The Situational Ethics of Stickers With Charity Majors  🔗 **Source:** [The Situational Ethics of Stickers With Charity Majors](https://www.youtube.com/watch?v=HVk6RRR-8Hw) > [!note]+ Document Notes > tags: [ai-ethics, generative-art, training-data-consent, design-vocabulary, purity-culture] > author: Charity Majors > source: O'Reilly (YouTube) > reference: https://www.youtube.com/watch?v=HVk6RRR-8Hw > date created: 2026-10-15 17:46 > > > [!info] Generated analysis — not the author's words. Sections below are machine-written > > from the captured source. Only text inside `> [!quote]` blocks is verbatim from the author. > > Do not quote or attribute any other line to the named author. > > # 🏷️ Title: The Situational Ethics of Stickers with Charity Majors > <!-- deep-read:sections start --> ## 🧭 Table of contents - [📝 Summary](#-summary) - [📖 Main content](#-main-content) - [🗒️ Notes](#️-notes) - [📚 References](#-references) - [🗣️ Paraphrase](#️-paraphrase) - [📰 Newsletter](#-newsletter) - [✅ Action items](#-action-items) - [🧪 Hypotheses](#-hypotheses) - [📊 Methodology](#-methodology) - [❓ Questions](#-questions) - [🔮 Future research](#-future-research) - [💡 Implications](#-implications) - [🧠 Learning styles assessment](#-learning-styles-assessment) - [📊 Accessibility assessment](#-accessibility-assessment) - [🌍 DEIB assessment](#-deib-assessment) - [✅ Logical validity](#-logical-validity) - [🛡️ DARVO analysis](#️-darvo-analysis) - [🎯 Conclusion](#-conclusion) - [🔄 Refactor appendix](#-refactor-appendix--the-situational-ethics-of-stickers-with-charity-majors) ## 📝 Summary - **What happened.** The O'Reilly channel published a talk titled "The Situational Ethics of Stickers with Charity Majors." The metadata file has no duration and no upload date (oEmbed only), so both are unknown here. The last transcript paragraph starts at 00:09:33, so the video runs at least 9:33. The speaker is identified only by the video title; the transcript never states a name. - **The claim.** The speaker started using AI image tools late, after refusing them over training-data theft, then got hooked and now won't pay a designer again. The speaker's ethical position: the training was probably wrong and the law is unsettled, but abstaining is a "purity" move, and people in tech have an "absolute obligation" to master the tools and help govern them. - **Why it matters.** It's a worked example of a person who shares the objections of anti-AI creators and still argues for use. The argument rests on three separations: law versus morals versus ethics, personal ethics versus universal rules, and participation versus abstention. - **Key stakeholders.** Artists whose work was scraped for training (named as unpaid and unacknowledged), for-profit AI companies, the speaker's "artist friends" and "lefty friends," designers who charge hundreds of dollars an hour, and the Supreme Court as the eventual legal referee. - **Risks and tradeoffs.** The talk concedes harm ("It feels very wrong") and then argues for continued use without saying whether any compensation to artists follows. The purity framing can read as a dismissal of people with concrete grievances. The legal claims are hedged correctly ("we don't know yet"), but the prediction that creators won't prevail ("I doubt they will") carries no stated basis. - **Bottom line.** The most useful part is the vocabulary-building account (00:01:28 to 00:02:53): book covers as complete visual languages that let a non-designer learn why a design works. The ethics argument is candid and personal, but it leaves the compensation question open. ## 📖 Main content **The arc.** The talk has three parts. Part one (00:00:00 to 00:04:18) is a personal story: a lifelong love of color, failed attempts to learn Photoshop, and a refusal to touch generative image tools out of solidarity with artist friends who saw the training as theft. A deadline broke the refusal. Early results were "terrible," but the speaker got hooked and started studying typography, visual languages, color palettes, and layout by feeding designs to a chat model and asking it to describe them. **The vocabulary argument.** The speaker has never been able to tell designers why they like option A over option B. Working with a designer billing "hundreds of dollars an hour" discourages lingering on that question. Working alone with the tools removes the cost of asking, so the speaker built a vocabulary: "entire taxonomies of ribbons and swishes and curls." Book covers serve as seeds because each is a complete, coherent visual language. A cover designed by Joanna O'Neal (name as transcribed; spelling unverified) seeds every image in the talk. **The ethics argument.** At 00:04:18 the speaker says ethics are "deeply personal" and rejects relativism ("I'm not saying that everything is relative"). The position has four steps: - **Legality is unknown.** The laws weren't written for this use, case law is narrow and contradictory, and anyone claiming certainty is "selling something full of smoke." - **Rightness is a separate question.** Whatever the court decides, it "feels very wrong" for for-profit companies to train on everyone's work and sell it back with no compensation or acknowledgement. - **Law, morals, and ethics differ.** The speaker hopes creators win in court and doubts they will. - **Abstention is the wrong response.** Original sin isn't unique to AI. Those with relevant skills have an obligation to learn the tools and shape their governance, because otherwise "the people with the fewest morals" do. **The purity turn.** At 00:07:20 the speaker ties the abstention argument to a personal history: raised as a fundamentalist on the right, taught that purity means leaving the schools, health care, and courts. The speaker has "devoted my entire adult life to the opposite proposition" of interdependence. Purity-seeking, the talk argues, becomes "narcissism and performance art" that centers the purist rather than the problem or the people harmed. The same impulse is said to fuel fundamentalism on both left and right. **The close.