@utzlei
iAccount based inSweden
About this account
- Account based in
- Sweden
- Connected via
- Web
Account-level information from X, not a live location or the device used for a specific post.
engineer. building.
🇪🇺 Stockholm
Joined August 2011
- Tweets1.6K
- Following1.2K
- Followers726
- Likes1.9K
- northvolt! vad hände?
- aa vet inte, svårt att konkurrera med kina typ
- usch vilken slapp analys
- aa vet
- men sverige finansierar väl hundratals forskare vars jobb bokstavligen talat är att studera industriell tillverkning och industripolitik
- aa
- och frågan om varför sverige och europa inte lyckades bygga en konkurrenskraftig batterifabrik borde väl vara topp 1 innan man gör nästa satsning
- aa
- jag sökte och hittade 45 sidor om detta i olika artiklar, det handlade mest om EU-stöd eller Skellefteå kommun och kopplades till fyra svenska forskare
- aa sjukt
- och fyra böcker, alla på svenska
- aa läst
- det investerades alltså 15 miljarder dollar i en av europas största industrisatsningar någonsin och 18 månader efter kollapsen verkar det inte finnas ett försök till analys kring varför
- aa
- trots att frågan är central för Europas ambition att återindustrialisera, minska det strategiska beroendet av Kina och upprätthålla långsiktig ekonomisk autonomi
- aa vet
- så vad hände?
- aa vet inte, svårt att konkurrera med kina typ
en vanlig tumregel för startups pre-AI var att investera 50/50 i produkt vs. sales/marketing/talk to users (GTM)
om man tror på denna allokering och vet att produkt nu har blivit 10x-100x snabbare att bygga
medan GTM har blivit bara 1.3x snabbare att göra (i alla fall människa till människa)
så bör tumregeln idag vara att investera 10% produkt / 90% GTM
kommer ta lång tid för bolag att internalisera detta
Elin Lütz retweeted
Very cool paper showing that government services are getting flooded because of AI... Echoing @avshah99's earlier findings!
I met a novelist today who said he writes 50,000 lines of fiction a day now thanks to AI. This is probably the limit case. He's a hotshot writer, he knows AI tools very well, and he's talking about a 12 hour day. But he's not naive. This is not 50,000 lines of slop-filled crap.
I read somewhere that spoken languages all seem to converge on roughly the same information rate (~39 bits/second)
speakers of denser languages slow down, while speakers of sparser languages speed up, and end up at about the same throughput
it made me think that even the fastest AI interfaces are inevitably running into the same human i/o limit of how quickly we can formulate or consume ideas. and that limit may not be that high.
so whether you type or ramble is perhaps secondary
One pattern I find useful for working with LLMs is a nice long ramble session. Sometimes the LLM needs more bits to understand what you're trying to achieve, but you're too lazy to type them. In these cases I like to lean back, switch to /voice and just ramble for like 10 minutes, total mess, anything goes, full stream of consciousness. Sometimes I declare it up top, something like "switching to speech recognition sorry for any typos...". Sometimes I turn it into a small interview of a few turns. But I find that the LLMs are somehow very good at reconstructing long incoherent rambles and often their echo of your own tangle of thoughts comes out quite a bit cleaner than what you started with. The result is that you improve the mind meld and have to correct things less from that point on.
last week i touched (a lot of) grass in the Faroe Islands
did not know much about this place before going which now feels strange given how special it is
a two-hour flight from Copenhagen to an 18-island self-governing archipelago with its own language, culture and deep norse gaelic roots
magical trip, thank you @byFounders