it doesn't matter the reason. This is a race and nobody will care or remember how the winners got there.
Mistral looks like it's fading away to irrelevance unless they can play alongside the similar sized models, or have some unique advantage other than being in Europe, for Europe. I was really excited for them back when they were startup that had the biggest European venture round ever. This space will have a few winners, and many losers. Google, plus either Anthropic or OpenAI most likely. Big models will see breakthroughs in inference performance/cost fall precipitously and small models will only exist on devices (Pixels and iPhones, cars, watches, bluetooth speakers, etc)
It’s not that I don’t agree with you, I am just pointing out why it’s hard to catch up to scaling laws given the European economic (capital) and political (US would be upset if they found out Europeans distill) constraints. China is only bound by economic constraints.
> This is a race and nobody will care or remember how the winners got there.
For consumer AI, yes. For coding assistants, probably.
For specific application "business" AI like the things Airbus announced the other day? Not at all. What matters for an Airbus using Mistral to build compliance documentation based on AI generated physics simulations is the enterprise relationship, reliability, compliance, forward deployed engineers helping with the fine tuning, quality, predictability, support. A Chinese lab having a better at benchmarks model that is cheaper is just irrelevant for that.
And IMO, the real money in AI is this type of "business AI" deployment. Developer tooling tends to converge on becoming commoditised. Once you're a core supplier for a big bank and embedded in their processes, you're there untill you screw up with the pricing (like Broadcom), and even then.
The world seems to be fragmenting, into those that see the value in the latest from Google, and those that resist changes like this. I search for how much oil does my <ICE vehicle> take" and get the exact answer in a single sentence, or I suppose I could click the links and wade through all the validation for choosing <ICE vehicle> and how often one should change the oil, and which brand of oil that blog is pushing etc etc.
I love Google's AI answers and their AI Mode tab. DDG is just Bing or a search vendor proxy, so I've never understood the fascination. At least Perplexity is different to Google. DDG seeing a 28% increase is like Google saying they saw a drop of 0.0000000001% in traffic.
HN crowd forget that the world isn't like us, they didn't grow up with Yahoo and Alta Vista, with Excite etc etc. Our SOP is to resist all change, anytime Apple brings out a new version it'll be the end of Apple according to HN - Apple - the biggest company in the world - what do they know about UI, "Liquid Glass sucks!" :) :)
We're a community in danger of pushing out those new to the tech world, recent graduates will be made to feel unwelcome if we continue to trash everything that the biggest companies in the world do, like we always know better. I implore the community to be more positive about the future, about the technologies that will take us into that future.
> I search for how much oil does my <ICE vehicle> take" and get the exact answer in a single sentence, or I suppose I could click the links and wade through all the validation for choosing <ICE vehicle> and how often one should change the oil, and which brand of oil that blog is pushing etc etc.
Don't you have to do that _anyway_? Unless you're just blindly trusting the AI to be correct, and if that's the case, please do enjoy gluing your cheese to your pizza.
Yeah, but perhaps we're getting to the point where the AI synthesis of all reachable data sources is going to be more truthy than trusting whatever random anonymous result you pick from the top half of the results page.
I'm also finding that the AI-cited links are often more helpful and authoritative than the top search results.
I like having a direct answer to my question "how much oil does my engine take" but as of today I do not trust the answer to be correct, so I still cross check several sources, ideally ones that appear to be authoritative.
I asked claude to dig up the current Ford Bulletin for the engine in my truck to tell me the recommended motor oil. And it found the updated recommendations properly. I wouldn't trust google AI because I know specifically that the recommendations changed, and I don't want whatever the published specs were when the engine was first manufactured, which is out of date (and found on lots of low quality blogs). I don't even trust claude, but it gave me the URL to click on to verify and summarized it well enough that I mostly trusted that it wasn't using the cited technical bulletin and not a bunch of random AI-slop web pages.
I wouldn't trust any 'confident stochastic next-word predictor' to tell me fact. There are official sources of information for these kinds of car maintenance questions.
Better than asking your backyard mechanic buddy who believes that you don't need anything other than 10W40 or something like that. You have always had to know who to trust and what level of work you need to do in order to get to an answer which is satisfactory. And in this case, Claude cited the manufacturer's bulletin which is the actual best source of truth that you're talking about.
Right, but standard Google search doesn't find them reliably either. Instead you get answers from enthusiast forums, "quick oil change" franchise sites, SEO slop, and maybe somehing from the actual manufacturer site if you scroll down.
Being critical about AI companies isn't what's pushing new people away from the tech world. The AI companies and the consequences of their actions are, as well as comments like this pretending the issues don't exist and that we need to just be positive about the "future".
And supposing these technologies do take us into the future: when said future is bleak and worse in most ways than what came before, people aren't going to be encouraged or enthusiastic about the tech world.
> search for how much oil does my <ICE vehicle> take" and get the exact answer in a single sentence
How do you know the answer is exact?
> or I suppose I could click the links and wade through all the validation for choosing <ICE vehicle> and how often one should change the oil, and which brand of oil that blog is pushing etc etc.
Where do you think that "exact answer" is being scraped and averaged-out from?
Are you going to enjoy a future where those different sources can't be found, so now Google requires you to have a subscription that includes data about vehicle repair? The great thing about the web before was that the information was available for everyone, it was decentralized. This is what they are trying to kill.
> The world seems to be fragmenting, into those that see the value in the latest from Google, and those that resist changes like this.
This reads as a strongly closed minded claim that has been "whitewashed into corporate appropriate speech" by an llm. If you cannot understand any validity to the other side of a debate then you are not engaging in discussion, no matter how your claims are dressed up.
You're entitled to your opinion. But "we should embrace the stuff big tech is doing" does not follow from "let's be welcoming to new entrants to the field". They are, of course, welcome to their opinions as well. But even if they and I disagree on things, that doesn't make them unwelcome. So no, I'm not going to embrace the slop Google is putting out based on a spurious concern over welcoming newcomers.
I've never seen such a reaction to Apple on HN. Quite to the contrary, there are a lot of fanboys here. A lot of macOS-only software is promoted here and I rarely see any complaints in the comments.
The "in" operator is not unusual or trivia, it's something people who code in python use all the time! Python does have a function (or more accurately a method), it's called __contains__ and it's invoked by "in". I would expect a python developer to be able to comment on whether python uses a fastsearch algorithm or not, they should be able to comment on BM and H algorithms.
Maybe this was a very junior position, but I'm with the interviewer here. Using regex would be very questionable - and a solid case of 1171.
reply