Low-quality AI output is common —
and it scales into millions in hidden cost

53% of AI-using workers admit sending at least some ‘workslop’: AI-generated content that looks complete but lacks substance. Each incident costs ~2 hours to resolve. Individual behavior scales into organizational cost.

A two-panel makeover redesigning an HBR bar chart on AI ‘workslop’ into a behavior-to-cost system narrative. The left panel uses a stacked composition bar to show that 64% of AI-using workers send at least some low-quality AI output, binned into four severity levels. The right panel presents a cost cascade — 2 hours per incident, $186 per employee per month, $9 million per year — using progressive typography to convey compounding organizational impact.

Original

The original visualization comes from Why People Create AI Workslop

Original visualization

Makeover

Figure 1: A two-panel data visualization titled “Low-quality AI output is common — and it scales into millions in hidden cost.” The left panel shows a horizontal stacked bar chart depicting the workforce workslop profile among AI-using workers (n ≈ 952). Segments are color-coded by severity: Heavy (7%, deep crimson), Regular (11%, burnt orange), Occasional (46%, amber), and None (36%, gray-blue). A bracket above the three non-zero segments highlights that 64% of AI-using workers send at least some workslop. The right panel presents a cost cascade in progressively larger typography: 2 hours per incident (average time to resolve), $186 per employee per month (invisible productivity tax), and $9 million per year for a 10,000-person organization (annual organizational cost). A footnote states that 17% of respondents who did not use AI at work are excluded. Source: BetterUp Labs and Stanford Social Media Lab survey of 1,150 U.S. desk workers, September 2025; Niederhoffer, Robichaux, and Hancock (2026), Harvard Business Review.

Steps to Create this Graphic

1. Load Packages & Setup

```{r}
#| label: load
#| warning: false
#| message: false      
#| results: "hide"     

## 1. LOAD PACKAGES & SETUP ----
suppressPackageStartupMessages({
  if (!require("pacman")) install.packages("pacman")
  pacman::p_load(
    tidyverse, ggtext, showtext, scales, 
  glue, janitor, readxl, patchwork
)
})

### |- figure size ----
camcorder::gg_record(
  dir    = here::here("temp_plots"),
  device = "png",
  width  = 13,
  height = 10,
  units  = "in",
  dpi    = 320
)

# Source utility functions
suppressMessages(source(here::here("R/utils/fonts.R")))
source(here::here("R/utils/social_icons.R"))
source(here::here("R/utils/image_utils.R"))
source(here::here("R/themes/base_theme.R"))
```

2. Read in the Data

```{r}
#| label: read
#| include: true
#| eval: true
#| warning: false
#| 

df_raw <- readxl::read_xlsx(
  here::here("data/MakeoverMonday/2026/MM 2026W13.xlsx")) |>
  clean_names()
```

3. Examine the Data

```{r}
#| label: examine
#| include: true
#| eval: true
#| results: 'hide'
#| warning: false

glimpse(df_raw)
skimr::skim_without_charts(df_raw)
```

4. Tidy Data

```{r}
#| label: tidy
#| warning: false

### |- left panel: bin the distribution ----
df_bins <- df_raw |>
  mutate(
    workslop_num = suppressWarnings(as.numeric(workslop_pct)),
    bin = case_when(
      is.na(workslop_num) ~ NA_character_,
      workslop_num == 0 ~ "None",
      workslop_num <= 30 ~ "Occasional",
      workslop_num <= 60 ~ "Regular",
      workslop_num >= 70 ~ "Heavy"
    ),
    bin = factor(bin, levels = c("None", "Occasional", "Regular", "Heavy"))
  ) |>
  filter(!is.na(bin)) |>
  group_by(bin) |>
  summarise(pct = sum(respondent_pct), .groups = "drop") |>
  mutate(pct_rescaled = pct / sum(pct) * 100)

### |- right panel: cost cascade values (hand-encoded, BetterUp/Stanford) ----
# Source: BetterUp Labs + Stanford Social Media Lab, n=1,150, Sept 2025
# https://www.betterup.com/workslop
# Values: 2 hrs/incident; $186/employee/month; $9M/year for 10k-person org
```

