zhobbs

Could it be TPU/GPU availability? ie, is it possible that it's more profitable for their compute to be serving Anthropic models, so Gemini customers are stuck with lower compute available.

It seems like future models could build off the 10 steps to take 10 more. It's not 10 steps of AI slop, it's a discovery to build off of just like how human scientific progress has worked.

Awesome project. When I read the display I hear a very specific voice in my head calling out the arrival times!

https://www.bart.gov/news/articles/2009/news20090309

And if no one pays for any of that content there will be zero ways to stay informed!

conversely, what's the purpose of using Codeberg over Github?

I'd like to hold it, it seems like it might be more one-hand grippable in this form factor.

I've never transferred data over a wire to or from my iPhone, interesting that this is important for some use cases.

The article quotes the CEO saying yield is comparable:

>TSMC’s new Arizona plant is already comparable with those in Taiwan when it comes to the measure of yield — the amount of good chips a production run produces per batch — Su told the audience at the forum.

I think it also shows that fabs who only have one customer (ie, Intel) aren't as competitive because they can't provide as much scale and are more sensitive to that customer's success.

Intel's fab would be doing much better if it spun it out a while ago and was making Intel, Nvidia, and Apple chips right now.

Possible I'm misunderstanding what you're trying to do, but ollama works well for me for local inference with qwen on my Macbook Pro (32GB).

Yeah, I think we're over analyzing it in this thread. Seems pretty light hearted and fun.

I think another generalization people make here is around complexity. Many developers work on apps that just aren't that complex. Glorified CMS's mostly doing CRUD with well established code patterns.

Sure, LLMs might create slop on novel problems, but a non-tech company that needs to "create a new CRUD route" and an accompanying form, LLMs are smart enough.

Yeah this will likely lead to margin compression. The best companies will be fine though, as brand and existing distribution is a huge moat.

Totally insane, $6.5B acquihire. But also, all stock deal and OpenAI appears to have an unlimited number of investors waiting to give them money if/when needed.

My understanding is that modern search engines are using embeddings / vector search under the hood.

So even if LLM's aren't directly passing a vector to the search engine, my assumption is that the search engine is converting to a vector and searching.

"You interact with embeddings every time you complete a Google Search" from https://cloud.google.com/vertex-ai/generative-ai/docs/embedd...

Just wonder if it matters? If Google spent 10x as much in the first 5 years of its life would it be a worse company now? Giant TAM, winner takes all (or most?), all that matters is winning.

>For what? There is no ROI at that price point. There is no monetisation potential.

I think your whole argument is based on this being true, but you didn't give much argument about why there is no ROI. 400M USD isn't hard to generate...even a moderate ad engagement lift on X would generate ROI and that's just 1 customer.

Imagine going back in time and showing every VC how great the search business will be in 20-30 years. The only rational response would be to make giant bets on 20 different Googles...and I think that's what's happening. These all seem like rational investments to me.

Curious if anyone knows, what legality of using a technique like this? I assume illegal, even though it's just making publicly available API calls?

Because it’s fun to create things like this in very little time. Some people wake up and play wordle in the morning, I like to spend 15 minutes making something fun and dumb with v0 or lovable.

is the prompt available somewhere?

There are advertisers who specifically want to reach fans (gamers, tv/movie fans, etc), the name change was to make that ad sale conversation stronger: https://about.fandom.com/mediakit

It's usually due to incentives and the time horizon you're optimizing for. If a manager is tasked with maximizing revenue over the next 12 months no matter what, then increasing the ad load is a lever you are probably going to pull.

If your goal is to create an enduring product that will slowly grow revenue and be around forever, then you're probably not backed by VCs or private equity, or you have a cash machine (google search, etc).

The reality is that some businesses shouldn't take VC money and shouldn't get so big. Maybe a wiki farm should just be a wiki farm profitably run by 5 friends or something.

It’s a good point, but one thing to consider is that ultra long hard tech might be structurally challenging as well. If fusion requires 100 years of capital and R&D it won’t be viable ever due to economic/societal/structural reasons.

Seems pretty well known at this point that risk-adjusted comp is higher at bigger stable co's. Personally find them boring, so I tend to work at startup/growth companies, where I can make enough for the lifestyle I like, and I enjoy my work.

The latency is too high for me using the wireless Mac display on Vision Pro. Would be a lot better if we could use that wire...maybe it's possible?

I think it's just a function of the Apple marketing overselling the passthrough quality and productivity use cases.

The manufacturing process isn’t the issue, it’s the layers of contractors, subcontractors, and lack of accountability.

It's a pretty recent version of Android, it probably supports Android Auto. So if you have a newer car or an aftermarket Android Auto device it might be best of both worlds.

Recording real world data can be cheap for some use cases, but often labeling it is very expensive.

Can you elaborate? As far as I can tell, effectively everyone is like this...