AI is not a god conjuring language out of thin air. It is a prediction machine, generating text by learning patterns from vast datasets. And those datasets are made of human sentences, written by actual people over decades. People like us.
So yes, it is entirely possible for someone to “write like AI.” Often what that accusation really means is: this is too clear, too structured, too competent to fit the low expectations we have grown accustomed to. The new sport of “spot the AI” because a paragraph does not sound like the daily sludge is lazy, reductive, and a little embarrassing.
I was writing someone I met recently and halfway through our conversation she said, “Are you writing this with AI?” A part of me wanted to be offended. Then I found myself feeling sorry for her. It sounded like no one had been writing to her with care, with love for words, with a dedication to accuracy, appropriateness, and the ordinary beauty of a well-made sentence.
Maybe because I have always had this. In university, the people I kept close were interested in words: in syntax, in not just the sense but the sound. Between friends and colleagues there were constant debates, not only about the big ideas but about the minutiae: context, form, the weight of a single choice. I remember guys like @richardalijos, with whom I wrote a weekly editorial for our Writers League publication (later folded into the Creative Writers Club). Some weeks, when Richard wrote, I would spend hours with his sentences. He wrote better than most of the students I knew, and I wanted to make sure mine was at least as good.
We did all this with hardly any internet, not to talk of AI. We wrote by hand, leafed through dusty books, then typed.
So when people wake up and tell us that using a word like “delve,” or using em dashes, or paying attention to detail makes us sound like AI, it is, to say the least, hilarious. It is also revealing. It tells you how low the bar has fallen. And I will even go further to say that if AI, with all of its dangers, ends up raising that bar for people who ordinarily do not care about words, then that is one small, grudging good to take from it.
@GaryMarcus the "memorability" of a song/book/movie is very proportional to the marketing it receives. it has always been. nothing to do with quality of content for a long time.
just pick an AI creation and pour some money on it. it will become memorable.
@EyalToledano@TaskmasterAI@github come ooon dude you have the taskmaster :))
kudos for all the great work btw, don't think I'm ungrateful. just prefer gemini :)
@BrianRoemmele as months go by, this started feeling like just clickbait.
how long will you train it? is there a timeframe?
will it be open source? what's the current status?
just showing us videos and saying `I'm training your ai on this` doesn't feel serious.
Six years ago, I walked into the lair of the tech gods - TED - to warn them democracy may not survive technology.
In fact, I was the one who almost didn’t survive.
This year I returned & called them data rapists.
Complicit in a coup.
Collaborators.
Excited to launch Browserable: an open-source and self-hostable browser automation library designed to let agents execute web tasks, track status and fetch results.
GitHub: https://t.co/c0AsCj5w1U
Watch Browserable search Amazon for a yoga mat atleast 6mm thick, non-slip, eco-friendly, and under $50.
the data, search again...)
stream your thoughts until you and all the agents reach a consensus, and start your final report
clearly separate the final report from the rest of the stream.
4/4
Did you know that @GoogleDeepMind gemini realtime streaming is much better than @perplexity_ai, chatgpt/grok deep search etc?
prompt in the reply below.
check the images to see how it works.
1/4
identifying the most prevalent sources to avoid more microplastics.
</replace with your search>
create at least 5 agents that will research different aspects of this research for at least 5 iterations with each agent (search, get data, create new searches based on ....
3/4