05032-Mendicant-Bias

Well, nothing stops i.e. the kde devs from integrating it into their DE with a suite of premade scripts and feeding the text straight to the LLM.
 
But yeah, I agree speech controls are mostly a pointless gimmick - except for accessibility, I can see it being useful for movement impaired users.
The "problem" is that LLMs are just not the right tool for creative writing; it's not what they are designed to do and the way they work kind of inherently prevents it. What you get out of an LLM is what it deems to be the most likely next word, which inevitably brings it down to a sort of "average human writing an essay" level of quality, even at its best. "The average book" is probably not something I'd be terribly interested in reading, even if the model was expanded to be able to output hundreds of coherent pages.

Pretty cool, but do note that if you want it to actually do things on your computer it seems you'll need to write your own scripts for it to reference.

Impressive what a road-legal car headlight is in Canada.
 
Here in Europe, this thing would have failed inspection because the lenses/housing aren't clear enough/aged.
Also modifying the headlights like this is a big no. Even if you had a kit for a  specific Honda Civic, you couldn't just use it on a random Civic model. Some of this is pretty silly: Replacing a halogen bulb with an LED that is 'identical' to a halogen bulb still requires the LED manufacturer to test and certify each model separately, meaning that these upgrades are only available for popular car models.
 

who wouldn't want fake money???

Summary
OpenInterpreter just announced a portable, open source, language model-powered voice interface for home computers. Like OpenInterpreter in a terminal on PC, the 01 can be run either via OI's servers, or 100% locally using a downloaded LLM and local host ❤️ (as OpenInterpreter does currently). The 01 developer preview is live on GitHub.
 
Quotes
 
My thoughts
This is one of the more exciting bits of AI news I've seen in a while. I love the open source ethos, and I believe it can offset some of the more worrying potentials of allowing corporate profiteering to be the primary driver of this endeavour. I've already been playing with a local instance of OpenInterpreter in a Windows terminal using local inference, testing it out as a code building tool and general helper. It's slightly fiddly, tech nerdy stuff, which is fine for me, but would never reach mass uptake in its current form. The 01 is super exciting, not only cos it's like having your own personal Majel Barrett Enterprise computer interface, but because it could break through that fiddly tech nerdy barrier and make a genuinely useful open source tool which everyone needs - the natural language computer interface - accessible for huge numbers of home computer users.
 
Sources
https://www.openinterpreter.com/01
https://github.com/OpenInterpreter/01
https://twitter.com/OpenInterpreter/status/1770821439458840846
 

It's a technique called Pulse and Glide and it does work to save fuel. As far as I know, in some countries it's not allowed because the uneven speed makes it harder to drive in a motorway, with people constantly coasting and accelerating.
 

 
Efficiency just depends on the engine and operating conditions of the engine. The chart above tells you that this particular engine has an optimal RPM/Torque at which it gives the best efficiency from chemical potential energy to mechanical energy. The further away, the lower the efficiency.
 
So rather than driving at a constant speed maybe in the yellow region, it's more efficient to operate the engine in a burst in the red region, then coast, then burst again and repeat.
 
This chart also tells you the best efficiency you get 40H to 80HP, and if you want the best acceleration instead you can operate at the peak where the engine is giving you 124HP, which is where you are at high RPM, and the engine is giving you high torque. (Power = torque*rpm*conversion constant)

In WAN show on 12th April at around 3rd hour, Linus and Luke talk about light acceleration being more efficient than stronger acceleration. This is a very common misconception, that I've encountered way to often, when working in this field.
 
Engines work optimally at optimal load, which, for ICE engines, is close to full load. For electric motors, it varies greatly based on design.
 
For optimal efficiency and best fuel mileage, you'd want to accelerate at near full load in highest gear, until reaching the revs where the friction of the piston against the the cylinder causes too much drag and a speed where the wind resistance causes too much drag.
 
In fuel mileage competitions, when a driver reaches this speed, they shut off the gas and let the kinetic energy turn the engine, or they just switch the engine off and shift into neutral, if they can save more fuel than what it takes to start the engine.
 
I'm not saying anybody outside a competition should drive like this, that would be horrible, but don't feather that gas pedal either. Keeping the engine at minimal load all the time won't help it and you are not winning on efficiency either.

Console Wars never change.. and we want to enlist on the side of Gamers. Tier Lists are a great way to rate gaming machines and there are PLENTY of rankings to go around. What PlayStation is the greatest of all time? What Xbox sits atop of the rest? Why is Nintendo always changing their naming?
 
 

An update, reviews are in, to predictable results. (predictable to anyone but the investors in the builders of this gadget)
 

 
How are still investors falling on the "projector works in daylight" scam... There sure are lots of dumb investors that fail to hire competent advice for their tech investment out there...

So it's like my phone but less accurate, and I can't see images?

