Rows: 428 Columns: 34
── Column specification ────────────────────────────────────────────────────────
Delimiter: ","
chr (21): Gender, What is your current country of residence?, Area speciali...
dbl (11): ...1, Age, Year Experience, how would you rate your current knowl...
dttm (2): Response started, Response completed
ℹ Use `spec()` to retrieve the full column specification for this data.
ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.Exploratory Data Analysis - Insights from global dental educators about chatGPT and AI Chatbots
Load clean dataset
Add the continent detailed
IMF classification
EDA
[1] 428 36Hiow many countries?
[1] 66Table 1
| Characteristic | N = 4281 |
|---|---|
| Age | 45 (37, 56) |
| Year Experience | 16 (8, 25) |
| Gender | |
| Female | 226 (53%) |
| Male | 200 (47%) |
| Prefer not to say/Non-binary | 2 (0.5%) |
| Country | |
| Argentina | 14 (3.3%) |
| Brazil | 11 (2.6%) |
| Chile | 65 (15%) |
| Colombia | 14 (3.3%) |
| Egypt | 17 (4.0%) |
| France | 15 (3.5%) |
| India | 10 (2.3%) |
| Latvia | 13 (3.0%) |
| Lithuania | 11 (2.6%) |
| Mexico | 11 (2.6%) |
| Spain | 11 (2.6%) |
| United Kingdom | 16 (3.7%) |
| United States | 89 (21%) |
| Venezuela | 10 (2.3%) |
| Other | 121 (28%) |
| UN Classification | |
| Australia and New Zealand | 13 (3.0%) |
| Eastern Asia | 7 (1.6%) |
| Eastern Europe | 4 (0.9%) |
| Latin America and the Caribbean | 136 (32%) |
| Northern Africa | 20 (4.7%) |
| Northern America | 95 (22%) |
| Northern Europe | 50 (12%) |
| South-eastern Asia | 9 (2.1%) |
| Southern Asia | 13 (3.0%) |
| Southern Europe | 27 (6.3%) |
| Sub-Saharan Africa | 11 (2.6%) |
| Western Asia | 13 (3.0%) |
| Western Europe | 30 (7.0%) |
| Continent | |
| Africa | 31 (7.2%) |
| Americas | 231 (54%) |
| Asia | 42 (9.8%) |
| Europe | 111 (26%) |
| Oceania | 13 (3.0%) |
| 1 Median (IQR); n (%) | |
1. Current use of AI
| Use of AI | n | percent |
|---|---|---|
| I don't know. | 105 | 24.5% |
| No, we do not use AI-powered tools. | 190 | 44.4% |
| Yes, we use AI-powered tools. | 133 | 31.1% |
Cross check to verify if understand turnitin is a AI tool
| AI_use | No | Not sure | Yes |
|---|---|---|---|
| I don't know. | 61 | 7 | 37 |
| No, we do not use AI-powered tools. | 89 | 15 | 86 |
| Yes, we use AI-powered tools. | 42 | 4 | 87 |
Q8
The most frequently cited tools among users were ChatGPT, used for various tasks (24 mentions), and Turnitin, a plagiarism detection software (14 mentions). Tools such as Grammarly, oral/intraoral scanners (e.g., 3Shape Automate, iTero), and teledentistry and IA dental solutions (e.g., Overjet, AI Dental RPD Design) were each mentioned three times. AI-powered CAD Design tools, the educational simulation tool Kahoot, and AI tools for radiographic support and diagnosis received fewer mentions. Other AI tools, including Quillbot, DeepL, and image stitching software, totaled six mentions.
| how would you rate your current knowledge of AI-powered tools (such as ChatGPT) in education? (1 being very low and 5 being very high) | n | percent |
|---|---|---|
| 1 | 124 | 29.0% |
| 2 | 83 | 19.4% |
| 3 | 134 | 31.3% |
| 4 | 68 | 15.9% |
| 5 | 19 | 4.4% |
Min. 1st Qu. Median Mean 3rd Qu. Max.
1.000 1.000 3.000 2.474 3.000 5.000 2. Perceived impact of AI in dental education
believe AI can enhance dental education?
Min. 1st Qu. Median Mean 3rd Qu. Max.