** Support, not abstention, is "more powerful" (the transcript renders this phrase with garbled words). If you support artists, pay them. Build solutions and foster dialogue. Current frustration with slop, spam, and mandates makes this "a great time to propose new ways" of working together. The last line gives the title its hook: "stickers are pretty and art is nice." **Contested or unsourced in this source.** - The legal status of training on copyrighted work: the speaker says unknown, and no case is named in the transcript. - "I doubt they will" (creators prevailing): a prediction without stated evidence. - Whether using the tools hurts artists who would otherwise be hired: not addressed, apart from the speaker's statement that they'll never go back to paying a designer for this work. - Whether the speaker pays any artist: not stated. **ASR note.** The auto-captions render ChatGPT as "Chat TBT" and "Chief Chat DBT," drop a word in "learn the or Photoshop," garble "abstension," and split "palettes" as "pal palettes." Quotes below preserve the transcript text. ## 🗒️ Notes #### ⏱️ [00:00:00](https://www.youtube.com/watch?v=HVk6RRR-8Hw&t=0) Love of color and the early refusal - The speaker loves "all the colors" if they're saturated, spent "many hours" trying to learn Photoshop, and never made anything that looked good. - When image generators arrived, the speaker was tempted but held off because of artist and "lefty" friends upset about stolen, uncompensated training data. #### ⏱️ [00:00:44](https://www.youtube.com/watch?v=HVk6RRR-8Hw&t=44) The deadline that broke the refusal - A time crunch before an event, with the usual artist unavailable, led to trying the tools "just once." - Early results were bad. The speaker was hooked "almost immediately" and stayed up late experimenting. #### ⏱️ [00:01:28](https://www.youtube.com/watch?v=HVk6RRR-8Hw&t=88) Learning design by asking why - Method: copy designs they liked, ask the chat model to describe them, then ask why option A beats option B. - Subjects picked up along the way: typography, visual languages, color palettes, layout. #### ⏱️ [00:02:10](https://www.youtube.com/watch?v=HVk6RRR-8Hw&t=130) The vocabulary gap with human designers - Long-standing frustration: the speaker knows what they like and want more or less of, but can't say why. - Book covers are the learning tool because each is a coherent visual language. #### ⏱️ [00:02:53](https://www.youtube.com/watch?v=HVk6RRR-8Hw&t=173) Why self-service lets you linger - At hundreds of dollars an hour, a designer's client takes what's given. With the tools, the speaker can dwell on taste for free. - A favorite cover (colors, stripes, retro font; designer credited as Joanna O'Neal) seeds every image in the talk. #### ⏱️ [00:03:35](https://www.youtube.com/watch?v=HVk6RRR-8Hw&t=215) From hobby to a decision not to hire out - The speaker built "entire taxonomies of ribbons and swishes and curls." - Realized they'd never pay someone to do this again because it's too much fun, which forced the ethics question. > [!quote] [00:04:18] > The first thing I want to say about this is ethics are deeply personal. #### ⏱️ [00:04:18](https://www.youtube.com/watch?v=HVk6RRR-8Hw&t=258) Ethics are personal, but not relative - Disclaimer: not "everything is relative" and not "just your opinion, man." This is where the speaker landed, and listeners should do their own work. - Opening question: was it legal? #### ⏱️ [00:05:06](https://www.youtube.com/watch?v=HVk6RRR-8Hw&t=306) Legal or not, it feels wrong - Laws weren't written for this use case, and narrowly decided case law is partly contradictory. > [!quote] [00:05:06] > We'll find out if it was legal or not when the Supreme Court weighs in is the only honest answer. - Regardless of legality: no compensation, no acknowledgement, for-profit companies selling work back to its makers. #### ⏱️ [00:05:51](https://www.youtube.com/watch?v=HVk6RRR-8Hw&t=351) Ethics, morals, and law are three things - Many reasons to be angry at the big AI companies, including training "without permission or compensation." - The speaker hopes creators prevail in court and doubts they will. #### ⏱️ [00:06:33](https://www.youtube.com/watch?v=HVk6RRR-8Hw&t=393) The obligation argument - AI isn't "uniquely evil" and doesn't carry "original sin." Disagreement with the "don't use" conclusion. - People in tech with adjacent skills should become experts "whose opinion matters" and join governance. "If not us then who?" #### ⏱️ [00:07:20](https://www.youtube.com/watch?v=HVk6RRR-8Hw&t=440) Unilateral disarmament as an appeal to purity - The speaker labels walking away an "appeal to purity" and says they were raised a fundamentalist on the right, taught purity especially for women. > [!quote] [00:07:20] > AI is not special. AI is just technology. #### ⏱️ [00:08:06](https://www.youtube.com/watch?v=HVk6RRR-8Hw&t=486) Interdependence versus purity - Taught to deal with a corrupt world by exiting its schools, health care, and courts. The speaker has spent adulthood arguing the opposite. - Purity "slips way too easily" into self-centered performance, away from the problem and the harmed. #### ⏱️ [00:08:48](https://www.youtube.com/watch?v=HVk6RRR-8Hw&t=528) Support artists, build solutions - Fundamentalism of any political side runs on the pursuit of purity. If you say you support artists, pay them. - Current frustration with slop, spam, wasted time, and mandates makes this a good moment to propose new ways of working together. #### ⏱️ [00:09:33](https://www.youtube.com/watch?v=HVk6RRR-8Hw&t=573) Closing line - Choice framing: nobody picks a pure world, only whether to participate in