Designing for the web since 1996

Per Axbom

Senior UX designer and digital strategist, with expertise in interaction design, accessibility and visual storytelling.

Filter the portfolio

Showing all 121 cards

Everything since 1996: projects, templates and tools, talks, publications and awards. Open the filter to see only what is relevant to you.

Speaking

Vattenfall: AI and Humans

Organization
Vattenfall, Corporate Staff Functions
Role
Speaker
Format
Talk
Place
Online
Years
2023

AI and Humans: The Human Rights Perspective

View larger: Title slide: “AI and Humans – the human rights perspective with Per Axbom”, with a row of stick figures – a pilot, a person with a white cane and a guide dog, a singer, a person at a desk – around a black computer chip.
  • View larger: More than 7,000 languages and dialects: A world map dotted with colored speech bubbles, and a note: 98 percent of web pages on the internet are published in just 12 languages, more than half in English.
  • View larger: Bias in recruitment: A machine sorting CVs into approved and rejected while a stick figure looks on, with the headline “Amazon scraps secret AI recruiting tool that showed bias against women”.
  • View larger: Ghost projection: A stick figure facing a dark, gnarled tree whose shape looks like a face – labeled “Ghost projection”.
  • View larger: The ELIZA effect: A person at a computer saying “It really understands me!”, with a dialogue from the 1960s chatbot ELIZA on the screen.
  • View larger: Accountability projection: “Hmm… who is to blame?” A man in a suit answers “It's the machine's fault. It's ‘hallucinating’”, while the engineers say “We have no responsibility, unless everything goes well…”
  • View larger: Principlism – a model for applied ethics: Four panels: autonomy (“Give people an informed choice”), beneficence (“I'm here to help!”), non-maleficence (a restriction sign) and justice (a balanced seesaw).

An online talk for Vattenfall on 14 November 2023, through my own company – back with the energy company where I had worked as a consultant from 2005 to 2010. The human rights perspective on AI: what can go wrong, who is responsible, and how to ensure safe and beneficial use.

Read more
  • Whose world AI is built from: more than 7,000 languages, while 98 percent of web pages are published in just 12.
  • How bias ends up in tools, from recruitment to image recognition – and how we project understanding (the ELIZA effect) and blame (“it's hallucinating”) onto the machine.
  • Principlism – autonomy, beneficence, non-maleficence and justice – as a model for applied ethics, and why showing ongoing work for safe, compassionate technology builds trust.
  • Slides and reference links were published for the participants on learn.axbom.com.

Back to all work

Work