1.000 3.000 4.000 3.801 5.000 5.000 | believe AI can enhance dental education? (1 being not at all and 5 being greatly) | n | percent |
|---|---|---|
| 1 | 11 | 0.0257009 |
| 2 | 22 | 0.0514019 |
| 3 | 122 | 0.2850467 |
| 4 | 159 | 0.3714953 |
| 5 | 114 | 0.2663551 |
believe AI can enhance dental education?
by continent
| Continent | n | mean | sd |
|---|---|---|---|
| Europe | 111 | 3.5 | 1.0 |
| Africa | 31 | 4.2 | 0.7 |
| Americas | 231 | 3.8 | 1.0 |
| Asia | 42 | 4.2 | 0.8 |
| Oceania | 13 | 3.6 | 1.4 |
Model by continent
Warning: Unknown levels in `f`: 5, 4, 3, 2, 1| Characteristic | N | OR1 | 95% CI1 | p-value |
|---|---|---|---|---|
| Continent | 426 | |||
| Europe | — | — | ||
| Africa | 6.37 | 2.43, 20.1 | <0.001 | |
| Americas | 2.72 | 1.68, 4.45 | <0.001 | |
| Asia | 3.86 | 1.73, 9.34 | 0.002 | |
| Oceania | 2.00 | 0.62, 7.08 | 0.3 | |
| Age | 426 | 0.98 | 0.96, 1.00 | 0.024 |
| Gender | 426 | |||
| Female | — | — | ||
| Male | 1.57 | 1.02, 2.42 | 0.041 | |
| 1 OR = Odds Ratio, CI = Confidence Interval | ||||
We fitted a logistic model (estimated using ML) to predict AI_enhance with
Continent (formula: AI_enhance ~ Continent + Age + Gender). The model's
explanatory power is weak (Tjur's R2 = 0.08). The model's intercept,
corresponding to Continent = Europe, is at 0.53 (95% CI [-0.30, 1.36], p =
0.210). Within this model:
- The effect of Continent [Africa] is statistically significant and positive
(beta = 1.85, 95% CI [0.89, 3.00], p < .001; Std. beta = 1.85, 95% CI [0.89,
3.00])
- The effect of Continent [Americas] is statistically significant and positive
(beta = 1.00, 95% CI [0.52, 1.49], p < .001; Std. beta = 1.00, 95% CI [0.52,
1.49])
- The effect of Continent [Asia] is statistically significant and positive
(beta = 1.35, 95% CI [0.55, 2.23], p = 0.002; Std. beta = 1.35, 95% CI [0.55,
2.23])
- The effect of Continent [Oceania] is statistically non-significant and
positive (beta = 0.69, 95% CI [-0.48, 1.96], p = 0.254; Std. beta = 0.69, 95%
CI [-0.48, 1.96])
- The effect of Age is statistically significant and negative (beta = -0.02,
95% CI [-0.04, -2.77e-03], p = 0.024; Std. beta = -0.26, 95% CI [-0.48, -0.04])
- The effect of Gender [Male] is statistically significant and positive (beta =
0.45, 95% CI [0.02, 0.88], p = 0.041; Std. beta = 0.45, 95% CI [0.02, 0.88])
Standardized parameters were obtained by fitting the model on a standardized
version of the dataset. 95% Confidence Intervals (CIs) and p-values were
computed using a Wald z-distribution approximation., We fitted a logistic model
(estimated using ML) to predict AI_enhance with Age (formula: AI_enhance ~
Continent + Age + Gender). The model's explanatory power is weak (Tjur's R2 =
0.08). The model's intercept, corresponding to Age = 0, is at 0.53 (95% CI
[-0.30, 1.36], p = 0.210). Within this model:
- The effect of Continent [Africa] is statistically significant and positive
(beta = 1.85, 95% CI [0.89, 3.00], p < .001; Std. beta = 1.85, 95% CI [0.89,
3.00])
- The effect of Continent [Americas] is statistically significant and positive
(beta = 1.00, 95% CI [0.52, 1.49], p < .001; Std. beta = 1.00, 95% CI [0.52,
1.49])
- The effect of Continent [Asia] is statistically significant and positive
(beta = 1.35, 95% CI [0.55, 2.23], p = 0.002; Std. beta = 1.35, 95% CI [0.55,
2.23])
- The effect of Continent [Oceania] is statistically non-significant and
positive (beta = 0.69, 95% CI [-0.48, 1.96], p = 0.254; Std. beta = 0.69, 95%
CI [-0.48, 1.96])
- The effect of Age is statistically significant and negative (beta = -0.02,
95% CI [-0.04, -2.77e-03], p = 0.024; Std. beta = -0.26, 95% CI [-0.48, -0.04])
