636 karma · joined March 31, 2026
Man, it has been purely made of fake content, even since before LLM were a thing.
And the notifications are so senseless that everyone disable them -- and people don't even answer the private messages.
There is nothing social about LI.
- A workflow + distribution channel of open models, which sells GPU
- A world-class AI infrastructure know-how, which is actually hard at this scale.
I can imagine NVIDIA contemplating the opportunity to move up the value chain and sell inferences from open models at scale.You might think it would be a commodity, but they would have a strong cost advantage since they make the chips.
Second, you can't compare technical (planned) maintenance with the uncontrolled intermittence of wind or solar. Not having sun or wind during peak demand is the actual problem. In Europe, peak demand is 6-8pm in winter: no sun, no wind.
The last problematic preventive maintenance, mandated by the supervising agency, which forced France to import MWh for a few months was in 2022. Anything human-made can have a downtime. It was actually a voluntary (although mandated) downtime in this case.
Moreover, you may not know that nuclear is forced to reduce output when intermittent energies increase their output. There is no good technical reason for this, it is just how the regulation has been made (you can imagine lobbying entities asking for it). You may have noticed more sun in June than in January, so yes nuclear may have had to reduce output. Not due to the technology.
If you avoid irrelevant comparisons of costs (e.g. vs intermittent sources), nuclear is actually attractively priced.
Actually, nuclear is safer than, say, hydraulic power.
With that said, it requires a country with good governance and an independent safety regulator (like in France).
Nuclear works 24x7 on its own. Intermittent energies don't.
Intermittent energies need a whole infrastructure on their side (usually gas / coal power plants) to guarantee 24x7 electricity. This cost should be factored in.
If you are not convinced, think about your cell phone service: would you pay the same for a network that works only when it is sunny ? No. Not the same.
NB: don't even mention batteries, when the largest scale projects, very expensive, can only supply power for a few minutes at grid scale.
But it does not make my hand bigger to hold my phone, and it adds a gesture i.e. friction to the UX.
Alas it seems we won't see any such phone that fits in a pant's pocket and that can be operated single-handedly. I am afraid that the 13 mini is the last of its kind.
Yet many reports of people dropping their gigantic phones. Maybe that's a business model after all.
Good luck :)
- with AI, bad engineers (bad designs?) lead faster to catastrophe - Frontier AI companies pay their SW engineers very much
--> we need highly-skilled sw engineers and the lesser ones will get out.
Kind of what outsourcing did but x100.
As a manager, I'll still have to think how I grow young SW engineers to top SW managers.
That was geared against telemarketers or abusers, so that the callee could fetch a police officer that would acknowledge the fraudulent call.
Today all this data exists in databases. It would be so easy to check call logs from telemarketers to check if they ever called someone on a do-not-call list.
Or maybe that's exactly the right example...
To be honest, scraping them on broadcast TV will probably have little influence, even though in the wrong direction, because TV is no longer the media with the most influence.
The real failure is that no one cared to regulate ownership of digital medias, i.e. social networks, Youtube, etc. These have shown documented examples of influencing elections directly.
I totally understand why the OP would use only FBV, however when writing REST APIs you will want CBV to reuse base classes such as ListView.
The paper is actually poor in data and mostly requotes other studies.
Great, but since no human has looked at them, how do you know that they are relevant and actually test undesired behavior?
Also, would you be able to point at which tests should be added in the future?
Who needs memory when your model is set in silicon ?
But is there a single UI abstraction that succeeded in affordability for newcomers while maintaining performance and solid architecture?
It is high time to solve the write-3-apps for each usecase (iOS, Android, Windows). I use MAUI on some projects but it is still not perfect.
The way forward for Apple is to open source more and more and let the Swift community build a cross-platform solution.
Open up your mind and learn other cultures.
At the end it is more productive to be focused on the right target, than to gain 15% of productivity because you use whatever tools. Note that this is not mutually exclusive, but putting brain power in thinking about the product, rather than about the productivity hacks, is what matters.
The problem is that most companies don't care about thinking. You may have more chances to get promoted because you are sweating and running everywhere, instead of calmly thinking, looking through the window and developing mental models.
Time to 1st token is faster on the M5 because of HW accelerators helping the prompt interpretation (and it is CPU-bound).
Token generation after that is GPU-bound and will profit from the higher bandwidth of the M4 Max.