library(tidyverse)
library(gganimate)
library(gapminder)
library(viridis)
library(ggplot2)
library(DataExplorer)
library(ggthemes)Project 2
Project 2: GLOBAL AI IMPACT
(McClure, 2024)
Introducing:
| Data Set Variables |
|---|
| Country |
| Year |
| Industry |
| AI adoption rate (%) |
| AI generated content vol. (TBs per year) |
| Job loss due to AI (%) |
| Revenue increase due to AI (%) |
| Human-AI collaboration rate (%) |
| Top AI tools used |
| Regulation status |
| Consumer trust in AI (%) |
| Market share of AI companies (%) |
With my major in Information Science, I aim to pursue a career in technology and AI. This data set from Kaggle immediately caught my interest because it draws from several reputable sources like the Stanford AI Index Report, MIT Technology Review, and more, and includes variables such as job loss, AI adoption rates, and AI market share.
Out of the 12 variables in the “Global AI Impact” data set, I chose to focus mainly on the job loss rate due to AI (%) and how it relates to country, industry, and year. I began by cleaning the data—checking for NA’s, renaming columns for clarity, and later narrowing it down to my key variables and working with average values to make it more manageable and focused.
Load necessary libraries
Read in AI impact data
AI_Impact<- read_csv("/Users/maishasubin/Desktop/DATA110/Global_AI_Impact_Dataset.csv")Rows: 200 Columns: 12
── Column specification ────────────────────────────────────────────────────────
Delimiter: ","
chr (4): Country, Industry, Top AI Tools Used, Regulation Status
dbl (8): Year, AI Adoption Rate (%), AI-Generated Content Volume (TBs per ye...
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.head(AI_Impact)# A tibble: 6 × 12
Country Year Industry `AI Adoption Rate (%)` AI-Generated Content Vol…¹
<chr> <dbl> <chr> <dbl> <dbl>
1 South Korea 2022 Media 44.3 33.1
2 China 2025 Legal 34.8 66.7
3 USA 2022 Automotive 81.1 96.1
4 France 2021 Legal 85.2 93.8
5 France 2021 Gaming 79.0 45.6
6 USA 2021 Retail 67.0 47.7
# ℹ abbreviated name: ¹`AI-Generated Content Volume (TBs per year)`
# ℹ 7 more variables: `Job Loss Due to AI (%)` <dbl>,
# `Revenue Increase Due to AI (%)` <dbl>,
# `Human-AI Collaboration Rate (%)` <dbl>, `Top AI Tools Used` <chr>,
# `Regulation Status` <chr>, `Consumer Trust in AI (%)` <dbl>,
# `Market Share of AI Companies (%)` <dbl>Necessary cleaning:
- Renaming variables without space
- Removing NA’s from all columns individually
# Renaming variables
AI_Impact2 <- AI_Impact |>
rename(ai_adoption_rate = `AI Adoption Rate (%)`
, ai_gen_content_vol = `AI-Generated Content Volume (TBs per year)`
, job_loss_ai_rate = `Job Loss Due to AI (%)`
, rev_increase_ai_rate = `Revenue Increase Due to AI (%)`
, human_ai_collab_rate = `Human-AI Collaboration Rate (%)`
, top_ai_tools = `Top AI Tools Used`
, regulation_status = `Regulation Status`
, consumer_trust_ai_rate = `Consumer Trust in AI (%)`
, market_share_ai_rate = `Market Share of AI Companies (%)`
, country = `Country`, year = `Year`, industry = `Industry`)
#Removing missing values: NA's
AI_Impact_nona <- AI_Impact2 %>%
filter(!is.na(ai_adoption_rate), !is.na(ai_gen_content_vol)