the compromised one. - The talk ends on "stickers are pretty and art is nice." The transcript ends at this paragraph, so any content after 00:09:33 is missing. ## 📚 References The transcript itself contains no links; the following come from the video description or are primary sources for named entities. The description wasn't captured (oEmbed only), so nothing below comes from it. - [The Situational Ethics of Stickers with Charity Majors — O'Reilly](https://www.youtube.com/watch?v=HVk6RRR-8Hw) — the source video. Duration and upload date are unknown. - [O'Reilly Media](https://www.oreilly.com) — channel owner and publisher. - [Midjourney](https://www.midjourney.com) — named at 00:00:00 as one of the image tools whose arrival tempted the speaker. - [ChatGPT](https://chatgpt.com) — the transcript renders it as "Chat TBT" and "Chief Chat DBT." Used both to generate images and to describe designs. - [Adobe Photoshop](https://www.adobe.com/products/photoshop.html) — the transcript drops a word ("learn the or Photoshop"), but Photoshop is clearly the tool the speaker tried first. - Joanna O'Neal — credited at 00:03:35 as the designer of the seed book cover. The transcribed spelling is unverified, and the book title isn't given. - U.S. Supreme Court — named at 00:05:06 as the eventual arbiter of legality. No specific case appears in the transcript. ## 🗣️ Paraphrase The speaker spent years unable to say what they liked about a design, and stayed away from generative image tools because friends who are artists said the training was theft. A deadline pushed them into trying one tool. They got better quickly by copying designs they admired, asking a chat model to describe them, and comparing options until they had words for their taste. Book covers worked best because each cover is a finished visual language. Without a designer's hourly rate ticking, they could stop and ask why. Then came the ethical question they'd been avoiding. Their answer is explicitly personal. Legality is unsettled, and anyone claiming certainty is overselling. Rightness is a separate matter, and the training without pay or credit "feels very wrong." They separate the harm from the conclusion. In their view, abstaining doesn't fix the harm and leaves the field to people with fewer scruples, so those with the skills should learn the tools and help govern them. They tie the urge to abstain to the fundamentalist upbringing they left, calling purity-seeking a way of centering oneself. The close is practical: pay artists, build better arrangements, and accept that we only choose how to take part in a world that's already compromised. ## 📰 Newsletter **Pretty stickers, hard questions.** O'Reilly's short talk on the "situational ethics" of AI image tools is worth ten minutes if you build for the web, because the argument maps straight onto WordPress. The speaker held out against generative tools for years over artists' unpaid work in the training data. A last-minute event deadline broke the hold-out. The speaker's position now: the training probably wasn't right, the law is unsettled, and walking away from the tools still isn't the answer. The speaker calls it an obligation to learn them and join their governance. The part that travels best is the design-vocabulary story. By feeding admired book covers to a chat model and asking why option A beats option B, the speaker built a working language of typography, palette, and layout. For theme and pattern authors who struggle to brief a designer, that exercise costs nothing to try. One honest gap: the talk says "if you say you support artists, pay them," but doesn't say whether the speaker does. If your project uses generated images, the WordPress Photos and Openverse teams have a live policy question here. Link: [the video](https://www.youtube.com/watch?v=HVk6RRR-8Hw). ## ✅ Action items #### DevRel - Developer advocacy: write a talk or post on "learning a visual vocabulary with AI," using the speaker's compare-and-explain loop as the worked example, with a clear disclosure about image provenance. - Education: run a workshop where participants pick a favorite cover or interface, ask a model to describe it, then write three taste statements ("I like X because Y"). - Developer experience: add a "which images are AI-generated" metadata convention to templates and starter themes so users can disclose provenance. - Developer marketing: if AI-generated imagery appears in campaigns, state the tool and whether any artist was paid, instead of leaving it implicit. - Community management: draft a short, neutral guide for threads where artists and tool users disagree, separating "is it legal," "is it right," and "should we use it." - Internal/external/cross-project: ask peer projects how they handle image provenance and publish the comparison. #### OSPO - Policy and governance: decide whether AI-generated images are allowed in project assets and docs, and write down the rule and its exceptions. - License compliance: record each image tool's output-ownership and training terms before generated assets ship in a GPL-licensed project. - Security management: no direct action from this source; nearest relevance: uploads of generated images should pass the same malware and metadata checks as any other media. - Software inventory: list image-generation services used in the build or design process as third-party dependencies. - Developer support: no direct action from this source; nearest relevance: a help page explaining how to credit and compensate artists for commissioned work. - Community engagement: if a project funds artists, publish the amounts and selection method. - Strategic alignment: set a KPI for money paid to human creators versus spent on generation, so "support artists, pay them" is measurable. #### Community - Maintainers: no direct action from this source; nearest relevance: