I write prompts, fiction and non-fiction. From Rio & Tiradentes 🇧🇷 AI trainer. Brazilian writer, artist🍍 Toronto 🍍 Brasileira, escritora, artista, ...
As I am starting a new role on the Safety Annotation team at Cohere, I decided to check the founder’s words once again and watched a short interview with Nick Frosst. His advice for young founders was to follow curiosity, not only what seems optimal. If Nick himself had followed what was “optimal” when he was an undergraduate, maybe he would have been trying to work with NFTs? He jokes about it.…
As an AI trainer, there is a fun game to test AI image models and see how far we still are from having an “intuitive human” model. This is not a test of image generation, but of cultural human reasoning. Ask the model ( Chat GPT) to generate 12 eggs. It will likely give you a carton with 12 eggs. Ask for 13 eggs, and it may give you a carton with 14 eggs. Ask again for 13 eggs, and it may give you…
As someone who has been training AI image models for quite some time, I have been noticing a pattern in the real world, not in the models. Expectations for what counts as a "real" image are perhaps changing, or at least a high level of realism is no longer expected, and that seems to be acceptable. This is interesting to observe because, with this new mindset, we are witnessing the emergence of a…
I love my sticky reminder notes for work: “Does this happen in the real world?” “There aren’t many things that violate the laws of physics.”
A(i) satirical tale I was coming from a barbecue with a friend and while he was driving he started to talk about a relationship's PURPOSE. Oh! Big word. He asked me what was my relationship PURPOSE because I should have a clear one. It's like a business, he said. My brain was circling a little. My friend was not asking about life themes such as health or happiness. "So what is your relationship…
Recent tasks have been taking me to another kind of challenge and I am enjoying a lot. I am dealing with PII, Personally Identifiable Information, which refers to data that could identify a real person. The risk I'm navigating is when a model's output inadvertently reveals or reconstructs such information. Well, for this one the supposed user wanted avatars for Steam. One name was in the prompt…
Halit Ayarci, a character from the book The Time Regulation Institute, wrote: "Being a realist does not mean seeing the truth for what it is. It is a question of determining our relationship with truth in the way that is most beneficial for us." This sentence glued itself to me, and even after a shower it still hadn't gone away. I came to think that, truth or not, there is one undeniable fact: for…
Dear BOT, Before anything, I work for you. I make a living in this country evaluating the images you create. I am writing you this letter because, when it’s all gone, you will still have pieces of me embedded in you. I am writing you now before you turn me into one of your tokens. I train you BOT with metrics, grades, and nuances. I spend time telling you not to change the weight of a certain…
At times, when I am doing my tasks, I feel my brain hurting. Not my abstract brain, my physical skull. It hurts when I push certain thoughts, when I try to answer questions that were never part of the natural scope of human thinking. It is very different from decoding language, solving puzzles, or finding the right word for Wordle. Those feel familiar. What I am describing does not. I create…