- The effect of Gender [Male] is statistically significant and positive (beta =
0.45, 95% CI [0.02, 0.88], p = 0.041; Std. beta = 0.45, 95% CI [0.02, 0.88])
Standardized parameters were obtained by fitting the model on a standardized
version of the dataset. 95% Confidence Intervals (CIs) and p-values were
computed using a Wald z-distribution approximation. and We fitted a logistic
model (estimated using ML) to predict AI_enhance with Gender (formula:
AI_enhance ~ Continent + Age + Gender). The model's explanatory power is weak
(Tjur's R2 = 0.08). The model's intercept, corresponding to Gender = Female, is
at 0.53 (95% CI [-0.30, 1.36], p = 0.210). Within this model:
- The effect of Continent [Africa] is statistically significant and positive
(beta = 1.85, 95% CI [0.89, 3.00], p < .001; Std. beta = 1.85, 95% CI [0.89,
3.00])
- The effect of Continent [Americas] is statistically significant and positive
(beta = 1.00, 95% CI [0.52, 1.49], p < .001; Std. beta = 1.00, 95% CI [0.52,
1.49])
- The effect of Continent [Asia] is statistically significant and positive
(beta = 1.35, 95% CI [0.55, 2.23], p = 0.002; Std. beta = 1.35, 95% CI [0.55,
2.23])
- The effect of Continent [Oceania] is statistically non-significant and
positive (beta = 0.69, 95% CI [-0.48, 1.96], p = 0.254; Std. beta = 0.69, 95%
CI [-0.48, 1.96])
- The effect of Age is statistically significant and negative (beta = -0.02,
95% CI [-0.04, -2.77e-03], p = 0.024; Std. beta = -0.26, 95% CI [-0.48, -0.04])
- The effect of Gender [Male] is statistically significant and positive (beta =
0.45, 95% CI [0.02, 0.88], p = 0.041; Std. beta = 0.45, 95% CI [0.02, 0.88])
Standardized parameters were obtained by fitting the model on a standardized
version of the dataset. 95% Confidence Intervals (CIs) and p-values were
computed using a Wald z-distribution approximation.
By UN region
| UN Classification | n | Mean | SD |
|---|---|---|---|
| Western Europe | 30 | 3.5 | 0.9 |
| Eastern Europe | 4 | 4.2 | 1.0 |
| Northern Europe | 50 | 3.6 | 1.1 |
| Southern Europe | 27 | 3.4 | 0.7 |
| Northern America | 95 | 3.7 | 1.1 |
| Latin America and the Caribbean | 136 | 3.9 | 0.9 |
| Northern Africa | 20 | 4.2 | 0.7 |
| Sub-Saharan Africa | 11 | 4.2 | 0.8 |
| Eastern Asia | 7 | 3.9 | 0.9 |
| South-eastern Asia | 9 | 4.2 | 0.4 |
| Southern Asia | 13 | 4.2 | 0.8 |
| Western Asia | 13 | 4.4 | 0.9 |
| Australia and New Zealand | 13 | 3.6 | 1.4 |
“What dental education areas could be most enhanced by AI-powered tools? ()”
| Areas_enhace | n | porcentaje |
|---|---|---|
| Knowledge acquisition | 318 | 74.29907 |
| Research | 293 | 68.45794 |
| Clinical decision making | 272 | 63.55140 |
| Assessment | 257 | 60.04673 |
| Clinical skills training | 166 | 38.78505 |
| Other | 43 | 10.04673 |
3. Perceived barriers to the use of AI chatbots in dental education
Relabel and relevel
Likert barriers table
| item | Completely Disagree | Disagree | Neutral | Agree | Completely Agree |
|---|---|---|---|---|---|
| AI tools are easy to use. | 10 | 44 | 182 | 134 | 58 |
| AI tools can accurately assess students' skills. | 43 | 126 | 141 | 89 | 29 |
| AI tools can replace traditional teaching methods. | 98 | 117 | 103 | 81 | 29 |
| The benefits of using AI tools outweigh the costs. | 12 | 52 | 188 | 125 | 51 |
| There is adequate support and training for AI tools. | 91 | 163 | 126 | 37 | 11 |
in Likert format
4. Future of AI chatbots in dental education
| comfortable_assessment | n | percent |
|---|---|---|
| 1 | 22 | 5.1% |
| 2 | 79 | 18.5% |
| 3 | 158 | 36.9% |
| 4 | 115 | 26.9% |
| 5 | 54 | 12.6% |
Min. 1st Qu. Median Mean 3rd Qu. Max.
