notfromhere

they probably mean gpt-5-low. but the small models are bad for parsing data where the data has strong implications

In the U.S. context this impacts agencies that regulate industry and offer beneficial state services (healthcare, food inspection, pollution monitoring) and not services that impact state control (DHS, the military, or domestic secret police (FBI)).

The longstanding goal has been to undo the new deal

Gpt5 writes clean, simple code and listens to instructions. I went from tons of Claude APi usage to usage to basically none overnight

if you have the right to travel between internal states then there should be no reason you can't travel to other sovereign states.

I don't think they're offering a VSCode fork, esp when the copy you pasted says 'move beyond the IDE'.

Seems more like an agent layer on top of their existing CDE product. Which is a bit different than a cursor/windsurf/etc clone.

Literally no credible research supports your conclusions. The only one being ideological here is you.

Drug responses can be weird. Plus if you’ve been told this will make you manic and hyperfocused, your body will respond accordingly even if biochemically that doesn’t make a lot of sense.

Its also high quality and their products last a long time

I don’t understand how this works when legally in many states they can only dispense a 30 day supply

You aren’t really going to get addicted to therapeutic doses. Recreational doses are like 5-10x what a doctor would prescribe, with restrictive laws in the U.S. you can only get a months supply at a time.

Would be pretty dumb to use your months dosage for 3 days of partying

Eh. When you take it the first time the euphoria is more that you can focus and the mind becomes quieter. It’s not like a party drug

you're doing a lot of imagining all over this thread. none of it is really based in any kind of objective reality.

i'm confused as to why you think you're suddenly the arbiter of the lived experience of others? all of the questions you're asking are things that are easily found online, but something makes me think you're not after knowledge.

100 years ago people smoked like chimneys and were basically self-medicating with nicotine. The classic 'coffee and a cigarette' breakfast is the most ADHD-coded thing imaginable.

Do you have ADHD? Because as someone that has it, it's definitely very much something that heavily impacts the individual with it, and not just society.

It's hard to consistent do things, form habits, maintain attention. You have almost a lack of object permanence, a hard time remembering long-term memories, etc. There are so many problems it causes that aren't even occupational related but lead to negative outcomes outside of your personal engagement in social and economic environments.

Yes, society isn't really designed for folks with ADHD, but also the prevalence of ADHD was somewhat disguised by the fact that mental health had a social stigma and that smoking was incredibly common and people were basically microdosing stimulants every hour as such. It's not that shocking to think that the decline in smoking made adhd much more apparent.

Why should I relieve you of the joys of self-directed learning?

https://en.wikipedia.org/wiki/Attention_deficit_hyperactivit...

Too many psychiatrists do very little continuing education in their practice. I have heard more than a few try to asset that ADHD can only be diagnosed for children.

As if a neurodevelopmental disorder just magically vanishes when you hit age of majority.

It's pretty wild that despite it being a disorder that has been documented for hundreds of years, people still make the argument that people are just lying.

in my own family line it is painfully obvious that it's a heritable disorder.

You may have misunderstood my comment because I'm arguing that openai's valuation implies a level of value capture that doesn't look realistic.

ADHD is a neurodevelopmental disorder that's highly inheritable.

And yet, study after study doesn't actually show any of that. What happens is that ADHD tend to progress as you get older

honestly I only find that happens when the medication is wearing off. I'm much more spatially aware when its working and that leads to fewer accidental injuries, I see a huge difference in my driving, parking, etc.

US has highly automated manufacturing but only in a few industries where they're still competitive or have trade protections.

Though I feel like generally the narrative being spun for folks is that factory jobs largely got automated, which is true only in the fact that the remaining factory jobs got automated. The vast majority of them went overseas because it was cheaper. Otherwise the country wouldn't be filled with rusting factory towns and you'd still be able to find lots of household goods made in the US.

The chemical thing kinda worked out. Eastman Chemical is the spinout from that and they're a $9B annual business.

The folks running Kodak kind of forgot that Kodak was really a chemicals company that supported photography, not a photography company. Hence why the pivot into that was ill-fated and doomed.

the magic quadrant is essentially a stack ranking of 'how much money are you paying us vs how much money we think we can get out of you'.

that's how you end up in scenarios where some shit IBM product is leading the chart against its objectively superior competitors.

Eh. They're not wrong. A lot of folks still pay Gartner money to be on their lists, but it's more of a feeling that you have to, and not because it actually leads to any results.

Having worked in both corporate and startup worlds, I've rarely seen anyone under 40 reference a Gartner report as credible or actually use that as a source of information. Everyone knows it's pay-to-play, not particularly credible, and as the younger generations age into these very senior roles, I have no doubt that Gartner will lose a lot of relevance.

Given that trust in "mainstream media" has pretty much collapsed everywhere, I don't really doubt that this will inevitably hit the obvious corporate gatekeepers as well. Enterprise/b2b is just 10-20 years behind on trends experienced elsewhere.

Sure but so do like a dozen other companies. Given that models bump past each other every few months, I haven’t seen anything that says they have any kind of competitive advatange.

Maybe the better analogy for LLM businesses isn’t SaaS but more like power generation.

If AI really becomes that ubiquitous then OpenAI capturing that value is no less ridiculous than ComEd capturing the value of every application of electric power.

local-first sandboxed envs for AI coding seems like the right idea, esp. compared to shipping all your code/context off to openai/anthropic. my main concern would be infra overhead. enterprise setups always end up messy, but if they get the ux/integration right, could actually make a difference for compliance-heavy orgs

You'd still use RAG, just use MCP to more easily connect an LLM to your RAG pipeline