I’ve come to a conclusion on defining taste:
Knowing what represents good taste comes from immersion to tasteful ideas in that domain. It’s trained.
Knowing what sucks is a more inherent, gut feeling. This for me, is what taste means. A filter.
You need both to be world class.
Don’t get me wrong, the technological progress is impressive. But if this takes you more than 10 seconds to realize it’s AI, it’s time for an AI psychosis break. You people are too far gone🤣
Introducing Griffin, the first model to pass the video Turing test.
48% of people who talked to it live thought it was a real human. Previous systems have had a pass rate <3%. It is #1 on NVIDIA's benchmark for full-duplex AI video.
It’s the first Human Interaction Model (HIM).
@ayushswrites Hell of an opportunity, Warp’s been on a tear this year.
Been interesting watching GTM engineering evolve from SDR with leverage into building agent infra and being the flex player across RevOps.
(1) Liberty’s pedestal shouldn’t count, she’s 46m*
(2) This is embarrassing for the US, SF is due for a liberty sibling. See below:
@elonmusk@pmarca@PalmerLuckey Oh what could have been, make Prometheus happen!
@Tocelot 100%!
CoS to a post-exit founder was the best opp I could have asked for leaving grad school.
Best way to explain it: “what would a CEO do if he had an extra 60 hrs/week?”
You move across a ton of deep problems few at a time, kills the wide but thin exposure you get at big 4.
On behalf of my fellow eye-havers in tech, I propose we formally end the Hey Dudes pandemic by banishing them from office dress codes
Effective immediately, they’ll be replaced by the timeless & tasteful, suede Clarks Wallabee
You can afford, and deserve, the upgrade
See below
@BAsnath44571 Suggesting I should sell a hosted web interface for TAMage? To me this was more about sharing what helped us out than starting a business around it.
Been working on a ML project that lets GTM teams easily segment their TAM through semantic vector clusters.
https://t.co/BAOiVPe3DH open-source CLI coming soon.
Lossy compression is the culprit of most startup hiring pain
- Orgs know they need to hire before they know who
- Compress a dynamic set of needs into the nearest discrete title
- Candidates compress in reverse to fit the JD
Dice roll
Who's building a new, more nuanced system?
So we've all *quietly* agreed that AI can't write final, unreviewed drafts of anything consequential right?
(Posts, Emails, Specs, etc...)
...or are there actually daring swashbucklers out there letting their slop cannons loop around the web off-leash?
@thericebowlgirl The “everyone will vibe code a replacement for their SaaS stack” crowd has always sounded like peak bubble psychosis for 3 reasons:
1) Macroeconomic inefficiency/redundancy
2) It overestimates the avg person’s AI literacy
3) Maintenance & integration effort pulls from actual job