, !is.na(job_loss_ai_rate), !is.na(rev_increase_ai_rate)
, !is.na(human_ai_collab_rate), !is.na(top_ai_tools)
, !is.na(regulation_status), !is.na(consumer_trust_ai_rate)
, !is.na(market_share_ai_rate), !is.na(country), !is.na(year)
, !is.na(industry))Viewing my data:
# To view the list of country's I am working with
unique(AI_Impact_nona$country) [1] "South Korea" "China" "USA" "France" "Australia"
[6] "UK" "Canada" "India" "Japan" "Germany" # To view the list of industry's I am working with
unique(AI_Impact_nona$industry) [1] "Media" "Legal" "Automotive" "Gaming"
[5] "Retail" "Education" "Healthcare" "Marketing"
[9] "Manufacturing" "Finance" # To view years I am working with
unique(AI_Impact_nona$year)[1] 2022 2025 2021 2023 2020 2024Testing multiple linear regression model in relation to job loss due to AI rate
- Exploration 1: with respect to all variables (for understanding purposes, regression analysis for exploration 2 coming up)
# Run linear regression models
lm_model <- lm(job_loss_ai_rate ~ ai_adoption_rate + year + ai_gen_content_vol + rev_increase_ai_rate + consumer_trust_ai_rate + human_ai_collab_rate + regulation_status
+ market_share_ai_rate + industry + top_ai_tools + country , data = AI_Impact_nona)
# Summary of model
summary(lm_model)
Call:
lm(formula = job_loss_ai_rate ~ ai_adoption_rate + year + ai_gen_content_vol +
rev_increase_ai_rate + consumer_trust_ai_rate + human_ai_collab_rate +
regulation_status + market_share_ai_rate + industry + top_ai_tools +
country, data = AI_Impact_nona)
Residuals:
Min 1Q Median 3Q Max
-31.0968 -8.8642 -0.2018 9.0328 29.8798
Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) 1389.07122 1180.20025 1.177 0.2409
ai_adoption_rate -0.01095 0.04194 -0.261 0.7944
year -0.68092 0.58402 -1.166 0.2453
ai_gen_content_vol 0.01115 0.03590 0.311 0.7564
rev_increase_ai_rate 0.09966 0.04279 2.329 0.0211 *
consumer_trust_ai_rate 0.05837 0.06111 0.955 0.3409
human_ai_collab_rate 0.02240 0.05448 0.411 0.6815
regulation_statusModerate 3.32802 2.47297 1.346 0.1802
regulation_statusStrict 0.26367 2.64207 0.100 0.9206
market_share_ai_rate 0.07256 0.07425 0.977 0.3299
industryEducation -1.77904 4.90145 -0.363 0.7171
industryFinance 0.02543 5.09110 0.005 0.9960
industryGaming 0.14968 4.36277 0.034 0.9727
industryHealthcare -3.09022 4.84103 -0.638 0.5241
industryLegal -2.35939 4.92809 -0.479 0.6327
industryManufacturing 2.75090 4.84781 0.567 0.5712
industryMarketing -7.97446 4.63354 -1.721 0.0871 .
industryMedia -7.36095 4.17869 -1.762 0.0800 .
industryRetail -4.60081 4.57440 -1.006 0.3160
top_ai_toolsChatGPT -6.13845 4.00379 -1.533 0.1271
top_ai_toolsClaude -3.81987 4.02315 -0.949 0.3438
top_ai_toolsDALL-E -0.91187 4.17351 -0.218 0.8273
top_ai_toolsMidjourney -2.84421 3.86866 -0.735 0.4633
top_ai_toolsStable Diffusion -1.97316 4.08296 -0.483 0.6295
top_ai_toolsSynthesia 1.30563 4.17744 0.313 0.7550
countryCanada 11.57367 5.36196 2.158 0.0323 *
countryChina 8.49437 5.05364 1.681 0.0947 .
countryFrance 8.48622 4.82318 1.759 0.0803 .
countryGermany 8.35650 5.19316 1.609 0.1095
countryIndia 5.67046 4.95703 1.144 0.2543
countryJapan 9.30590 5.08581 1.830 0.0691 .
