Spotlight search relevancy is a complete joke. If only they did some embedding based search across the system and paid attention to basic precision recall numbers. This has gone from bad to worse quickly.
I think a more comprehensive and simpler explanation is that the people protesting just hate this administration. They don't go about making lists like this and then think they need to go protest. They just see a guy they despise and start protesting. Hate is really powerful.
I feel that the mob doesn't understand nuance and right now that mob is fighting for control for definitions of words and what is moral and ethical without giving you the freedom to choose for yourself and accepting it without malice. It's vicious and tiring and definitly not productive.
A while back there was a lot of hype on ultra wide band that is some sort of pulse code modulation and it being trails t to jamming. Can military jammers jam these too? Or is this similar to spread spectrum and the same jammers work for UWB as well?
Size should be a product metric so that it is tracked and optimized. Same goes for memory consumption. It's easy for product to come up with features and run the enxt cool experiment but your users don't care about any of that.
The Netflix series Turning Point talks about fall of the Berlin wall and the relevance of NATO after the Warsaw pact ended. Putin wanted NATO to end but his actions only made it stronger. It looks like this will collapse from the inside without the Russian involvment.
> actions completely legitimate
So if I'm a dictoator in Africa who hates US and the west, I'd first think oh that evil empire already did that so I can also do it too? Why would they even care what US does? This makes no sense.
I’m curious which specific African countries have the capability to pull off what US did with this level of precision and literally no casualties. Moreover these countries already hate US, so shouldn’t they logically hate the US actions as well?
Do these Chinese cars last over 5 years? I remember that it was a long hard path for Kia and other Korean car companies to catch up to the Japanese car quality.
This is just plain corruption and the government officials were paid off. The parents of students should have been given the option to request the devices if needed and that would have been 1000x cheaper.
The ffmpeg wrapper use case is interesting. Reusing some core software for speed and reliability but to get better usability by leveraging LLMs seems like the sweet spot for SaaS use cases. A lot of these CRM products are basically wrappers around DBs and sheets like core products with terrible UI. The new breed will solve the UI ease of use issues.
Hard disagree. Tim should focus on fixing their software. It has become extremely buggy and it needs to be fixed. No one buying an iPad cares about running some custom browser and supporting it is pointless and is what makes the software emote complex and worse. He should take better care of his paying customers rather than engaging with opinionated activists.
I'm curious why use a hosted browser instead of just spinning one up locally and since you already have he electron app. Why not just use a different Chrome profile for isolation and interact with that?
NPR mentioned something about Europe not mandating this and some other vaccines the US mandates but the expert basically dismissed that saying 30 years ago US decided to do its own thing. I felt this was biased and was just taking an anti-incumbent / anti-Trump stance. Does anyone have a more scientific view of EU policy with risk assessments?
> I’m predicting that the uncertainty and discouragement of H1-Bs will lead to a destruction of jobs in the U.S. and a suppression of salaries as high quality software engineers don’t move to the U.S., allowing the center of gravity of the industry to shift out of the U.S.
Statistically speaking if you take a bunch of low-mid level salary / skill folks and move them out of US, the average salary should increase no? R&D jobs in countries like India is still low compared to US. These are the high skill and pay jobs and they will continue to be here.
Is there data on what site search engine they use? This is hard to get as it is sitting deep in the backend but will be super useful information. built with doesn’t have this, but they do have a list mapping search engine (Bloomreach, Coveo, Algolia) to the website probably based on private data dumps. Being able to look this up for a website will be very useful.
One thing the analysis for textiles vs cars misses I the complexity of the supply chain and the raw materials / components that need to be procured to make the end product. Steel/textiles have simple supply chains and they went through a boom/bust cycle as the demand plateaued. But cars on the other hand will not go through the same pattern - there are too many logistical things that need to line up and the trust factor in each of those steps as well as the end product is quite high.
Software is similar to cars - the individual components that need to be properly procured and put together is very complex and trust will be important - will you trust that you as a restaurant owner vibe coded your payment stack properly or will you just drop in the 3 lines to integrate with Stripe? I think most of the non-tech business owners will do the latter.
ChatGPT already knows about these Khan Academy videos so there is not much to get if there is another app out of an app. However, it could be use cases where it can pull out real time information like e-commerce stock or sales deals that users might find very useful as this won’t be available within ChatGPT’s old training data.
The article is not making a proper distinction of scale and is probably due to the small scale problem that they solved. What is small scale and <10K documents / files can be easily processed with grep, find etc. For something at larger scale >1M documents etc. you will need to use search engine technology. You can definitely do the same agent approach for the large scale problem - we essentially need search, look at the results and issue follow up queries to get documents of interest.
All that said, for the types of problem the OP is solving, it might just be better to create a project in Claude/ChatGPT and throw in the files there and get done with it. That approach has been working for over 2 years now and is nothing new.
Ads haven't made it yet, they are charging money for the purchase made:
> Merchants pay a small fee on completed purchases, but the service is free for users, doesn’t affect their prices, and doesn’t influence ChatGPT’s product results.
If each LLM sessions is linked to the domain and restricted just like how we restrict cross domain communication, this problem can be solved? We can have a completely isolated LLM context per each domain.
Does anyone have insights into what is at the backend of all this? I know there is Playwright, Broser Use, StageHand as some of the technologies people use. If everyone of these is using one of these, what exactly is the differentiator?
Meta Got Caught Gaming AI Benchmarks
Meta released two new Llama 4 models over the weekend -- Scout and Maverick -- with claims that Maverick outperforms GPT-4o and Gemini 2.0 Flash on benchmarks. Maverick quickly secured the number-two spot on LMArena, behind only Gemini 2.5 Pro.
Researchers have since discovered that Meta used an "experimental chat version" of Maverick for LMArena testing that was "optimized for conversationality" rather than the publicly available version.
In response, LMArena said "Meta's interpretation of our policy did not match what we expect from model providers" and announced policy updates to prevent similar issues.