x187463
x187463 on Hacker News
Is this basically the bitter lesson applied to the very act of creating model weights? Absolutely wild, and the implications for RSI are noteworthy.
Carbon scoring below Opus 5.5 on coding highlights how magnificent Opus 5.5 is as a coding agent. With just a $20/month sub, it performs a ton of work at a very high quality.
For those of us stuck in an office building, at a desk, AI models have enabled handling multiple tasks as a same time as fast as we can think. I'm absolutely doing multiple times the work I was doing before and the time between tasks is basically zero. Same pay, though, and drastically increased productivity and sense of effort.
Right, I'd rather lose 15% of cost to waste/inefficiency than increase cost by 25% to put money in someone's pocket.
Alternate headline: "Profit-seeking organizations seek profit"
I've always been frustrated by the intense focus the modern economy has on advertising. If I am interested in a product, I would much rather discover it by searching. Having advertisements forced upon me at every turn is a massive waste of resources (time and attention) given only a fraction of a percent of the products have any relevance to my life. I block every form of advertising I can reasonably block. If a company raises the friction sufficiently, I'll find a way around it. I'm mostly an amoral participant in the economy. I pay for YouTube, Disney, etc. but if they make giving them money an unpleasant experience, I have no qualms employing a usenet subscription and *arr stack to dodge the whole thing.
> The crash occurred around 2:30 a.m. on Interstate 80 westbound at the Interstate 780 interchange, where the Tesla’s dash camera showed the 71-year-old Vallejo man entering the freeway from the right shoulder while the car was traveling at about 70 miles per hour, according to the California Highway Patrol.
> The Tesla braked and veered to the left as the man walked into the car’s path but was unable to avoid crashing the right front portion of the car into the pedestrian, CHP officials said.
FSD attempted to avoid the pedestrian who, for some reason, walked into 70MPH freeway traffic. FSD has fast reaction times to this sort of thing so there's a good chance the guy either came out from behind something or made a very unexpected motion into traffic. A dash cam video is necessary to draw any conclusions about fault or the reasonableness of FSD's response.
I appreciate how Anthropic keeps the Pro tier priced around OpenAI's Plus tier. The extra tier with OpenAI invites additional product subscription gates. Functionally, Anthropic will have a frontier model driving a personal agent available a 1/5 the cost of OpenAI. Astra may be a bigger model, but is there a meaningful qualitative difference between a personal agent running Opus 5.5 vs GPT6 Astra?
They're all equally contributing to mathematics?
This thread contains an enormous amount of negativity that just doesn't make sense here. OAI are giving developers and security researchers access to AI models so they can check their software for vulnerabilities. What's the problem? Sure, there's marketing around the success stories.
Some people are acting like this is only a problem because labs are releasing models that can find and exploit these vulnerabilities in the first place. This is a nonsense claim. Humans have been discovering vulnerabilities and privately weaponizing exploits just fine for decades. We are collectively raising the floor for security.
The difference is the open and self-hosted models are not far behind the closed models. You can't rugpull if you don't own the rug.
Self-hosted and/or open models are fast approaching (or have reached) the quality bar necessary to handle a great deal of software development work. The models needed to cause the OP's problem are becoming commoditized.
Using AI as a tool to investigate the mysteries of the universe is a huge win for humanity. Generating slop images to confuse people online is not. AI is tool, we like when tools are used for good.
This is not representative of what's actually going on, though. It's much more like an artisan carpenter who wants to build houses using bespoke hand-tools and precise joinery. That's great, and there are beautiful works of art created by those skills. But in the last year robots were developed that build homes exactly to spec. The robots work 24/7 and finish entire houses in a few days. That same skilled carpenter is being hired to monitor the robots and check over their work. There is, economically speaking, no place in the process for him to apply his craft. In the time he rips a single board, the robots build the wall. In the time he drives a handful of nails, the drywall is up and the paint is drying. He costs $200/hr, the robots cost $1/hr.
The carpenter is free to exercise his craft as he see's fit, but it's going to take place mostly in his personal workshop, for personal reasons, not to develop homes or run a business in the way he did before.
We can talk for ages about the quality of the robot's work, the maintainability of the resulting home, etc. But it's clear that, at this time, the market and ecosystem are embracing the robots whole-heartedly so those of us used to the old ways are having to rapidly adjust.
My dude, reading AI writing on a product page is not in the universe of being sexually assaulted.
I dislike one-shot LLM prose as much as the next person, but there's no chance this use case ever goes away. Writing is difficult and takes a long time. Using an LLM to write for you gets your message across with the least effort. It's here to stay. Our only hope is that the models simply start writing better.
It's not just about who they hire, it's about who is brought into the country, which is much more under the purview of the federal government than traditional hiring decisions. They can hire anybody they want, but they'll have to keep them remote.
This was my first thought so I scrolled to find this comment. It's a fantastic episode. It's a shame Reply All fell apart. However, PJ Vogt went on to make a new podcast called Search Engine and it's also great.
Colin Furze has entered the chat.
You can't fathom how 'top human engineers' could take advantage of exclusive access to a frontier model at ultrafast speeds and limitless token budgets to conduct research?
I've found basically all of the release presentations are completely irrelevant and seem to target a very narrow subset of the population. They're always reserving something like a restaurant or a sports venue or generating a new slide deck for their business meeting. Perhaps these companies would do better if they could make a use case for regular people. You know, the 90% of the folks out there working normal jobs and not spending infinite money. I don't think I know a single person who has made a restaurant reservation more than once or twice (mostly never) in their entire life. Maybe I'm the weirdo here, but I can't relate to any of these demonstrations.
This has 'social media will bring us all together' vibes.
As for prosecution, none of the victims in this case have any interest in pressing charges.
At this point, it's kind of a hackernews thing. Simon posts them as a single comment in the relevant thread. It's okay for this place to have a little bit of a sense of community, and you can just ignore the comment.
Is that under-appreciated? It's kind of the whole deal aside from environmental concerns.
These things remind me of my dad's 90's ford ranger. I would love a small functional pickup. I don't need to drive a battle-tank around town. If all works out for these trucks, it's a top contender for my money in the near future.
If it makes you feel any better, nobody is playing with the toys, now.
I'm very exciting for Nvidia to meaningfully enter this space. I know they've been working on autonomous vehicles for a while now, but it seems like they are approaching a real product. Hopefully, they produce something that can be used on consumer vehicles. We really need good competition in this space. The US market is limited to Tesla FSD and no other manufacturer is even close. I'm not confident individual manufacturers could meaningfully develop their own solutions. A strong third-party option is a great direction for the industry.
> Because if a "condition" doesn't impact you, will it help to be aware of it?
Fast and cheap full body scans could provide the data necessary to tune out the noise.