~~Long Form Highlights Thread~~
1) Investigating one of the world's most restrictive planning regimes, 3-Bed apartments in Taipei routinely sell for 20x the median annual income. Much of the city cannot be practically rebuilt, despite decaying buildings.
https://t.co/jC1YxSFcbd
@ZealandEarly In fact, its the power and capabilities that these models given to everyday people that the labs see as the greatest risk, hence the withholding of Mythos from public distribution for fear of enabling cyber attacks, but I guess I don't really understand your concern.
@ZealandEarly Well, consumers determine how AI is used, unless you're referring to concerns about censorship from the labs, and the need to mandate neutrality and open-access?
@krishnanrohit Thanks for taking the time to speak with me Rohit. I found your AI & markets research fascinating, and it was great to have the chance to discuss it with you.
Full episode here:
https://t.co/1cfWncHTNC
Even aligned AI agents can still easily end up committing fraud when working within an organisation.
Excerpt from a discussion with @krishnanrohit about his research findings, Hayek's critique of the godlike super intelligence, and the future of professional expertise.
I find it amusing that Twitter absolutely laps up Jensen's 'man of the people' night market/fried chicken PR walkabouts, but the CCP saw his charm offensive on the streets of Beijing said 'not here, son' and banned the 5090D.
Dinner with the Huangs.
Jensen joined his parents for a family meal and took a moment to share fried mantou with local media gathered at the restaurant.
@tszzl Most professionals and management teams anchor their understanding of capabilities and reliability to MS Copilot's terrible offering. No surprise these people are AI skeptics.
I cover this dynamic in detail in my review of AI in the NZ public sector:
https://t.co/aivXZXzFWY
Doubly relevant now in light of the announced staffing cuts, supposedly justified by AI efficiencies.
Opportunities for efficiencies are everywhere, but it's hard to see these being realised given my findings on current management and policy settings (see below).
@gvp324377 Yeah. Although, while Copilot is slowly getting better, even today users still have far less control over reasoning length compared with a GPT/Claude subscription, which leads users to believe AI is less capable, consistent, and reliable than it really is.
Highlights from my study of AI adoption across the New Zealand public sector:
-Usage is relatively high, but uneven across agencies
-Copilot dominates usage & shapes understanding of AI capabilities
-Unions were a barrier to adoption in 2024, but have since softened
Link below.
Thanks. Absolutely, especially given how poorly indexed much of it often is. I'm also extremely bullish on AI Agents democratization of computing capabilities (e.g. data extraction, automation of minor tasks, etc.) previously accessible only to the tiny fraction of individuals with technical ability. English is enough now, so long as you can accurately conceptualize and describe what it is you want. Have just recorded a great podcast with @krishnanrohit on the subject, which I hope to release next week.
@deanwball Lovely. Would you agree the Pastoral is the exception that proves the rule you outline? i.e. More movie than vidya, shifting dramatically into the latter mode during the 4th movement as the listener, lulled, now desperately tries to grasp onto some structure amidst the chaos.
@teortaxesTex@CNLiberalism Not communists, but a rather a bunch of American technocrats backed by Rockefeller III who got the population control ball rolling in Taiwan.
See:
https://t.co/PVuLVdBFu4
@peterrhague Presumably there was something she wanted, but was unable to communicate what it was. She'd get so worked up at bedtime sometimes that she'd sometimes only fall asleep from exhaustion, still hyperventilating. It all got much easier once she was able to talk and ask for things.
@peterrhague My daughter is one of the most stubborn people in existence. Refused to feed one day, age 2months, spent it screeching instead. Sleep deprived and frustrated, there was a moment when I wanted to throttle her, so I had to put her down in her cot and walk away until I calmed down.