Andrew Hewitt
776 posts

Andrew Hewitt
@ComposerHewitt
Bafta Nominated Composer, It's What's Inside, The Double, Villains, Submarine, Mojave, Stanford Prison Experiment, Garth Marenghi's Darkplace
Los Angeles, CA Katılım Mayıs 2022
3.1K Takip Edilen423 Takipçiler
Andrew Hewitt retweetledi

This should be part of every curriculum.
cinesthetic.@TheCinesthetic
Learning the alphabet with ARNOLD SCHWARZENEGGER.
English

@GoogleAIStudio great. though element-editing needs to be able to select instantly from project's css labels.
English

@FurqanR it stopped charging for a while. but has now started again: tools, agents, discord msgs/actions, are being auto-routed through non-Qwen3.6 models, resulting in charges.
English

@DiscussingFilm struggling to compute.
and most creatives I know are struggling to survive.
English

@spyderdobro I'd add that true talent and reliability actually go hand in hand. They're not 'ethically separable', or sim. Talented ppl are auto-reliable simply cos they auto-care.
English

@buffys I'd love to know how these positions are appointed. The process, requirements, etc.
English

@porksmith congrats on the Oscars duties, Dr Sanchez. Yours, Stig B. #Darkplace
English

I asked GPT-5.4 Pro how I can use it to discover or repurpose new drugs. It was an intentionally vague prompt to see what it would say. Then, as I read through its highly insightful response, I noticed this big slap in the face, in its final paragraph, ouch! 😅:
"The big idea is not “use GPT-5.4 pro to discover drugs.” That is too mystical and too sloppy. The better idea is to use GPT-5.4 pro to compress the highest-friction cognitive work in repurposing: reconciling conflicting evidence, designing kill-shot experiments, red-teaming endpoints, and translating scattered evidence into a trial that can actually teach you something. The fastest serious system is the one where small models do the shoveling, the big model does the judging, biology kills weak ideas early, and the mini trial is designed to return a clean posterior update rather than a cloud of entrepreneurial incense.
The natural next move is to pick one disease area and turn this into a concrete candidate-ranking rubric, assay stack, and pilot-trial blueprint."
English

@feraljokes every episode of X-Files Season One went precisely as required
English





