a contribution policy line on AI-generated images in pull requests. - Contributors: add a note on how to disclose AI assistance in design contributions. - Sponsors: consider a sponsorship track that pays illustrators for project art. - Ecosystem: convene a cross-project conversation on provenance labels for generated media, with artists at the table. #### WordPress - Core: no direct action from this source; nearest relevance: whether default media metadata can carry an "AI-generated" flag. - Core Performance: no direct action from this source. - Core AI: include training-data provenance and artist compensation in the responsible-AI guardrail discussion, as distinct from legality. - **Core Program Team:** note that the speaker separates law, morals, and ethics; use that split when scoping any AI image feature discussion. - Mobile: no direct action from this source. - Playground: no direct action from this source. - Community: host a meetup session on the purity-versus-participation debate, with moderation rules set in advance. - Training: draft a lesson on building a taste vocabulary (typography, palette, layout) from existing designs. - Marketing: if campaigns use generated art, add a one-line credit policy and pay commissioned illustrators when budgets allow. - Support: no direct action from this source. - Design: run a vocabulary exercise on block-theme patterns: pick three, write why each works, and turn the language into design-token naming. - Documentation: add a short "images and provenance" page to the style guide. - Photos: write a policy on AI-generated submissions, covering disclosure, labeling, and rejection criteria. - TV: no direct action from this source. - Polyglots: no direct action from this source. - Accessibility: no direct action from this source; nearest relevance: generated images still need alt text. - Meta: no direct action from this source. - CLI: no direct action from this source. - Hosting: no direct action from this source. - Tide: no direct action from this source. - Openverse: decide how to label and filter generated works, and whether a "generated" license-compatible category exists. - Themes: no direct action from this source; nearest relevance: bundled theme imagery should have clear provenance. - Plugins: no direct action from this source; nearest relevance: plugins that generate images need a documented training and data policy. #### The WPOCC - Programs: consider an artist-compensation microgrant track if The WP Open Community Collective funds creative work. - Projects: no direct action from this source. - Funding experiments: pilot a transparent ledger showing payments to human creators alongside any generation spend. ### FAIR - Technical independence: no direct action from this source; nearest relevance: provenance metadata for media that travels with the asset. - Policies and covenants: the "pay artists" statement maps to funding transparency; consider a covenant line on creator compensation. - Productized governance artifacts: add a "generated media provenance" field to operator scorecards. ## 🧪 Hypotheses 1. A person who explains their design preferences by comparing two options and asking a model "why" builds a usable design vocabulary faster than one who reviews designs without a comparison prompt. 2. Describing a position as "personal" and "where I've landed" (00:04:18) lowers audience defensiveness compared with stating it as a universal rule. 3. Audience members who hold a strong abstention view rate the speaker's "purity" framing as less persuasive than members who hold no prior view. 4. A requirement to publish compensation paid to artists changes how often projects adopt generated imagery. 5. Skilled practitioners who refuse a technology have less influence on its governance than practitioners who use it, holding experience constant. 6. Anchoring a design vocabulary to existing coherent artifacts (book covers) produces more consistent output than starting from abstract style words. ## 📊 Methodology - **H1.** Randomized study with 40 novices split into a compare-and-explain group and a view-only group. After two weeks, both groups describe 10 unseen designs. Blind raters score vocabulary precision on a rubric. - **H2.** A survey experiment shows two scripts of the same ethical argument, one framed personally and one framed as a rule. Measure agreement and reactance on a validated scale. - **H3.** Pre-screen viewers on prior stance toward AI imagery, show the 00:07:20 to 00:08:48 segment, and compare persuasion scores across stance groups. - **H4.** A natural experiment across projects that adopt a compensation-disclosure policy, comparing generated-asset share before and after, with matched control projects. - **H5.** Track participation in AI governance forums (comments, drafts, committee seats) by stated tool use, controlling for tenure. - **H6.** A within-subject comparison of generated designs seeded by a book cover versus a style-word prompt, scored for internal consistency by blind designers. - **Controls.** Pre-register thresholds, use blind rating, and report effect sizes, not only p-values. ## ❓ Questions - Does the speaker pay any artist now, and if so, how much and how selected? - Which tool did the speaker use, and what are its stated training sources? The transcript says only "Chat TBT" and Midjourney. - What is the basis for "I doubt they will" on creators winning in court? - Does the "obligation to participate" apply to people outside tech, such as illustrators who object? - Where does the speaker draw the line between abstention as purity and abstention as principled refusal, such as a boycott with a stated demand? - If the speaker's former designers lose work, does the "pay them" principle apply to them? - Which "solutions" does the speaker propose beyond "foster dialogue"? - Does the personal-ethics framing still hold for someone who can't make an informed choice about their own work being used? ## 🔮 Future research - **Compensation models.