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.
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.
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.
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.
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.
HN has a huge user base. If the article is pro-AI, you'll see top comments praising AI. If the article is anti-AI, you'll see top comments deriding AI. The article and headline is just an attractor to those whose opinion is aligned. It's not unique to AI, basically every thread has your typical nerd-fights, but AI has been particularly polarizing for many reasons.
Not only is this NOT the case, but all Tesla vehicles since something like 2021 have included matrix lights. They have adaptive beams to automatically darken sections of the headlight beam to avoid blinding other drivers.
What's 'allowed' in a speed run is decided by the community who do the speed running. Further, run leaderboards are broken into categories based around defined parameters. So, perhaps there would be a category unique to hardware manipulation such as this. For major speed running games, the rules are often very well defined and enforced.
On a Macbook, natural scrolling feels right on the touchpad. The real crime is not having a separate setting for a mouse. I had to use applescript tied to a keybind so I can use natural scrolling on the touchpad and toggle to regular scrolling when using a mouse wheel.
I wonder how a focus on per-token API profits will impact the incentives to improve token efficiency and drive down costs through optimized compute. I suppose as long as a few leading labs are competing, we'll see progress in this regard, but it's certainly less in their interest than it is with a flat subscription pricing model.
> Without guidance, LLMs tend to paint themselves into a corner, because they’re generating code to solve individual prompts, not thinking holistically about an application’s architecture.
I've found I can prevent the LLM, in many cases, from thrashing on a bug/feature for long periods of time by switching into plan mode and, even in the middle of a conversation, having it reassess the structure around the problem, first. If you keep prompting about the same bug, it may keep producing variations of the problem code. But forcing it to stop and 'think' for a bit, has yielded much better results.
I've been finding a lot of ~5 year old satellite and street view data. It's only anecdotal, but it seems like Google is not updating their imagery as often as they used to.
I must believe, as with the medical profession's uptake of AI tools, if these tools prove themselves to be reliable and meaningfully helpful, she will experience more than enough professional influence to learn the tools.
Measuring token usage as a productivity metric is like measuring keystrokes. Don't mind me, just over here rolling my face on the keyboard for an hour so I can take Friday off...
...except each keystroke has an associated cost, the sum of which may equal or exceed my salary.
+1 for NAND-to-Tetris. I combined it with a visual logic simulator so I could actually see the structures beyond the VHDL. I would love to go back and do Part 2.
I'm at 96% FSD miles since they started tracking a few months ago. It works very well. It was unreliable until about 12-18 months ago, but it's been great since v13.
GM/Ford/etc. driver assistance systems are nowhere close to the latest FSD. It would be nice if there were other options, but there isn't. You can't buy a Waymo. FSD can go from parked in the driveway to parked in a parking spot at your destination without any human intervention.