By Hinda Haned

Founder at Owls & Arrows | Responsible AI Consultancy

How did we get here?

Our fascination with artificially-enhanced minds isn't new. Human depictions of artificial intelligence span centuries, reflecting our evolving relationship with technology and our projections about consciousness itself.

In ancient Greece, the poet Hesiod wrote of Talos, a giant bronze automaton crafted by Hephaestus to guard the island of Crete. This mythological figure represents one of humanity's earliest imaginings of artificial beings designed to serve and protect. Mary Shelley's Frankenstein brought artificial intelligence into the modern age, exploring profound questions about consciousness, responsibility, and control. The novel established themes that continue to influence how we think about AI today. Fritz Lang's 1927 film Metropolis introduced cinema's first iconic robot: a mechanical double of a woman that both fascinated and terrified audiences. Later, Stanley Kubrick's HAL 9000 in ‘2001: A Space Odyssey’ presented AI as a glowing red eye, eerily calm and disturbingly human in its calculated responses. These cultural references show how our visions of AI have always reflected more than mere technology: they echo our deepest hopes, fears, and projections about ourselves and our relationship with our creations.

The problem with current AI imagery

The problem is, these images suggest that AI is something far-off, alien, or impossibly powerful. They reinforce a narrative of AI as either a looming threat to humanity or a magical solution to all our problems. What they rarely show is the real face of AI: messy data pipelines, underpaid workers collecting or annotating data, teams of developers debating model design while often having little say in the broader decisions about how their work is used, product managers juggling trade-offs and managing misaligned or resistant stakeholders, or policy advisors trying to assess risks against business incentives. In short: AI is not about machines: it’s about people, processes, and politics.

The Human element

AI systems emerge from fundamentally human processes. Behind every algorithm lies a team of people making decisions about data collection, model architecture, training procedures, and deployment strategies. These human elements: the people, processes, and politics of AI, remain largely invisible in current visual representations.

Improving how we visualize AI can transform public understanding and discourse. More accurate imagery can:

  • Ground expectations: Realistic visuals help audiences understand AI's current capabilities rather than science fiction fantasies.
  • Promote critical thinking: When people see the human infrastructure behind AI, they're more likely to ask important questions about accountability, bias, and fairness.
  • Encourage informed participation: Better imagery can help non-technical audiences engage more meaningfully in AI policy and governance discussions.
  • Reduce anxiety and hype: Accurate representations can counter both AI anxiety and unrealistic expectations by showing the technology's actual scope and limitations.

Three practical steps for better AI imagery:

If we want to move away from clichéd, misleading visuals of AI, we need practical alternatives. Here are three simple, effective ways to improve how we illustrate AI in presentations, media, and everyday communication:

  1. Choose process-specific visuals. Talking about data labeling? Use an image that shows humans annotating text or drawing bounding boxes around images. Explaining image recognition? Use visuals of labeled datasets, heatmaps showing object detection, or side-by-side comparisons of predicted vs. actual outputs.
  2. Avoid humanoid robot imagery. Robots, especially metallic, humanoid ones, are often used as stand-ins for AI. But they can mislead audiences into associating AI with sentience or general intelligence. Most AI systems today are software-based and task-specific. Focus on representations that match the domain: dashboards, neural networks, or task flows.
  3. Highlight the people behind AI. AI is made by people. Highlight the teams and roles involved in building and deploying AI systems: data scientists, annotators, domain experts, software engineers. If you're explaining model development, include visuals of team collaboration, interviews with data annotators, or behind-the-scenes shots of AI development labs. If you're illustrating AI in a business context, show how people interact with and guide the technology, whether that’s reviewing outputs, refining processes, or making final decisions.

Moving forward: Better images of AI initiative

The Better Images of AI initiative is coordinated by We and AI, a UK‑based non‑profit coalition of researchers, artists, academics, and organizations like BBC R&D and the Leverhulme Centre for the Future of Intelligence.

Their freely accessible image library offers Creative‑Commons licensed visuals designed to replace clichéd, misleading AI tropes, such as humanoid robots and glowing brains, with more accurate, inclusive, and diverse portrayals. Visit their database to download and use these images in media, education, or policy contexts, and help shape a thoughtful, grounded visual narrative around AI.

So the next time you see a humanoid robot, ask yourself: Whose version of AI is this? And what are we not seeing?

This post was first published by Maeve Hosea and Hinda Haned on Substack.