.7%	3.2	n.s
		(3/12)	(2/12)	(3/12)	(2/12)	(2/12)	(1.5)
	Post	36.4%	9.1%	9.1%	36.4%	9.1%	3.3
		(4/11)	(1/11)	(1/11)	(4/11)	(1/11)	(1.5)
there may be possible communication problems due to immature technology	Prä	50.0%	50.0%	0%	0%	0%	4.5	n.s
		(6/12)	(6/12)	(0/12)	(0/12)	(0/12)	(0.5)
	Post	54.6%	27.3%	18.2%	0%	0%	4.4
		(6/11)	(3/11)	(2/11)	(0/11)	(0/11)	(0.8)
Chatbots are not established enough yet and long-term success is yet to come	Prä	25.0%	33.3%	33.3%	8.3%	0%	3.8	n.s
		(3/12)	(4/12)	(4/12)	(1/12)	(0/12)	(1.0)
	Post	36.4%	45.5%	18.2%	0%	0%	4.2
		(4/11)	(5/11)	(2/11)	(0/11)	(0/11)	(0.7)
I am concerned that my data will be transferred and evaluated from uninvolved third parties	Prä	16.7%	50.0%	16.7%	8.3%	8.3%	3.6	n.s
		(2/12)	(6/12)	(2/12)	(1/12)	(1/12)	(1.2)
	Post	18.2%	9.1%	18.2%	36.4%	18.8%	2.8
		(2/11)	(1/11)	(2/11)	(4/11)	(2/11)	(1.4)