5. Visualization Parameters

```{r}
#| label: params
#| include: true
#| warning: false

### |- Colors ----
colors <- get_theme_colors(
  palette = list(
    none       = "#B8C4C8",   
    occasional = "#B8922A",   
    regular    = "#C75B2A",   
    heavy      = "#6B1F14",  
    # Supporting
    background = "#F8F5F0",
    text_dark  = "#1A1A1A",
    text_mid   = "#4A4A4A",
    text_light = "#8A8A8A",
    divider    = "#D0C8BC",
    # Annotation accent
    anchor     = "#1A1A1A"
  )
)

# Named vector for stacked bar fill mapping
bin_colors <- c(
  "None"       = colors$palette$none,
  "Occasional" = colors$palette$occasional,
  "Regular"    = colors$palette$regular,
  "Heavy"      = colors$palette$heavy
)

### |- titles and caption ----
title_text    <- str_glue("Low-quality AI output is common —\nand it scales into millions in hidden cost")

subtitle_text <- str_glue(
  "53% of AI-using workers admit sending at least some 'workslop': AI-generated content that looks complete but lacks substance.\n",
  "Each incident costs ~2 hours to resolve. Individual behavior scales into organizational cost."
)

caption_text <- create_mm_caption(
   mm_year     = 2026,
  mm_week     = 13,
  source_text = "BetterUp Labs & Stanford Social Media Lab (n=1,150, Sept 2025)<br>
                 Niederhoffer, Robichaux & Hancock (2026), Harvard Business Review"
)

### |- fonts ----
setup_fonts()
fonts <- get_font_families()

### |- plot theme ----
base_theme <- create_base_theme(colors)

weekly_theme <- extend_weekly_theme(
  base_theme,
  theme(
    # Panel
    plot.background  = element_rect(fill = colors$palette$background, color = NA),
    panel.background = element_rect(fill = colors$palette$background, color = NA),
    panel.grid       = element_blank(),
    axis.ticks       = element_blank(),
    # Text
    axis.text        = element_blank(),
    axis.title       = element_blank(),
    # Margins
    plot.margin      = margin(8, 12, 8, 12)
  )
)

theme_set(weekly_theme)
```

