alex_sf
Joined 13 years ago
Karma
1,209
Submissions
410
Member since
Nov 6
Recent submissions
Comment · 5 days ago
I mean this doesn't seem very insane? Basic info about the model, the ability to actually use it. Google in particular is awful about getting access to their models from the UX side: I don't even bother trying them out unless it's via openrouter.
Comment · 1 month ago
I don't understand how to make a modern CPU. I'm not involved in the manufacturing of it. From my perspective, there may as well not be any human involvement. I can still use the resulting chip (in an larger system of other things I can't make and wasn't involved in) to argue with you on the internet.
It becomes another abstraction, really. As long as we can use it for something useful, it's still valuable.
Comment · 1 month ago
> Is this really why the rule was made? Did they really have safety of driving in mind or is this just in the interest of the profits of airlines?
Yes. There are multiple public documents showing this, ex:
> The FAA has determined it is not appropriate to mandate the use of CRSs in aircraft now. We remain concerned that if we require children under 2 years old to be in an approved restraint system (which requires a passenger seat), affected operators might find it necessary to charge a fare for transporting these children. (Currently most, if not all, operators do not charge a fare for children under 2 years old who are held in an adult's lap.) In turn, for economic reasons some adults might decide to drive in automobiles to their destinations rather than fly. The FAA is concerned because automobile injury and fatality rates are higher than aircraft injury and fatality rates. As a result, there would be a net increase in transportation injuries and fatalities as families opt, for economic reasons, to drive rather than fly to their destinations.
https://www.federalregister.gov/documents/2005/08/26/05-1678...
> * Subsidize the 2nd seat for parents with children from tax dollars
That's not something the FAA or DOT can do.
> * Criminalize many intentional driving violations
That's not something the FAA or DOT can do.
> * Build high speed rail
That's not something the FAA or DOT (unilaterally) can do.
Comment · 1 month ago
They do when appropriate after considering all concerns. The NTSB is solely concerned with safety. The best rulemaking is not literally safety-first or nothing would actually get done.
Comment · 1 month ago
This is really naive. 'Anthropic won't give them out'. It's amazing how creative AI doomers can get about the upcoming apocalypse, and can't seem to fathom that men with guns always override the opinions of nerds in SF.
Comment · 1 month ago
Nothing that is really landmark legislation. Dem legislators also didn't do anything about net neutrality. They deferred entirely to the FCC (which is largely why the rules were subsequently struck down).
Comment · 1 month ago
Do you think a private company should be able to accumulate enough wealth and power to dictate policy to the US federal government?
Comment · 1 month ago
Just to clarify:
> 2) It generates structured output natively - guaranteed to be correct
It's not guaranteed to be correct: it's guaranteed to be _formatted in a particular way_. You can get the same thing with grammars on any LLM.
Jev and Jev-like models have other advantages, but I feel like people forget grammars exist for LLMs.
Comment · 1 month ago
You totally can. The latency is just about ~3 miles per hour.
Comment · 1 month ago
I have people skills; I am good at dealing with people.
Comment · 1 month ago
The models will only get better and inference cost will go down.
I don’t see any reason to think the same thing that happens with all tech won’t happen here.
Comment · 1 month ago
I take the specifications from the customer and type them into the AI
Comment · 1 month ago
> It’s either a maintenance issue. e.g., Opus recommended and implemented a fix for a database corruption crash. This was ~400 lines of code with many moving parts. I reviewed, and found out Android Room library already handles this recovery case, and all I needed was a 10 liner PR that catches this exception and ignores it.
I understand this, but I just can't bring myself to care. I've been doing professional software work for almost two decades. These sorts of improvements/time savers are great without AI. With AI? Whatever. It's fine.
When the underlying lib has an issue, it'll be quicker to debug with the whole thing in context.
Comment · 1 month ago
Whatever values you stick in there are about as accurate as the scale is, so.
Comment · 2 months ago
Intelligence per unit of compute will almost certainly keep increasing. That doesn't mean having more compute isn't still way, way better.