countrySouth Korea 8.00702 5.10165 1.569 0.1184
countryUK 7.38738 5.20701 1.419 0.1579
countryUSA 1.38714 4.91546 0.282 0.7781
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Residual standard error: 13.65 on 166 degrees of freedom
Multiple R-squared: 0.1957, Adjusted R-squared: 0.03586
F-statistic: 1.224 on 33 and 166 DF, p-value: 0.2046- Exploration 2: with respect to country, industry, AI adoption rate and year
# Run linear regression models
lm_model <- lm(job_loss_ai_rate ~ year + ai_adoption_rate + industry + country, data = AI_Impact_nona )
# Summary of model
summary(lm_model)
Call:
lm(formula = job_loss_ai_rate ~ year + ai_adoption_rate + industry +
country, data = AI_Impact_nona)
Residuals:
Min 1Q Median 3Q Max
-30.9961 -10.0059 0.6298 9.3155 27.4944
Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) 1273.18734 1162.47210 1.095 0.2749
year -0.61809 0.57506 -1.075 0.2839
ai_adoption_rate -0.01554 0.04140 -0.375 0.7078
industryEducation -4.01746 4.66596 -0.861 0.3904
industryFinance -0.90974 4.95570 -0.184 0.8546
industryGaming -1.08993 4.21329 -0.259 0.7962
industryHealthcare -4.34091 4.79464 -0.905 0.3665
industryLegal -2.46743 4.81411 -0.513 0.6089
industryManufacturing 2.39900 4.74296 0.506 0.6136
industryMarketing -9.15181 4.59830 -1.990 0.0481 *
industryMedia -7.30683 4.17454 -1.750 0.0818 .
industryRetail -6.04044 4.53904 -1.331 0.1850
countryCanada 12.05361 5.07190 2.377 0.0185 *
countryChina 8.91410 4.79659 1.858 0.0648 .
countryFrance 8.52955 4.62530 1.844 0.0668 .
countryGermany 8.69716 5.00203 1.739 0.0838 .
countryIndia 7.15881 4.65214 1.539 0.1256
countryJapan 7.41173 4.80797 1.542 0.1249
countrySouth Korea 7.27710 4.82130 1.509 0.1330
countryUK 6.64916 4.82398 1.378 0.1698
countryUSA 2.29563 4.72688 0.486 0.6278
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Residual standard error: 13.76 on 179 degrees of freedom
Multiple R-squared: 0.1193, Adjusted R-squared: 0.02087
F-statistic: 1.212 on 20 and 179 DF, p-value: 0.2488Exploration 2: Linear Regression analysis:
- Equation for my model:
job_loss_ai_rate = 1273.19 - 0.618(year) - 0.0155(ai_adoption_rate) - 4.017(industryEducation) - 0.909(industryFinance) - 1.090(industryGaming) - 4.341(industryHealthcare) - 2.467(industryLegal) + 2.399(industryManufacturing) - 9.152(industryMarketing) - 7.307(industryMedia) - 6.040(industryRetail) + 12.053(countryCanada) + 8.914(countryChina) + 8.529(countryFrance) + 8.697(countryGermany) + 7.159(countryIndia) + 7.412(countryJapan) + 7.277(countrySouth Korea) + 6.649(countryUK) + 2.296(countryUSA)
- Analyzing model based on p-values:
From the analysis above, industryMarketing (p = 0.0481) is statistically significant, suggesting that Marketing industry has a weak but significant relationship with job loss due to AI. Also, countryCanada (p = 0.0185) is significant as well with a p value of less than 0.05.
Rest of the variables have (p > 0.05) suggesting statistically insignificant values, which means they do not have a strong impact on job loss in this model.
- Adjusted R^2 values:
Adjusted R square = 0.02087 (low), indicating that the model explains less than 2.1% of the variation in job loss due to AI.