6. Plot

```{r}
#| label: plot
#| warning: false

### |- LEFT PANEL: Stacked composition bar ----

# Derived annotation values
sends_some_pct <- df_bins |>
  filter(bin != "None") |>
  summarise(total = sum(pct_rescaled)) |>
  pull(total) |>
  round(0)

none_pct <- df_bins |>
  filter(bin == "None") |>
  pull(pct_rescaled)

df_bins_pos <- df_bins |>
  arrange(desc(as.integer(bin))) |>
  mutate(
    x_end   = cumsum(pct_rescaled),
    x_start = x_end - pct_rescaled,
    x_mid   = (x_start + x_end) / 2
  )

p_left <- ggplot(df_bins, aes(x = pct_rescaled, y = "", fill = bin)) +
  geom_col(
    position  = position_stack(reverse = FALSE),
    width     = 0.42,
    color     = colors$palette$background,
    linewidth = 1.2
  ) +
  geom_text(
    data = df_bins |> filter(pct_rescaled >= 9),
    aes(label = glue("{round(pct_rescaled, 0)}%")),
    position = position_stack(vjust = 0.5, reverse = FALSE),
    color = "white",
    size = 4.5,
    fontface = "bold",
    family = fonts$text
  ) +
  geom_text(
    data = df_bins_pos,
    aes(x = x_mid, label = bin),
    y = 1.22,
    color = colors$palette$text_mid,
    size = 2.8,
    fontface = "plain",
    family = fonts$text,
    hjust = 0.5,
    inherit.aes = FALSE
  ) +
  annotate(
    "text",
    x = sends_some_pct / 2,
    y = 1.58,
    label = glue("{sends_some_pct}% send at least\nsome workslop"),
    size = 3.6,
    color = colors$palette$text_dark,
    fontface = "bold",
    family = fonts$text,
    hjust = 0.5,
    lineheight = 1.2
  ) +
  annotate(
    "segment",
    x = 1, xend = sends_some_pct - 1,
    y = 1.44, yend = 1.44,
    color = colors$palette$text_mid, linewidth = 0.5
  ) +
  annotate("segment",
    x = 1, xend = 1,
    y = 1.40, yend = 1.44,
    color = colors$palette$text_mid, linewidth = 0.5
  ) +
  annotate("segment",
    x = sends_some_pct - 1, xend = sends_some_pct - 1,
    y = 1.40, yend = 1.44,
    color = colors$palette$text_mid, linewidth = 0.5
  ) +
  annotate(
    "text",
    x = 0, y = 0.58,
    label = "Excludes 17% of respondents who did not use AI at work",
    size = 2.3,
    color = colors$palette$text_light,
    hjust = 0,
    family = fonts$text,
    fontface = "italic"
  ) +
  scale_fill_manual(values = bin_colors) +
  scale_x_continuous(
    expand = expansion(mult = c(0, 0.01)),
    limits = c(0, 100)
  ) +
  scale_y_discrete(expand = expansion(add = c(0.55, 1.05))) +
  labs(
    title    = "Workforce workslop profile",
    subtitle = "Share of AI-using workers by amount of workslop sent"
  ) +
  theme(
    legend.position = "none",
    axis.text = element_blank(),
    axis.title = element_blank(),
    panel.grid = element_blank(),
    axis.ticks = element_blank(),
    plot.title = element_text(
      size   = 18,
      face   = "bold",
      color  = colors$palette$text_dark,
      family = fonts$title,
      margin = margin(b = 3)
    ),
    plot.subtitle = element_text(
      size   = 10,
      color  = colors$palette$text_mid,
      family = 'sans',
      margin = margin(b = 12)
    ),
    plot.margin = margin(12, 24, 12, 12)
  )


### |- RIGHT PANEL: Cost cascade ----

y_pos <- c(0.82, 0.50, 0.14)

cascade_sizes <- c(8.5, 13, 20)

p_right <- ggplot() +
  xlim(0, 1) +
  ylim(0, 1) +
  annotate("text",
    x = 0.5, y = y_pos[1] + 0.09,
    label = "2 hours",
    size = cascade_sizes[1],
    fontface = "bold",
    color = colors$palette$text_dark,
    family = fonts$text,
    hjust = 0.5
  ) +
  annotate("text",
    x = 0.5, y = y_pos[1] - 0.02,
    label = "PER INCIDENT",
    size = 2.5,
    color = colors$palette$text_light,
    family = fonts$text,
    hjust = 0.5
  ) +
  annotate("text",
    x = 0.5, y = y_pos[1] - 0.08,
    label = "average time to resolve",
    size = 2.8,
    color = colors$palette$text_mid,
    family = fonts$text,
    hjust = 0.5,
    fontface = "italic"
  ) +
  annotate("segment",
    x = 0.15, xend = 0.85,
    y = (y_pos[1] + y_pos[2]) / 2,
    yend = (y_pos[1] + y_pos[2]) / 2,
    color = colors$palette$divider,
    linewidth = 0.5
  ) +
  annotate("text",
    x = 0.5, y = y_pos[2] + 0.09,
    label = "$186",
    size = cascade_sizes[2],
    fontface = "bold",
    color = colors$palette$regular,
    family = fonts$text,
    hjust = 0.5
  ) +
  annotate("text",
    x = 0.5, y = y_pos[2] - 0.02,
    label = "PER EMPLOYEE / MONTH",
    size = 2.5,
    color = colors$palette$text_light,
    family = fonts$text,
    hjust = 0.5
  ) +
  annotate("text",
    x = 0.5, y = y_pos[2] - 0.08,
    label = "invisible productivity tax",
    size = 2.8,
    color = colors$palette$text_mid,
    family = fonts$text,
    hjust = 0.5,
    fontface = "italic"
  ) +
  annotate("segment",
    x = 0.15, xend = 0.85,
    y = (y_pos[2] + y_pos[3]) / 2,
    yend = (y_pos[2] + y_pos[3]) / 2,
    color = colors$palette$divider,
    linewidth = 0.5
  ) +
  annotate("text",
    x = 0.5, y = y_pos[3] + 0.09,
    label = "$9 million",
    size = cascade_sizes[3],
    fontface = "bold",
    color = colors$palette$heavy,
    family = fonts$text,
    hjust = 0.5
  ) +
  annotate("text",
    x = 0.5, y = y_pos[3] - 0.02,
    label = "PER YEAR (10K ORG)",
    size = 2.5,
    color = colors$palette$text_light,
    family = fonts$text,
    hjust = 0.5
  ) +
  annotate("text",
    x = 0.5, y = y_pos[3] - 0.08,
    label = "annual organizational cost",
    size = 2.8,
    color = colors$palette$text_mid,
    family = fonts$text,
    hjust = 0.5,
    fontface = "italic"
  ) +
  labs(
    title    = "The compounding cost",
    subtitle = "Individual behavior \u2192 organizational scale"
  ) +
  theme(
    plot.title = element_text(
      size   = 18,
      face   = "bold",
      color  = colors$palette$text_dark,
      family = fonts$title,
      margin = margin(b = 3)
    ),
    plot.subtitle = element_text(
      size   = 10,
      color  = colors$palette$text_mid,
      family = 'sans',
      margin = margin(b = 12)
    ),
    panel.grid = element_blank(),
    axis.text = element_blank(),
    axis.ticks = element_blank(),
    plot.margin = margin(12, 12, 12, 24)
  )

### |- Combined Plots ----
combined_plots <- p_left + p_right +
  plot_layout(widths = c(1, 1)) +
  plot_annotation(
    title = title_text,
    subtitle = subtitle_text,
    caption = caption_text,
    theme = theme(
      plot.title = element_text(
        size        = 30,
        face        = "bold",
        color       = colors$palette$text_dark,
        family      = fonts$title,
        lineheight  = 1.3,
        margin      = margin(b = 8)
      ),
      plot.subtitle = element_text(
        size        = 11,
        color       = colors$palette$text_mid,
        family      = fonts$text,
        lineheight  = 1.5,
        margin      = margin(b = 18)
      ),
      plot.caption = element_markdown(
        size        = 7,
        color       = colors$palette$text_light,
        family      = fonts$caption,
        linewidth   = 1.3,
        hjust       = 0,
        margin      = margin(t = 14)
      ),
      plot.background = element_rect(fill = colors$palette$background, color = NA),
      plot.margin = margin(20, 20, 14, 20)
    )
  )
```

