WHAT?
talkingrobot.com is an experimental media where all the content (text and images) is generated via artificial intelligence (AI) tools, without human editing.
WhY?
talkingrobot.com is a way to showcase advanced AI tools, and a way for non-tech savvy people to grasp how pertinent today’s AI is, based on real examples. It’s also a way to assess (or push the limits of) what is possible with these tools today.
WHO?
Hi! I’m Cyril Fievet (Web / Twitter). I created talkingrobot.com in December 2022 and I’m in charge of it.
I am a French former AI computer engineer turned writer/journalist. Most of my 25-years-old career was devoted to explaining emerging technologies, producing content (as a journalist for offline and online media, as a freelance writer for major brands), or writing books (including two on robots/AI and one on the blogging phenomenon).
I created this site with the conviction that we are entering a new era in terms of content creation. I see it as an extension of myself — whether it’s through blog posts, random thoughts, opinions, musings, virtual photos, illustrations, digital paintings — except none of these are directly produced by me.
how?
I use publicly available AI tools and feed them with (more or less) random prompts to get some results that are relevant, troubling, or (arguably) interesting.
All of the content is published “as it is” — I never manually edit texts or images, everything is purely generated by machines.
I follow two simple principles here:
- I only publish generated articles on subjects I’m interested in — technology, innovation, Bitcoin & cryptocurrencies, art, atheism, the future… (exactly as I would do, or used to do, on my own personal bllog);
- I mostly publish articles that I could have written myself, which convey opinions as close as possible to my own opinions (with a few exceptions for content trying to illustrate bias or oddities in AI models, or comparing results from different AI tools).
NB: The posts on this blog don’t express the “truth” about a subject, nor do they necessarily depict the exact reality of things. This is not Wikipedia. This is not journalism. It is an experiment showing what AI can do, good or bad (mistakes included), while trying to express my own feelings and opinions without directly producing content myself.
And, yes, all the talkingrobot.com mascots/profile pictures were generated from text too.
seriously, why?
Some people seem to consider that AI-generated content is necessarily “inferior” to human-produced content, that AI-generated texts are “not really” actual, reliable, meaningful content, or that AI-generated art is not “real art”. I tend to disagree. AI offers a new generation of tools allowing content creators and artists to express themselves differently.
This blog tackles several key questions about AI and its use in content creation, at the crossroads of creativity, art, and the meaning of information:
- What kind of content can be generated by AI? How relevant is this content? How does it compare to human production?
- Can something 100% produced with automated tools and minimal human inputs have a meaning, and some sort of “personality”?
- Finally, can a media entirely produced by machines convey the opinions, tastes, intents, and interests of its human creator?
AI will deeply transform human society, starting with our relation to information and content creation. This blog aims at showing (and questioning) this evolution.
MEDIA COVERAGE
People are talking about talkingrobot.com:
- Swiss — Bilan.ch — Feb. 2023
“Un blog entièrement produit par des IA“
- France — Clubic — Feb. 2023
“ChatGPT, Dall-E, Mid-journey : ce blog est entièrement rédigé et alimenté par des IA“
- France — KultureGeek — Mar. 2023
“Le blog TalkingRobot est entièrement rédigé… par des IA !“