- Exploration 3: for Canada, since statistically significant
# Filter data for Canada only
Canada_data <- AI_Impact_nona %>% filter(country == "Canada")
# Run linear regression model
lm_model <- lm(job_loss_ai_rate ~ year + industry, data = Canada_data )
# Summary of model
summary(lm_model)
Call:
lm(formula = job_loss_ai_rate ~ year + industry, data = Canada_data)
Residuals:
Min 1Q Median 3Q Max
-11.060 -3.912 0.000 3.912 11.060
Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) 7109.292 2583.420 2.752 0.0332 *
year -3.495 1.278 -2.735 0.0340 *
industryEducation -24.398 11.338 -2.152 0.0749 .
industryGaming -10.944 10.707 -1.022 0.3461
industryHealthcare -16.270 11.392 -1.428 0.2032
industryLegal -6.562 10.681 -0.614 0.5615
industryManufacturing -2.085 13.133 -0.159 0.8791
industryMarketing -10.948 11.338 -0.966 0.3715
industryMedia 8.635 13.133 0.657 0.5353
industryRetail -34.190 13.071 -2.616 0.0398 *
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Residual standard error: 9.243 on 6 degrees of freedom
Multiple R-squared: 0.8118, Adjusted R-squared: 0.5295
F-statistic: 2.876 on 9 and 6 DF, p-value: 0.1058Statistically significant values with adjusted R^2 = 0.53, indicating that the model explains about 53% of the variation in job loss due to AI.
Exploration 3: testing visualization for Canada regression analysis
Correlation visualization for Canada to analyze the data…
# Selecting only the relevant columns for the pair plot to analyze correlation
selected_columns <- Canada_data [, c("job_loss_ai_rate", "year", "industry")]
plot_correlation(selected_columns)
Bar graph analysis over the years, according to industry…
Canada_data |>
ggplot(aes(x = year, y = job_loss_ai_rate, fill = industry)) +
geom_bar(stat = "identity") +
theme_minimal() +
labs(title = "Job loss Due to AI from 2020-2025 According to Industries")
Here we see missing data for 2024, hence not going ahead with Canada filtering.
- Exploration 4: with respect to just country and year
# Run linear regression models
lm_model <- lm(job_loss_ai_rate ~ year + country, data = AI_Impact_nona)
# Summary of model
summary(lm_model)
Call:
lm(formula = job_loss_ai_rate ~ year + country, data = AI_Impact_nona)
Residuals:
Min 1Q Median 3Q Max
-27.587 -10.044 -1.018 10.962 28.293
Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) 1455.8502 1113.6291 1.307 0.1927
year -0.7106 0.5507 -1.290 0.1985
countryCanada 12.2076 4.9642 2.459 0.0148 *
countryChina 10.1665 4.6699 2.177 0.0307 *
countryFrance 7.7135 4.5463 1.697 0.0914 .
countryGermany 8.8091 4.9065 1.795 0.0742 .
countryIndia 7.0528 4.5736 1.542 0.1247
countryJapan 7.1399 4.6383 1.539 0.1254
countrySouth Korea 6.9100 4.7200 1.464 0.1449
countryUK 8.2868 4.7178 1.757 0.0806 .
