Rishika M

2026 is the first year where i have not slipped up and written last year’s date even once. foreboding omen somewhat

it’s so cool that the only use case for ai is to distort everyones perception of reality. this needs to be heavily regulated immediately, especially when a government can do this

I do *not* want an AI "summary" of an email, or a book, or a life. I do not want an AI summary of a winter sky, or my father's hands, or the hope in my child's eyes. I do not want an AI summary of the human heart, or the first little shiver of lust, or the long good work of love.

As you like the pollen shots so much, here is a bunch of sunflower pollen on some hummingbird hawk moths wingscales.

Another reason I refuse to use ChatGPT and AI is because I enjoy the elation and smug superiority I feel when I figure something out myself. I can’t explain how good it makes me feel. I’ll doubt myself, then prove myself wrong, and I like that feeling. I like the challenge.

It’s not “elitism” to say people should be trained in the subject they’re pontificating about. It’s how disciplines work. You want to be taken seriously in history? Then get in the trenches: read, research, revise, be wrong, start again. That’s the job.

With all the A.I. generated imagery out there, I would like to share some of the thoughts that go into a painting created by a human being...

remember that very brief period where google searches would autopopulate a summary from Wikipedia instead of having an AI automangle a summary from Wikipedia? that was nice

This is an extremely beautiful view. Add Chennai central to the mix and 3 beautiful buildings next to each other.

Ppl don’t know how ChatGPT works. It doesn’t “know” things. It autocompletes sentences. It makes things up.

My latest #TharoorThink column mounts a robust defence of the English language in India while rejecting the Macaulay mindset. We can use English to connect to the world while refusing to allow it to disconnect us from ourselves:

reminds me of: the world is a museum of passion projects

The systems that diagnose cancer don't need to "read" stories like Harry Potter, Ironman, copyrighted novels nor artwork. Those types of systems must be calibrated specifically on cancer related data.

I wonder if OpenAI realize how many authors, artists, musicians, coders, etc also have terms of use that prohibit their work being scraped into generative AI models...

Prompting an AI isn't art. If you're only describing what you want without doing any of the actual work, that makes you a CLIENT, not an artist. It’s like ordering from a restaurant and claiming to be the chef.

AI users comparing themselves to directors really feels like a deliberate, surface level misunderstanding of the job. Guillermo del Toro was a special effects makeup designer for a decade. James Cameron is an artist. Gareth Edwards, Wes Ball, and Neill Blomkamp come from VFX, Rob Zombie comes from music, Barry Sonnenfeld was a cinematographer. Many directors write, edit, or score their own movies. John Carpenter has done all three. Many frequently wield cameras on set - David Lynch, Zack Snyder, Steven Soderberg, Alfonso Cuaron, Michael Bay, even Tarantino has. Some will say, "Yeah, exactly, I do all that now," but in reality it's mostly an editing exercise. That's the only place you can exert any real control. Up until then, you're using trial and error to trick a machine with the right set of words and inputs into giving you something approximating what you think you wanted. And even then, you're throwing out nearly all of it. There's so much more to directing than people realize or have an appreciation for, and reducing it to 'vision' and telling others what to do as if you're wielding them as tools is just not reality. Directors work very hard, but film is also a highly collaborative effort in the broadest way possible and the people working on a film outside the director exert a lot of influence on the final picture. Even the work of 'auteurs' relies on the creativity of others. With AI, the vast majority of influence on the output is not your own, and worse, you may not even know what those influences are because of the black box slot machine nature of AI systems. Many AI people don't even write their own prompts anymore, by their own admission. They'll get an LLM to flesh it out for them, or write the thing entirely based on just an idea. One person even used accessibility features to get scene descriptions from a movie to use. People aren't just cobbling together the outputs, they're cobbling together the inputs too. The process is so automated and out of one's hands that it's more consumption that creation. If the data centers or electrical grid went down tomorrow, your abilities go down with them. It's skills on subscription, they can go away or change at any time. Imagine becoming fully reliant on these things to support you and one day you just lose access for one reason or another. This goes beyond the title of director and what is or isn't art, your abilities are largely in the hands of corporations if you don't skill up outside of AI, and I honestly think that should scare people. You should be relying on yourself and your abilities, not those being sold to you by rent seeking corporations. Don't automate your creativity to the point that you've got nothing without internet access and 10 active subscriptions.

Congratulations to Ullasa Kodandaramaiah and Akhil Sadiq for their recent paper on how camouflage colour patterns influence the detection and recognition of prey by predators in "Behavioral Ecology and Sociobiology". @tvmiiser link.springer.com/article/10.100…

Twitter can be such a wholesome place :)

I love how authors in classic books spend three pages describing a street, whilst casually mentioning the death of a character in 1 line.