** No source here evaluates concrete payment schemes (licensing pools, opt-out registries, per-use royalties) against the "pay them" principle. - **Vocabulary-building efficacy.** The speaker's learning method is anecdotal. A controlled test would show whether it transfers beyond one learner. - **Purity framing in tech ethics.** Research on whether the purity-versus-participation split predicts behavior in other technology disputes would test the claim that purity-seeking "fuels every single fundamentalism." - **Governance participation.** Evidence that early adopters gain governance influence is asserted, not shown. - **Law as it settles.** The speaker's "we don't know yet" will age. A follow-up after the courts rule would show which claims held. ## 💡 Implications - **Practice.** Designers and developers can build a taste vocabulary by comparing paired options and asking for reasons. The method costs time, not money. - **Policy.** The law, morals, and ethics split suggests project policies should state which of the three each rule rests on, so disputes don't collapse into legal arguments. - **Security.** No direct implication from this source. - **Education.** Book covers and other finished visual languages work as teaching seeds for non-designers. - **Funding.** "If you say you support artists, pay them" sets a testable standard. Projects that use generated art can publish payments to human creators as evidence. - **Community.** Abstainers and users need a shared framework for disagreement. Labeling a position as personal, as the speaker does, gives it a lower-friction opening. ## 🧠 Learning styles assessment - **Reading level estimate.** Flesch-Kincaid Grade Level of about 6 to 7 by hand estimate (short sentences, mostly one- and two-syllable words), not computed with a tool. Spoken, conversational English with a few abstract terms ("taxonomies," "original sin") that raise the effective level slightly. #### 👀 Visual learners - Present: the talk is about visual language and uses a book cover as its seed image. The transcript can't confirm what's on screen. - Improve: show the before and after of an early "terrible" image versus a later one, and label the typography and palette choices on screen. #### 🔊 Aural learners - Present: a single narrator in a personal, story-led register with clear signposting ("The first thing I want to say"). - Improve: a two-minute audio recap of the three-part ethics argument would help listeners who can't see the slides. #### 🗣️ Verbal learners - Present: strong definitions by contrast (ethics, morals, and law are "three different things") and a vivid analogy to fundamentalism. - Improve: define "original sin" and "unilaterally disarm" for audiences outside the speaker's cultural references. #### 🧩 Kinesthetic learners - Present: the speaker describes a hands-on method (copy, describe, compare) that a viewer could repeat. - Improve: give a step-by-step version of the compare-and-explain exercise, with a sample prompt. #### 🧠 Logical learners - Present: a clear chain: legality unknown, rightness separate, abstention ineffective, obligation to participate. - Improve: state the premises explicitly and say what evidence would change the conclusion. The "unilaterally disarm" step skips the middle options. #### 🧑‍🤝‍🧑 Social learners - Present: the speaker invites dialogue ("Let's foster dialogue") and mentions friends and peers throughout. - Improve: add a discussion prompt and a moderated space for the artists the talk discusses. #### 🧘‍♀️ Solitary learners - Present: a reflective, first-person account that works as a prompt for self-examination. - Improve: close with two reflection questions, such as "what would change your mind?" and "whom do you pay?" #### 🌿 Naturalistic learners - Present: color and saturation come up as personal passions, but no environmental thread appears. - Improve: the energy and resource cost of image generation goes unmentioned and could be raised alongside the ethics. ## 📊 Accessibility assessment Judged as published: auto-captions only (per the transcript), no chapters in the metadata, and no description captured. #### Screen reader semantics - Present: YouTube supplies the page landmarks and player controls; the transcript is machine-readable. - Improve: visual content (the book cover, designs, any slides) isn't described in the narration, which fails WCAG 2.2 AA 1.1.1 (Non-text Content) and 1.2.5 (Audio Description, Prerecorded) for anyone who can't see the screen. #### Keyboard navigation - Present: the YouTube player is keyboard operable with visible focus. - Improve: no barriers added by the content. 2.1.1 (Keyboard) and 2.4.7 (Focus Visible) pass via the platform. #### Contrast and readability - Present: unverifiable from a transcript. - Improve: any slide text over saturated color palettes, which the speaker loves, risks failing 1.4.3 (Contrast Minimum). Check each slide with a contrast tool. #### Text resizing and reflow - Present: the player scales and captions resize. - Improve: burned-in slide text doesn't reflow, which fails 1.4.4 (Resize Text) and 1.4.10 (Reflow) for text baked into the frame. #### Media and captions - Present: auto-captions exist, and this analysis comes from them. - Improve: the captions garble ChatGPT ("Chat TBT," "Chief Chat DBT"), drop a word ("learn the or Photoshop"), and break "abstension." Proper nouns like Joanna O'Neal are unverifiable. That's a substantive 1.2.2 (Captions, Prerecorded) gap. No chapters exist in the metadata, so 2.4.5 (Multiple Ways) is weak for navigating a long talk. #### Forms and interactions - Present: no forms in the video. - Improve: not applicable in-video. Any comment prompt on the page needs labeled inputs per 3.3.2 (Labels or Instructions). #### Standards reference - Primary gaps against WCAG 2.2 AA: **1.2.2** (inaccurate captions on names and tool terms), **1.2.5** (no audio description for visuals), **1.1.1** (images conveyed visually only), **1.4.4 / 1.4.10** (baked-in text). Caption correction is the highest-impact and lowest-cost fix. ## 🌍 DEIB assessment #### Cultural breadth - Present: the speaker draws on a US fundamentalist upbringing and left-right political framing. - Improve: add perspectives from artists outside the US and from regions with different copyright regimes. Training-data law varies by country, and the talk only references the US Supreme Court. #### Inclusive language - Present: direct, second-person, and no slurs. The speaker acknowledges that purity norms fell hardest on women. - Improve: "lefty friends" is a casual shorthand that may read as dismissive. Naming what those friends objected to (training without consent) is more precise. #### Neurodiversity support - Present: a clear three-part structure (story, ethics, call to action) and plain sentences. - Improve: the transcript shows many false starts and repeated phrases, which can be hard to parse in audio. Chapter markers would help navigation. #### Representation - Present: one speaker, who foregrounds a woman designer (Joanna O'Neal) and a personal perspective shaped by gender-related purity norms. - Improve: no artist whose work was scraped speaks in the transcript. A direct voice from an affected creator would balance the argument. #### Psychological safety - Present: the speaker frames the ethics as personal, invites others to "do your own work," and names the harm openly. - Improve: "narcissism and performance art" applied to abstainers may land as a judgment on people with real grievances. Separating principled refusal from performative refusal would reduce that risk. ## ✅ Logical validity #### Strong reasoning - **Conceptual distinction, 00:05:51.** "Ethics, morals, and the law are three different things." The separation lets the speaker admit legal uncertainty and moral discomfort at once without contradiction. To strengthen it, define each term in one line. - **Epistemic honesty, 00:05:06.** "Anyone who says they know is selling something full of smoke." Declining to claim legal certainty is the correct stance on unsettled law, and it's backed by a mechanism: the laws weren't written for this use and case law is narrow. - **Counterfactual argument, 00:06:33.** "If not us then who?" has real force: if skilled, conscientious people abstain, governance falls to others. To strengthen it, name who the others are and show evidence that participants gain influence. - **Concession against interest, 00:05:06 and 00:05:51.** Granting that it "feels very wrong" and that creators have grounds for anger raises credibility, because the speaker is the one benefiting. - **Self-location, 00:04:18.** Presenting the view as a personal landing point avoids claiming universality and invites scrutiny. #### Fallacies - **False dilemma, 00:07:20.** "the argument that in the face of powerful new tools, we should simply unilaterally disarm and walk away" frames the options as total use or total withdrawal. Boycotts with demands, selective use, paying for licensed tools, and advocacy for opt-out all sit between. *Fix:* name the middle options and argue against them directly. - **Guilt by association / poisoning the well, 00:08:48.** "The pursuit of purity is what fuels every single fundamentalism on the left or the right" ties abstainers to fundamentalism. Some abstention is principled and aims at a concrete demand. *Fix:* separate performative purity from strategic refusal and say which one the talk targets. - **Unsupported prediction, 00:05:51.** "I doubt they will" (creators prevailing) has no stated basis, though it sits alongside an honest admission of legal uncertainty. *Fix:* give the reason or label it a gut estimate. - **Appeal to consequence, 00:06:33.** "The people with the fewest morals" taking the field if good actors leave assumes the field won't be taken regardless. *Fix:* show that abstention changes who governs. - **Unresolved tension (not a fallacy, but a gap), 00:05:51 to 00:08:48.** The speaker says the training feels very wrong and that supporters should pay artists, then never says whether payment to any artist happens. *Fix:* state what the speaker pays or will pay. - **Anecdotal generalization, 00:00:44 to 00:02:53.** One person's learning arc stands in for the claim that these tools help non-designers learn design. *Fix:* qualify as personal experience or cite a study. #### Tone of voice - Candid, warm, and confessional at the start ("Seriously. Seriously, I swear."), then argumentative and urgent at 00:06:33 ("We have an absolute obligation"), then personal and slightly defensive when the fundamentalist history comes up. The close at 00:09:33 is wry and resigned: the line about stickers lightens an argument that has just turned grave. Contractions and second person run throughout. The shift from storytelling to moral claims happens around 00:04:18 and is signposted. ## 🛡️ DARVO analysis Partly applicable. DARVO (Deny, Attack, Reverse Victim and Offender) describes a response to accusation. This talk is a self-examination answering an imagined critic, not a response to a specific accuser, so full DARVO doesn't appear. Each prong is addressed below. - **Deny:** none. The speaker affirms that the training was likely wrong and that artists have reason to be angry (00:05:06, 00:05:51). The speaker doesn't deny the harm. - **Attack:** mild. At 00:08:06 and 00:08:48, abstention is characterized as "narcissism and performance art" and linked to fundamentalism. The target is a stance and a motive, not named individuals. The attack on motive, though, risks painting people with concrete grievances as