countryUSA 2.1282 4.6732 0.455 0.6493
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Residual standard error: 13.81 on 189 degrees of freedom
Multiple R-squared: 0.06236, Adjusted R-squared: 0.01275
F-statistic: 1.257 on 10 and 189 DF, p-value: 0.2576Going ahead with exploration 2 and 4, and calculating average for clear visualization
AI_means <- AI_Impact_nona %>%
group_by(year, country, industry) %>% # grouping data sets based on these variables
summarize(job_loss_ai_rate = mean(job_loss_ai_rate)) # calculating the mean for job_loss_ai_rate
AI_means2 <- AI_Impact_nona %>%
group_by(year, country) %>% # grouping data sets based on these variables
summarize(job_loss_ai_rate = mean(job_loss_ai_rate)) # calculating the mean againExtra regression analysis after calculating average job_loss_ai_rate for (plot1)…
# Run linear regression models
lm_model <- lm(job_loss_ai_rate ~ year + industry + country, data = AI_means)
# Summary of model
summary(lm_model)
Call:
lm(formula = job_loss_ai_rate ~ year + industry + country, data = AI_means)
Residuals:
Min 1Q Median 3Q Max
-31.2145 -9.5189 0.7435 8.9062 27.1043
Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) 1672.9500 1213.1147 1.379 0.1699
year -0.8166 0.6001 -1.361 0.1756
industryEducation -4.9205 4.8882 -1.007 0.3157
industryFinance -0.2458 4.8958 -0.050 0.9600
industryGaming 0.4682 4.2938 0.109 0.9133
industryHealthcare -3.0538 4.9351 -0.619 0.5370
industryLegal -3.6880 4.8936 -0.754 0.4522
industryManufacturing 4.3189 4.9680 0.869 0.3860
industryMarketing -8.8418 4.6043 -1.920 0.0567 .
industryMedia -7.4350 4.2789 -1.738 0.0843 .
industryRetail -4.5661 4.7636 -0.959 0.3393
countryCanada 12.4774 4.9138 2.539 0.0121 *
countryChina 8.6999 4.8176 1.806 0.0729 .
countryFrance 9.1434 4.5887 1.993 0.0481 *
countryGermany 8.0344 4.9729 1.616 0.1082
countryIndia 8.3348 4.7443 1.757 0.0810 .
countryJapan 6.4985 4.9279 1.319 0.1892
countrySouth Korea 7.9971 4.7397 1.687 0.0936 .
countryUK 7.1328 4.8261 1.478 0.1415
countryUSA 1.4727 4.7165 0.312 0.7553
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Residual standard error: 13.3 on 152 degrees of freedom
Multiple R-squared: 0.1487, Adjusted R-squared: 0.04229
F-statistic: 1.397 on 19 and 152 DF, p-value: 0.1357Adjusted R-square still less, explaining 4.23% of job loss. P value is also higher than 0.05: 0.1357
Another extra regression analysis after calculating average job_loss_ai_rate for just country and year for (plot2) …
# Run linear regression models
lm_model <- lm(job_loss_ai_rate ~ year + country, data = AI_means2)
# Summary of model
summary(lm_model)
Call:
lm(formula = job_loss_ai_rate ~ year + country, data = AI_means2)
Residuals:
Min 1Q Median 3Q Max
-18.3420 -6.1481 0.7986 5.1637 17.5168
Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) 1075.0872 1429.7581 0.752 0.4558
year -0.5220 0.7069 -0.738 0.4639
countryCanada 14.0157 5.4993 2.549 0.0141 *
countryChina 10.4709 5.2395 1.998 0.0515 .
countryFrance 6.2226 5.2395 1.188 0.2409
countryGermany 10.3049 5.5066 1.871 0.0675 .