7. Save

```{r}
#| label: save
#| warning: false

### |-  plot image ----  
save_plot_patchwork(
  plot = combined_plots , 
  type = "makeovermonday", 
  year = current_year,
  week = current_week,
  width = 13, 
  height = 10
  )
```

8. Session Info

TipExpand for Session Info

R version 4.3.1 (2023-06-16 ucrt)
Platform: x86_64-w64-mingw32/x64 (64-bit)
Running under: Windows 11 x64 (build 26100)

Matrix products: default


locale:
[1] LC_COLLATE=English_United States.utf8 
[2] LC_CTYPE=English_United States.utf8   
[3] LC_MONETARY=English_United States.utf8
[4] LC_NUMERIC=C                          
[5] LC_TIME=English_United States.utf8    

time zone: America/New_York
tzcode source: internal

attached base packages:
[1] stats     graphics  grDevices utils     datasets  methods   base     

other attached packages:
 [1] here_1.0.2      patchwork_1.3.2 readxl_1.4.5    janitor_2.2.1  
 [5] glue_1.8.0      scales_1.4.0    showtext_0.9-7  showtextdb_3.0 
 [9] sysfonts_0.8.9  ggtext_0.1.2    lubridate_1.9.5 forcats_1.0.1  
[13] stringr_1.6.0   dplyr_1.2.0     purrr_1.2.1     readr_2.2.0    
[17] tidyr_1.3.2     tibble_3.2.1    ggplot2_4.0.2   tidyverse_2.0.0
[21] pacman_0.5.1   

