barrell
If they don’t end up switching models by the end of the year, I think that says a lot about the models. If they’re just waiting for an appropriately large or small model, or waiting to get large 4 put through its paces before switching, I think that’s a pretty ego-less way to go about it, and could speak pretty highly in a way.
Mistral large 3/medium 3.5/small 4 have all been pretty dated for a while so if they just needed a stopgap, I could understand. Now that large 4 is out though, we will see if that was the case.
I prefer using a European provider though so this is actually cheaper than I can find elsewhere.
Turns out they’re using GLM in le chat. Quite concerning if I’m being honest, and this is coming from a huge mistral fanboy.
I wish we could toggle mistral large 4 in le chat so I could get a feel for it.
Too many models now to remember specifics - I believe deepseek v4 did this and one of the gpt models (5? 5.1? 6?) did this if I recall correctly. I’m also fairly confident Google has had a few promotional periods on launch of the model, I seem to remember them starting the trend.
This is just my headcanon though.
Mistral models were the best for my use case when they came out, but they’re mostly almost a year old now. Hard to keep justifying using them, especially with all the price cuts this year on other models.
Many comments, but I'm surprised no one has mentioned the ease and quickness LLMs have to advise you to leave relationships. Doesn't matter the context, how little you tell them, or how close the person is, an LLM is always ready to drop them at the tip of a hat.
It's weird enough chatbots try to be your friend. Doubly weird that the search engine tries to be your friend. The downright dystopian part is how often they tell you to leave your real friends.
I’ve spent the last three years tweaking away at this algorithm. There are so many little details that have had to come together to make it an enjoyable experience. For example, take the one problem of balancing multiple language: What is the best way to balance multiple languages? How should you switch between them, and how often? What happens if you are over performing in one language and underperforming in another? What happens if you have more reviews in one language than another that isn’t in line with your priorities? How do you deal with competing learning speeds and language priorities? Spiky review burdens? What happens when you change your language priorities?
This is probably the most trivial part of the algorithm, and it took years of trial and error and dozens of iterations to get it to feel right. It requires knowing about SRS, the FSRS algorithm, language learning principles, and even then tons of experimentation, trial and error, and hundreds of days of daily usage.
I’ve asked LLMs every step of the way what to do. I’ve probably taken <1% of the advice I was given. Just asking an LLM “build an SRS app that balances multiple languages” would produce a result, but one that would definitely be terrible to use, and probably would end up slowing you down, not speeding you up. Explaining to it any of the issues that you would come across when using it will likely suggest to changes that will only make things worse, based on the thousands of terrible ideas I’ve seen confidently produced.
The hard part of a product is not the idea, but the experience of using it. The valueable work is normally not in the big feature specs, but the hundreds of little paper cuts that have all been smoothed over. The time saved is not because you don’t have to code it, but is from not having to become an expert in several different fields.
I would not want that in my product.
For example from the original analogy, if you told an entrepreneur that because they didn’t use AI, they were not an entrepreneur, but instead “unskilled” or a “beginner”, that would be disparaging; pejorative.
I do chat with the web interfaces of LLMs quite a bit as my workflow. I’m always baffled by how rarely it arrives at the correct solution.
I pose a question with code snippets, it gives me some incorrect answer. I pose a better question, and it gives me another wrong answer. Wash rinse repeat several times; not until the LLM gets the right answer, but until I pose the question specific enough that I can answer it myself.
I do find this loop of question refinement quite beneficial, but man it makes me terrified of vibe coded applications. 9/10 times the LLM doesn’t even get in the ballpark of the problem, let alone the solution.
You would not want a surgeon who says that they’ve tinkered around with surgery. You would not want a pilot who says they’re tinkered around with piloting.
Synonyms are quite literally beginner; dilettante; novice; pretender; uninitiate.
More in line with your definition would be artisan or craftsman; although fwiw craftsman can have a negative slant in modern (American) English as some people use it in lieu of blue-collar workers.
The same comparison could be said the other way around: there are entrepreneurs, who care deeply about the product, inside and out; and scammers, who don’t care about the problem itself or the beauty of a a neat solution, but only the sales of the end product.
Entrepreneurs currently are in hell, and a scammers are in heaven.
I don’t think either is more apt or more offensive, they just appeal to whichever side of the fence your on.
I’m sure LLMs have only made it worse, possibly pushing it over an edge where hand coding as a freelancer is not viable. I do not know.
Personally, I tell myself that we’re in the lowest of the low at the moment. Either AI will get substantially better, or it will get substantially worse. The amount of debt the industry is loaded up with is not sustainable. Something is going to have to crack. I expect in another 2 years for things to “settle down” on some side of the fence, although which side I do not know.
My quipy advice though is to build something for yourself. Be the boss that doesn’t require AI. We need more builders behind products, more indie products built because they’re cool and the founder is passionate about the problem; not more business types and cash grabs.
That’s all easier said than done though I know.
Yes coorporate “ship as fast as possible” is draining, but that doesn’t mean LLC coding isn’t also to blame.
I have little to no interest in AI coding - I try it every couple months but always regret it, git reset, and go back to coding by hand.
It feels quite lonely these days. The general energy and enthusiasm from hn was a big source of motivation, or maybe community. Now using hn feels quite isolating.
I do see the occasional post about going back to hand coding though, and their frequency seems to be on the rise, so hopefully that correction continues
(I don’t need to see no AI, to each their own, just would be nice to see an equal amount of hand coding content one day)
His predictions though? I don’t think I’ve ever read anything from him to get his read on how things will play out.
Just because one makes the wrong predictions does not mean that the data used to make them was wrong. A lot of people seem to dismiss his reporting because of his takeaways
I think most true native apps don't have these issues. Unfortunately there is very little incentive to build on the macOS platform these days, which is where I point more of the finger at Apple. UX, UI, & DX are all quite a bit worse for Mac native.
However, I live with Activity Monitor open at all times. The computer constantly grinds to a halt, and I have to force quit apps at least 10 times per day.
Not really Apple's fault though, just most apps use around 4GB of RAM these days even when they're idle. Some use upwards of 16GB.
I've watched the computer slowly get obsoleted over time, not because the hardware is too old, but because software just get's worse every year.