countryIndia 5.7824 5.2395 1.104 0.2754
countryJapan 5.4882 5.2395 1.047 0.3002
countrySouth Korea 4.8686 5.2395 0.929 0.3575
countryUK 8.6471 5.2395 1.650 0.1055
countryUSA -0.1162 5.2395 -0.022 0.9824
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Residual standard error: 9.075 on 47 degrees of freedom
Multiple R-squared: 0.219, Adjusted R-squared: 0.05285
F-statistic: 1.318 on 10 and 47 DF, p-value: 0.2489Heat map analysis for filtered data set ‘AI_means’ to understand trends according to industry
ggplot(AI_means, aes(x = factor(year), y = industry, fill = job_loss_ai_rate)) +
geom_tile(color = "white") +
scale_fill_gradient(low = "lightyellow", high = "red") +
labs(title = "Heatmap of AI-Driven Average Job Loss by Industry and Year",
x = "Year", y = "Industry", fill = "Average Job Loss (%)") +
theme_minimal()
# To see the industry's with higher rates
AI_means %>%
arrange(desc(job_loss_ai_rate))# A tibble: 172 × 4
# Groups: year, country [58]
year country industry job_loss_ai_rate
<dbl> <chr> <chr> <dbl>
1 2021 China Gaming 49.7
2 2025 Japan Manufacturing 49.6
3 2024 South Korea Retail 49.6
4 2022 USA Manufacturing 49.3
5 2024 France Gaming 49.3
6 2021 UK Automotive 49.1
7 2021 Germany Automotive 48.5
8 2020 India Retail 48.3
9 2023 Canada Media 48.2
10 2021 India Manufacturing 47.4
# ℹ 162 more rowsHeat map analysis for filtered data set ‘AI_means2’ to understand trends according to country
ggplot(AI_means2, aes(x = factor(year), y = country, fill = job_loss_ai_rate)) +
geom_tile(color = "white") +
scale_fill_gradient(low = "lightblue", high = "darkblue") +
labs(title = "Heatmap of AI-Driven Average Job Loss by Industry and Year",
x = "Year", y = "Industry", fill = "Average Job Loss (%)") +
theme_minimal()
Final Visualizations…
Plot 1: AI-Driven Average Job Loss by Industry and Year
library(gifski) # For converting animation as GIF for rendering purposes
## Animated graph
graph1 <- AI_means %>% # using AI_mean data set
ggplot(aes(x = year, y = job_loss_ai_rate, color = industry, group = industry)) +
geom_line(linewidth = 1.2) +
geom_point(size = 3, alpha = 0.8) +
theme_fivethirtyeight() + # Theme for the plot’s overall appearance
scale_x_continuous(breaks = 2020:2025) + # Setting x-axis to display year values from 2020 to 2025.
labs(
title = "AI-Driven Average Job Loss by Industry and Year",
x = "Year",
y = "Average Job Loss Rate ",
color = "Industry",
caption = "Source: Stanford AI Index Report, MIT Technology Review, et al."
) +
theme_dark(base_size = 13) +
theme(
plot.background = element_rect(fill = "black", color = NA),
panel.background = element_rect(fill = "black", color = NA),
legend.background = element_rect(fill = "black", color = NA),
legend.key = element_rect(fill = "black"),
plot.title = element_text(color = "white", face = "bold", size = 16),
plot.subtitle = element_text(color = "white", size = 13),
axis.title = element_text(color = "white"),
axis.text = element_text(color = "gray90"),
legend.text = element_text(color = "gray90"),
legend.title = element_text(color = "white"),
panel.grid.major = element_line(color = "gray30"),
panel.grid.minor = element_line(color = "gray20"),
plot.caption = element_text(color = "gray80", size = 10, hjust = 1.5) # For making the caption visible in final plot
) +
scale_color_viridis_d(option = "viridis")
# Animate it with transition_reveal
graph1_animation <- graph1 +
transition_reveal(year) + # Gradually reveals the graph over the years
labs(subtitle = "Year: {round(frame_along)}") # A dynamic subtitle that updates for each frame
# Render the animation
animate(graph1_animation, height = 500, width = 800, fps = 10, duration = 10, end_pause = 40, res = 100, renderer = gifski_renderer("ai_jobloss_industry.gif")) # Setting the frames per second to 10, total duration of the animation 10 secs, 40-frame pause at the end of the animation before it stops, resolution of the output image at 100 DPI.Plot 2: AI-Driven Average Job Loss by country and Year
library(gifski)
# Create the base line graph
graph1 <- AI_means2 %>%
ggplot(aes(x = year, y = job_loss_ai_rate, color = country, group = country)) +
geom_line(linewidth = 1.2) +
geom_point(size = 3, alpha = 0.8) +
theme_fivethirtyeight() + # Theme for the plot’s overall appearance
scale_x_continuous(breaks = 2020:2025) + # Setting x-axis to display year values from 2020 to 2025.
labs(
title = "AI-Driven Average Job Loss by Country and Year",
x = "Year",
y = "Average Job Loss Rate ",
color = "Country",
caption = "Source: Stanford AI Index Report, MIT Technology Review, et al."