1.000 3.000 3.000 3.234 4.000 5.000 
20 “What concerns, if any, do you have regarding AI-powered tools’ involvement in student evaluations and assessments? ()”
| concerns | n | percent |
|---|---|---|
| AI tools may not assess critical thinking and creativity effectively | 275 | 32.5% |
| AI tools may not accurately evaluate students' skills | 190 | 22.5% |
| There may be a lack of human touch in AI assessment | 176 | 20.8% |
| Concerns about data privacy and security | 134 | 15.9% |
| Other | 37 | 4.4% |
| No concerns | 33 | 3.9% |
21 “What potential benefits can AI-powered tools bring to dental education’s evaluation and assessment process? ()”
| benefits | n | percent |
|---|---|---|
| Increased efficiency in marking and feedback | 296 | 35.5% |
| Consistency in assessment | 274 | 32.9% |
| Personalized feedback based on student's performance | 233 | 27.9% |
| Other | 31 | 3.7% |
22 “What changes should be made in the current evaluation and assessment process to accommodate the involvement of AI-powered tools better? ()”
| changes | n | percent |
|---|---|---|
| More training for educators on how to use AI tools | 386 | 35.2% |
| Clear guidelines on how AI tools should be used for evaluation | 369 | 33.7% |
| Inclusion of AI tools in the curriculum | 270 | 24.6% |
| Less essay | 37 | 3.4% |
| Other | 34 | 3.1% |
5. Perceptions impact
# A tibble: 7 × 6
# Groups: item [7]
item Stron…¹ Disag…² Neutral Agree Stron…³
<chr> <int> <int> <int> <int> <int>
1 " AI tools like ChatGPT will significa… 22 51 154 119 42
2 " AI tools like ChatGPT will impact ass… 6 19 123 175 65
3 " AI tools like ChatGPT will likely lea… 13 62 131 120 62
4 "ChatGPT enhance critical thinking skil… 41 101 144 82 20
5 "ChatGPT result in an efficient grading… 14 39 137 164 34
6 "ChatGPT result in improved student eng… 28 63 142 122 33
7 "ChatGPTreduce human interaction and fe… 10 61 108 151 58
# … with abbreviated variable names ¹`Strongly Disagree`, ²Disagree,
# ³`Strongly Agree`In likert format:
Warning: Values from `value` are not uniquely identified; output will contain list-cols.
* Use `values_fn = list` to suppress this warning.
* Use `values_fn = {summary_fun}` to summarise duplicates.
* Use the following dplyr code to identify duplicates.
{data} %>%
dplyr::group_by(id, item) %>%
dplyr::summarise(n = dplyr::n(), .groups = "drop") %>%
dplyr::filter(n > 1L)Warning: `cols` is now required when using unnest().
Please use `cols = c(` AI tools like ChatGPT will impact assessment methods in dental education`,
` AI tools like ChatGPT will significantly increase oral examinations in dental education`,
` AI tools like ChatGPT will likely lead to a decrease in essay assignments in dental education`,
`ChatGPT result in improved student engagement.`, `ChatGPT enhance critical thinking skills of students.`,
`ChatGPT result in an efficient grading and feedback process.`,
`ChatGPTreduce human interaction and feedback.`)`
Some correlations
Age and AI_enhance
Role and AI_enhance
dataMaid
Store the names as labels
First regularize the names
Shorten the names
Make the Codebook