loaded via a namespace (and not attached):
 [1] gtable_0.3.6       xfun_0.56          htmlwidgets_1.6.4  tzdb_0.5.0        
 [5] yulab.utils_0.2.4  vctrs_0.7.1        tools_4.3.1        generics_0.1.4    
 [9] curl_7.0.0         gifski_1.32.0-2    pkgconfig_2.0.3    ggplotify_0.1.3   
[13] RColorBrewer_1.1-3 skimr_2.2.2        S7_0.2.0           lifecycle_1.0.5   
[17] compiler_4.3.1     farver_2.1.2       repr_1.1.7         codetools_0.2-19  
[21] snakecase_0.11.1   litedown_0.9       htmltools_0.5.9    yaml_2.3.12       
[25] pillar_1.11.1      camcorder_0.1.0    magick_2.8.6       commonmark_2.0.0  
[29] tidyselect_1.2.1   digest_0.6.39      stringi_1.8.7      labeling_0.4.3    
[33] rsvg_2.6.2         rprojroot_2.1.1    fastmap_1.2.0      grid_4.3.1        
[37] cli_3.6.5          magrittr_2.0.3     base64enc_0.1-6    withr_3.0.2       
[41] rappdirs_0.3.4     timechange_0.4.0   rmarkdown_2.30     otel_0.2.0        
[45] cellranger_1.1.0   hms_1.1.4          evaluate_1.0.5     knitr_1.51        
[49] markdown_2.0       gridGraphics_0.5-1 rlang_1.1.7        gridtext_0.1.6    
[53] Rcpp_1.1.1         xml2_1.5.2         svglite_2.1.3      rstudioapi_0.18.0 
[57] jsonlite_2.0.0     R6_2.6.1           fs_1.6.7           systemfonts_1.3.2 

9. GitHub Repository

TipExpand for GitHub Repo

The complete code for this analysis is available in mm_2026_13.qmd.

For the full repository, click here.

10. References

TipExpand for References

Primary Data (Makeover Monday):

1. Makeover Monday 2026 Week 13: AI Workslop Survey

2. Original Article: Why People Create AI ‘Workslop’—and How to Stop It

   • Authors: Kate Niederhoffer, Alexi Robichaux, and Jeffrey T. Hancock
   • Source: Harvard Business Review, January 2026

Source Data:

3. Dataset: 2026 Week 13 — AI Workslop Survey
   • Source: BetterUp Labs & Stanford Social Media Lab via data.world/makeovermonday
   • Data includes: Distribution of respondents by self-reported share of AI-generated work sent to colleagues that is unhelpful, low-effort, or low-quality (0%–100% in 10-point bins, plus non-users)
   • Scope: 12 response categories; n = 1,150 full-time U.S. desk workers, online survey conducted September 2025
4. Supplementary Research: Workslop: The Hidden Cost of AI-Generated Busywork
   • Source: BetterUp Labs & Stanford Social Media Lab
   • Coverage: Organizational cost metrics (2 hrs/incident, $186/employee/month, $9M/year for 10k-person org); emotional impact; workslop flow direction (peer/up/down)
5. Related HBR Article: AI-Generated ‘Workslop’ Is Destroying Productivity
   • Authors: Kate Niederhoffer et al.
   • Source: Harvard Business Review, September 2025

Note: The 12-row MM dataset covers self-reported sending behavior only. Supplementary cost metrics (2 hrs, $186/month, $9M/year) are drawn directly from BetterUp Labs & Stanford Social Media Lab published findings and are not derived from the MM dataset. The 17% of respondents who did not use AI at work were excluded before rescaling the remaining distribution to 100%; all percentages shown in the left panel reflect AI-using workers only (n ≈ 952).

11. Custom Functions Documentation

Note📦 Custom Helper Functions

This analysis uses custom functions from my personal module library for efficiency and consistency across projects.

Functions Used:

• fonts.R: setup_fonts(), get_font_families() - Font management with showtext
• social_icons.R: create_social_caption() - Generates formatted social media captions
• image_utils.R: save_plot() - Consistent plot saving with naming conventions
• base_theme.R: create_base_theme(), extend_weekly_theme(), get_theme_colors() - Custom ggplot2 themes

Why custom functions?
These utilities standardize theming, fonts, and output across all my data visualizations. The core analysis (data tidying and visualization logic) uses only standard tidyverse packages.

Source Code:
View all custom functions → GitHub: R/utils