) +
theme_dark(base_size = 13) +
theme(
plot.background = element_rect(fill = "black", color = NA),
panel.background = element_rect(fill = "black", color = NA),
legend.background = element_rect(fill = "black", color = NA),
legend.key = element_rect(fill = "black"),
plot.title = element_text(color = "white", face = "bold", size = 16),
plot.subtitle = element_text(color = "white", size = 13),
axis.title = element_text(color = "white"),
axis.text = element_text(color = "gray90"),
legend.text = element_text(color = "gray90"),
legend.title = element_text(color = "white"),
panel.grid.major = element_line(color = "gray30"), # Customizing grid lines to a light gray color.
panel.grid.minor = element_line(color = "gray20"),
plot.caption = element_text(color = "gray80", size = 10, hjust = 1.5) # For making the caption visible in final plot (ChatGPT's help)
) +
scale_color_viridis_d(option = "plasma")
# Animate it with transition_reveal
graph1_animation <- graph1 +
transition_reveal(year) + # Gradually reveals the graph over the years
labs(subtitle = "Year: {round(frame_along)}") # A dynamic subtitle that updates for each frame
# Render the animation
animate(graph1_animation, height = 500, width = 800, fps = 10, duration = 10, end_pause = 40, res = 100, renderer = gifski_renderer("ai_jobloss_country.gif")) # Setting the frames per second to 10, total duration of the animation 10 secs, 40-frame pause at the end of the animation before it stops, resolution of the output image at 100 DPI.Essay in Continuation
Overall, the trend I observed in the job loss percentages across various industries from my Plot 1 was that in 2020, as expected, the job loss rates were higher compared to other years, especially when compared to 2023 and 2025, which saw recovery from COVID. Manufacturing, gaming, automotive, and retail sectors experienced consistently higher values, aligning with reports like those from Medium (Hewitt, 2024). This report highlights jobs in customer service, manufacturing, transportation, finance, and retail as being particularly vulnerable to replacement by robotics in the near future.
Looking at Plot 2 and its ‘heatmap’, there was a noticeable increase in job loss between 2020 and 2021. What stood out to me was the drastic drop in job loss in the USA in 2024, compared to a peak of about 50% job loss in Germany during the same year. However, I couldn’t find much data to support a rise in unemployment in Germany due to AI in 2024. Additionally, I wanted to explore an area chart for my plot 1 with respect to industry’s however I could not get it to work. I was also hoping to merge ‘highcharter’ with animated graph for clarity, but I couldn’t figure out how to.
In summary, countries like the USA, Australia, Japan, and Germany are seeing a notable increase in job loss due to AI, with industries such as manufacturing, automotive, and retail bearing the most impact. However, the full significance of the data is still something I need to explore further.
Bibliography
- https://medium.com/@generup22/15-industries-that-ai-will-severely-disrupt-by-2034-a3416e77b894*#:~:text=Customer%20service%2C%20manufacturing%2C%20transportation%2C**,and%20parts%20of%20government%20bureaucracy.*
- https://www.statista.com/statistics/227005/unemployment-rate-in-germany/
- https://www.wsj.com/articles/it-unemployment-rises-to-5-7-as-ai-hits-tech-jobs-7726bb1b
- ChatGPT ’s help with showing captions in my animated graph as it disappeared.
- Tutorial followed for animated graph: https://www.youtube.com/watch?v=SnCi0s0e4Io