These companies have pivoted from being cash generation machines to being data center building companies. It’s a huge bet that might pay off but the market is starting to notice that where there used to be revenue generation there is now infrastructure spend.
That representation is also old, incredibly well documented, and used to describe how to reason about chess. There are of course text guides to other games in training data but they rely upon depictions of what’s happening that aren’t purely text so the game harness is always going to have to make novel decisions about represent the game as text.
Yea it’s wild watching so many smart people convince themselves that LLMs are general purpose AIs. Don’t get me wrong they are incredibly powerful tools. However being surprised that text models cannot play video games particularly well is like being surprised weather models cannot.
Why aren’t those all spiritual questions? They seem like it to me. Maybe not religious questions but at the very least they are questions which, if approached honestly, force you to grapple with what it means to be you.
Describing the issue as “violent” is wild. Reading through a bit, it’s massive, it’s clear no one involved has the moral high ground here. The polite response is to close the issue if you believe it’s genuinely off topic.
Still not quite sure what you mean by obvious because to me “Stop. You know nothing. You have shipped 0 features by hand. No one has ever depended on your code.” Is much more violent than “please do not vibe fuckup this software”.
There’s a difference between being a hater and acknowledging the reality of the technology and those building it. I want all of those things too. However I do not understand why LLMs will get them for us. Instead I see a few really powerful people looking to get more powerful. I see a powerful tool being presented as god in a box when. I see the most resources ever spent on a singular thing being spent in a way that’ll _best case_ be mostly obsolete in a few years.
I want real AI. I want cures for cancer. I want too want to live in a post scarcity world. We had most of the technologies to do that before this. However the companies and investors involved in the AI build out chose to sit on massive reserves instead of trying to directly solve those problems. There exist proposals which solve hunger, the energy transition, etc and together they wouldn’t amount for even half of what’s been spent.
That tells me those involved want nothing other than money and power.
I love that ember is still going. It’s obviously not the number one choice these days, might never have been, but it represents approaches to frontend web dev that are very different from the standard so it’s always great to see the diversity of ideas out there.
There’s an old Malcom Gladwell podcast episode, I think the show was Revisionist History, where he says he’s an interview nihilist. As long as the person seems reasonably capable, and can probably do a bit of what you need, hire them. Interviews are so hard to get right that what you’re saying ends up being most effective.
That sounds like a corollary of the Diamond Paradox. If there's any cost at all to price-shopping, then once you're at one store that sells the thing you want, it's usually not worth continuing to shop around.
Tangent: people do a lot of stressing out lately about which LLM they should be using for a certain task. I bet this advice applies there, too.
Unfortunately from an organizational perspective, a bad hire could cause so much damage through incompetence let alone malice, that making no hire the default unless they're a perfect cinnamon roll of a fit, is actually a good strategy.
> a bad hire could cause so much damage through incompetence let alone malice
The fact that an organization cannot deal with such a case is a bigger problem in the first place. Eliminating incompetence and malice is among the basic skills of an organization.
What if a bad apple is hired despite all the checks? The system should be able to detect and eliminate bad apples before they give “so much damage” regardless of when they are hired.
Of course the organization should do its best to avoid bad hires. It should do so because of the opportunity cost of not hiring the right person, not because of the damage that they might give to the organization.
I was actually hired at my current job to replace a worker who was let go because pretty much from day one, she was acting suspiciously, consistently with someone who was outsourcing their work to contractors and splitting their paycheck with them. Posing a significant risk to the confidentiality of corporate IP and data, from day one, despite putting on a convincing game face throughout the interview and selection process.
In light of this kind of threat, strongly favoring no hire, and even policies like mandatory on-prem work, start making a whole lot of sense.
The case supports my argument though. Trying not to hire a bad candidate is not a resilient strategy. Someone might fall through the cracks. You still have to screen the candidate after the hire. Heck, even not hiring is not a good strategy, because someone inside may turn bad.
Coding agent harnesses strike me as similar to blog generators. They can be as simple or as complex as you’d like. Plugins help with adoption. And if you want it’s real easy to write your own that does exactly what you want.
In a reductionist view yeah but blog generators and agent harnesses sit at a different spectrum than an EHR/Excel/whatever other insanely complex edge case ridden work you can think of
reply