In one sentence
DALL-E is an impressive tool, but in 2022 it could not replace an illustrator’s ability to translate ambiguous client goals into precise, consistent, professional images. Even if AI eventually surpasses humans technically, illustration and art may persist because people value human intention and excellence, not just outcomes.
Overview
Emmanuel, a professional illustrator and storyboard artist, compares DALL-E’s output with work he normally creates for advertising clients. He tests a Citroën Ami storyboard frame and a personal vaporwave-style illustration. DALL-E produces visually interesting results, but struggles with exact composition, object identity, character actions, continuity, expression, style, and the unstated details an experienced illustrator supplies instinctively. He concludes that the tool might help with moodboards, but not yet with professional storyboarding. He then argues that photography, digital art, chess, martial arts, and music show how human practices can survive technologies that outperform people in narrow capabilities.
Core ideas
Professional illustration is specification, not just image generation
A client needs a drawing that communicates a particular camera angle, action, expression, composition, object, and story beat. DALL-E could produce plausible images, but not reliably satisfy all those constraints at once.
Experts carry tacit knowledge
The author initially under-described details that felt obvious to him: smiling, looking forward, who was driving, and seatbelts. An illustrator automatically handles such requirements; a text model needs them explicitly stated, and even detailed prompts may not guarantee compliance.
Continuity is a major bottleneck
A storyboard is a sequence, not a collection of attractive single images. The same characters, vehicles, designs, and spatial relationships must remain coherent across multiple angles and scenes. The author sees this as especially difficult for generative tools.
Speed comparisons favor the human in this test
He estimates that generating and prompting several DALL-E images took 10–15 minutes without producing a usable professional result. Drawing his own frame took roughly 30 minutes, including a sketch for client validation and revisions—making the human workflow more valuable than the raw generation time suggests.
Client value is precision and communication
Advertising campaigns may cost hundreds of thousands or millions of dollars. The author doubts serious clients will trade dependable visual communication for modest savings when storyboards must align directors, agencies, camera crews, effects teams, and other stakeholders.
A tool does not confer artistic judgment
Owning a powerful camera does not make someone a professional photographer. Likewise, access to DALL-E does not automatically provide composition, taste, narrative understanding, or the ability to interpret a client’s intent.
Human practices can survive superior machines
Chess engines are far stronger than human players, yet human chess remains popular because people enjoy human competition and achievement. The author expects a similar pattern in art: machine capability may expand the field without eliminating the appeal of human creation.
Art is defined by intention as well as output
The closing claim is philosophical: art depends on a will to create, while AI is a tool without independent intention. Technology can transform or extend artistic practice, but cannot by itself erase the human desire to make art.
Practical takeaways
- When evaluating image generators for professional work, test them against exact client constraints—not merely whether the outputs look attractive.
- Separate single-image ideation from production requirements such as continuity, revisions, consistent characters, and precise staging.
- Treat prompting as a specification task: list actions, relationships, expressions, camera position, object details, and intended communication.
- Use generative images cautiously for moodboards or brainstorming, while retaining human oversight for client-facing clarity and narrative coherence.
- Remember that automation may reduce the value of some tasks without eliminating the broader profession; the surviving value may shift toward direction, selection, editing, and interpretation.
Caveats and counterpoints
- The essay is an August 13, 2022 snapshot of early DALL-E, not an assessment of current image-generation systems. Its technical conclusions should not be generalized unchanged to 2026.
- The author tests a small number of personal examples and does not conduct a controlled comparison across models, prompting methods, workflows, or professional illustrators.
- The argument focuses on advertising storyboards and authored illustration. Other areas—such as generic concept exploration, stock imagery, social content, or low-budget production—may face greater automation pressure.
- The claim that expensive clients would not replace illustrators is economic speculation, not proof. Cost, deadlines, and acceptable quality vary considerably by client.
- The human-value argument explains why art may survive culturally; it does not guarantee that every illustrator’s employment, rates, or volume of work will remain unchanged.
Questions worth revisiting
- Which parts of illustration work are most valuable because they require judgment rather than rendering skill?
- How much consistency and editability does a client need before a generated image becomes production-ready?
- Could AI shift illustrators toward creative direction, visual development, and client interpretation rather than eliminate them?
- Does the audience’s interest in human creation preserve the artist’s livelihood, or only preserve a niche cultural role?
- Which of the author’s 2022 objections—prompt precision, continuity, style, or client trust—has changed most significantly since publication?
Return to this when…
Return to this essay when thinking about the difference between generating plausible images and solving a professional visual-communication problem. It is especially useful as an early, firsthand account of why impressive demos did not automatically equal reliable replacement of illustrators.