self-centered. - **Reverse victim and offender:** none. The speaker doesn't claim to be the injured party. The personal history (a fundamentalist upbringing) explains a view; it isn't offered as a grievance. - **Emotional cues and rhetoric:** the shift from confession to moral urgency ("absolute obligation") and the rhetorical "If not us then who?" carry the weight. The related fallacies are guilt by association and false dilemma, covered under Logical validity. ## 🎯 Conclusion The talk's strongest contribution is practical and personal: a method for building a taste vocabulary by comparing designs and asking a model why, with book covers as seeds. The ethics argument is honest about its limits: the law is unsettled, the training feels wrong, and the speaker says so. Its weakest point is the leap from "abstention is purity" to "we have an obligation to use the tools," which skips the middle options and never says whether any artist gets paid. **Next steps for readers:** try the compare-and-explain exercise on three designs you like and write three taste statements. Separate your own position into legal, moral, and ethical parts, as the speaker does. If you use generated images in a project, decide and publish what you pay human creators, since the talk's own standard is "pay them." For the missing pieces (tool names, the cover title, any description links), check the O'Reilly video page rather than the auto-captions. ## 🔄 Refactor Appendix — The Situational Ethics of Stickers with Charity Majors > Density: 2–5 Literature atoms (H3 pad), 0–3 Permanent candidates (H4 pad). > Emitted: 4 Literature, 2 Permanent. Resonance filter applied: the book-cover vocabulary > method stays in Notes; the four atoms below carry the argument. > > Konik claim-header rules: present tense, declarative, ≤10 words, searchable, unique. > > Atomicity: one declarative sentence per atom. "and"/"also" → split. > > Tag pool: #bmf · #addendum · #xref · #articleseed · #storystem · #share · #aphorisms ### 📚 Literature Split Pad #### [H1] Legal status of AI training remains genuinely unknown > [!quote] [00:05:06] > We'll find out if it was legal or not when the Supreme Court weighs in is the only honest answer. **Context & use**: - 00:05:06, right after the speaker says laws weren't written for this use case and case law is narrow and contradictory. - Keeps the legal question apart from the moral one, which the speaker then answers separately. **Tags**: #xref #share <!-- atomicity: 1 idea ✓ --> #### [H2] Tech workers have an obligation to master AI tools > [!quote] [00:06:33] > We have an absolute obligation to use these tools to master them to be experts to be people whose opinion matters **Context & use**: - 00:06:33, the central normative claim, offered after conceding the training harm. - The speaker's reasoning is counterfactual: if conscientious experts leave, governance falls to "the people with the fewest morals." Evidence for that outcome isn't given. **Tags**: #xref #articleseed <!-- atomicity: 1 idea ✓ --> #### [H3] Purity-seeking turns ethics into performance > [!quote] [00:08:06] > I also think that the pursuit of purity slips way too easily into a kind of narcissism and performance art **Context & use**: - 00:08:06, tied to the speaker's fundamentalist upbringing and a life spent arguing for interdependence. - The claim targets motive and risks lumping principled refusal with performative refusal. Pair it with that caveat. **Tags**: #xref #aphorisms <!-- atomicity: 1 idea ✓ --> #### [H4] Everyone participates in a compromised world already > [!quote] [00:09:33] > We only get to choose whether or not to participate in the compromised world we already have. **Context & use**: - 00:09:33, the closing frame, just before the stickers line. - Compresses the talk's whole stance: no clean exit exists, only choices about how to take part. **Tags**: #xref #aphorisms <!-- atomicity: 1 idea ✓ --> ### 🧠 Atomic Split Pad #### [P1] Separate law, morals, and ethics before deciding **Claim** (one sentence, Courtney's voice): A decision about a contested technology goes better when I answer the legal, moral, and ethical questions separately, because each can have a different answer. **Evidence pointer**: [[#H1]] **Scope & conditions**: Holds when the law is unsettled or lags the technology. Breaks when the three questions give the same answer, or when the legal answer is settled and binding. The speaker's own use of the split ends in a personal ethic, so it doesn't produce a rule others must follow. **Links**: - [[AI ethics frameworks]] — because extends - [[Responsible AI guardrails for WordPress]] — because applies-in guardrail design for AI features - [[Purity versus participation]] — because contrasts on whether abstention is an ethical act <!-- atomicity: 1 idea ✓ --> #### [P2] Participation claims need a stated compensation mechanism **Claim** (one sentence, Courtney's voice): An argument that says "use the tools and pay the artists" holds only when it names who gets paid and how much. **Evidence pointer**: [[#H2]] · [[#H4]] **Scope & conditions**: Applies to any argument for using a tool whose creation harmed a group. Doesn't apply when the harm has no identifiable payee. The talk states the principle and gives no mechanism, so the claim stays open. **Links**: - [[Creator compensation models]] — because extends - [[Image provenance in open source projects]] — because applies-in project asset policy - [[Every claim carries a fact]] — because evidence-for the standard I hold my own writing to <!-- atomicity: 1 idea ✓ --> ### ✏️ Fleeting Queue - [F1] Confirm the speaker's identity and the event this talk came from; the transcript gives no name and the metadata has only the O'Reilly channel — https://www.youtube.com/watch?v=HVk6RRR-8Hw #fleeting #inbox - [F2] Find the seed book cover credited to "Joanna O'Neal" at 00:03:35 and verify the spelling and title #fleeting #inbox - [F3] Open question: does the speaker pay any artist, given "if you say you support artists, pay them" at 00:08:48? #fleeting #inbox - [F4] Article seed: "Legal, right, and ethical are three questions" as a template for evaluating any contested tool #fleeting #inbox #articleseed - [F5] Try the compare-and-explain exercise on three book covers and write three taste statements #fleeting #inbox - [F6] Check whether purity-versus-participation has research behind it in technology ethics before citing the claim about fundamentalism #fleeting #inbox ### → Things 3 (max 2 actions) - [ ] Find the speaker's identity, the event, and the seed book cover title for this talk and update the note — research, 15 min #things - [ ] Draft a one-page image provenance and creator-payment policy for a project that uses AI imagery — policy draft #things <!-- deep-read:sections end --> 📄 **Full text:** [[Full Document Contents/Articles/The Situational Ethics of Stickers With Charity Majors.md|The Situational Ethics of Stickers With Charity Majors by Charity Majors]] ## Highlights - [00:04:18] seriously grapple with the question of is this ethical or not. Is this is it okay? What about artists? What about ethics? The first thing I want to say about this is ethics are deeply personal. This is not the same as I'm not saying that everything is relative or everything is subjective. I'm not saying that everything is equally okay or it's just your opinion, man. I'm telling you that this is where I've landed. You'll have to do your own work and and and you should, but I'm happy to share where I've landed and and it's this. Was it legal? [00:05:06] I don't know. We don't know yet. Anyone who says they know is selling something full of smoke. The laws weren't written for this use case. Case law today has been narrowly decided. A lot of it is contradictory. We'll find out if it was legal or not when the Supreme Court weighs in is the only honest answer. But I do I will say this whether it was legal or not, I don't think it was right. It feels wrong. It feels very wrong for any any for-profit company to train their models off of everyone's work and then sell it back to us. It doesn't feel right with no compensation, no ([View Highlight](https://read.readwise.io/read/01m50s3h44ss3pb5y9na335vzn)) - Note: [H1] Legal status of AI training remains genuinely unknown - Tags: atom/literature/h1 - [00:05:51] acknowledgement. However, ethics, morals, and the law are three different things. Honestly, there's a lot what the big AI companies are doing right now that is troubling [laughter] is very much a problem. There are a lot of reasons to be angry about this, including how the data was trained on on works without permission or compensation. And I really I truly do hope that creators prevail in the court. I doubt they will, but one can hope, right? I agree that this is a huge problem, [00:06:33] but I disagree that this means that AI is uniquely evil, somehow has original sin. Uh I disagree that this means that we should not use these tools. I actually think it is the opposite. I think we have work to do. I think we have an obligation especially those of us in tech with highly relevant adjacent skill sets. We have an absolute obligation to use these tools to master them to be experts to be people whose opinion matters and to participate in their governance. We if not us then who the people with the fewest morals do we ([View Highlight](https://read.readwise.io/read/01m50s3kh6n8ptjrrr14pz4cpp)) - Note: [H2] Tech workers have an obligation to master AI tools - Tags: atom/literature/h2 - [00:07:20] just leave the court to them? No. No. AI is not special. AI is just technology. And the argument that in the face of powerful new tools, we should simply unilaterally disarm and walk away. It's an appeal to purity. And I was raised a fundamentalist on the right. I was raised to believe that purity was one of the greatest things that anyone, especially woman, could aspire to achieve. I was raised to believe that the way you deal with a corrupt world [00:08:06] full of original sin is by taking yourself out of it, out of its schools, out of its health care system, out of its courts, out of the secular world. I have devoted my entire adult life to the opposite proposition that we are all interdependent that we are all implicated. I think that unilaterally disarming in the face of powerful new tools is neither wise nor desirable. I also think that the pursuit of purity slips way too easily into a kind of narcissism and performance art where it centers ourselves in our quest for for perfection instead of centering the problem itself or the people who are ([View Highlight](https://read.readwise.io/read/01m50s3p4jhdf2s5326mr5fb02)) - Note: [H3] Purity-seeking turns ethics into performance - Tags: atom/literature/h3 - [00:08:48] harmed by it. The pursuit of purity is what fuels every single fundamentalism on the left or the right. And it may be tempting in a time like this, but it is not effective. is always more powerful than abstension. If you say you support artists, support artists, pay them, but let's grapple with the harms. Let's build better ways. Let's build solutions. Let's foster dialogue. Everyone is frustrated right now with the slop, with the with the spam, with the having their time wasted, with the mandates. It this makes [00:09:33] it a great time to propose new ways, new ideas of working together and being together. We don't get to choose to live in a pure world. We only get to choose whether or not to participate in the compromised world we already have. And stickers are pretty and art is nice. ([View Highlight](https://read.readwise.io/read/01m50s3rt7pmmbey24mna9hp5e)) - Note: [H4] Everyone participates in a compromised world already - Tags: atom/literature/h3