EDIT: or effectively navigate by having a map I can see? Voice-only navigation leaves a lot to be desired.

these guys make me seem good at minecraft
 
and i suck at minecraft

That setup took a lot of effort, the silent guy sure deployed some annoying feedback!
 
I look forward to Linus and Ludwig playing. I'm sure there will be a lot of funny overreaction there too.
 
This is the thread where the games were discussed:
 

If you had 16GB of RAM, in theory you could run an 8B LLM on a smartphone with enough NPU performance to do the inferencing. So yes, these  0.08B to 3B models could run on an iPhone natively to perform complex iOS tasks.

Finally, an intelligent Siri that would be actually useful.

https://www.forbes.com/sites/kateoflahertyuk/2024/03/15/apples-new-ai-move-just-changed-the-game-for-all-iphone-users/?sh=3b8ce72b7277

Apple bought DarwinAI earlier this year, and the firm’s employees have joined the iPhone maker’s AI division, according to Bloomberg, which cites “people with knowledge of the matter, who asked not to be identified because the deal hasn’t been announced.”

Another advantage of DarwinAI that will benefit Apple specifically is the company has developed tech that can make AI systems smaller and faster. “That could be helpful to Apple, which is focused on running AI on devices rather than entirely in the cloud,” Bloomberg writes.

So yes, rumored to be implemented in iOS 18.

In the end Siri like workloads are best done on device given your want low latency and you have so many possible devices that could be making requests that doing this server side would just cost a small fortune.   Just piping Siri to a cloud based LLM would have so many downsides from a load perspective since the Siri load is likly has lots of spikes throughout the day, and you cant pipe requests half way round the world to have a single server handle that load due to massive added latency so you would end up with a lot of compute spread out around the world with much of it only be used at peak times for that timezone. 

Being able to do as much as possible locally on the device, even possibly letting that on device model figure out how to query remove data sources (even remote LLMs) would filter out most of this load... it would be a huge wast of $ to have an huge cloud based LLM handle people adding groceries to thier shopping list, or tuning on and off lights in the house (this sort of things makes up most of the Siri requests)... people are not asking Siri to write them a 100 page essay. 

I haven't watched, and won't. And those type of jokes (and type of show) isn't my cup of tea.
 
But if someone is offended, why don't you just not watch that show? I bet there are a billion other YT channels and one of them must be appropriate for your taste. 
 
I watch South Park, Monty Python and so on. I bet there are many people who feel offended by that. But I don't want those people going around and somehow force the makers to make a non-offensive version. Just watch something else if you don't like it. 

This can be used as an example of innuendo about it if you want.
No one seems to have a problem with it and the business is apparently doing well.

Suggestive, yet kinda funny too - Yes it's a play (Joke) about it.

Oh no that's so sad, I'm definitely gonna let that sink in for a while now.
 
Anyway, make sure to add me to the "literally worse than h*tler" list.
 
The WAN show did nothing wrong.

Welp, that's enough internet for today.

So you let a sexual harassment joke slip into the TechLinked video on April 3rd... Too soon buds, um rather poor taste. I'll let this one slip as a late April Fools joke, but, do better guys.

What? rent free.
 
Not everything men say is "sexual harassment".

I would have been more surprised if Intel announced profits for their Foundry division. They are investing heavily into catching up with TSMC manufacturing, which is fiendishly expensive, even with the generous USA and EU government handout they have been getting.
 
Intel says the 18A process is on track, which is really great news. Intel is leaving behind the notorious 10nm forever delay.
 
I'm more surprised that the investors have been caught by surprise by the news... Isn't the market supposed to "price in" expected news? That's like... their whole job!

Think this is supposed to be 0.08B or 80M

Financial market analysts don't know their asses from their elbows when it comes to tech, even "tech focused" ones. Intel making a massive loss in the lead up to new technology and products is a huge sign of good potential and that's why I would be recommending to invest because you'll miss the boat otherwise. That is obviously not actual financial advice but if you actually know the sector and the technology developments happening then nobody knows better than yourself. I mean that is exactly how I 3x-5x my money on AMD stock, I knew Zen was coming soon, I knew it was tracking rather well and the only direction the stock was going to go was up. Intel is a little different since it's stock value isn't in the actual gutter heh.

That might explain an investor presentation they did earlier this week. I wrote more about it at link below, although I focused on the tech more than the financials.
 
Without looking at the latest results, they did say they were reorganising how they're reporting foundry results. Before, product design teams ate some of the costs, but now foundry is more separated. Be aware it might not be a like for like comparison. They've been doing a lot to get their fabs up to speed and it isn't cheap. Getting the newer nodes going will save them money so that "loss" today is investing for the future. The more you spend, the more you save. Wait, wrong company 😄 
 
In short, I don't feel this is significant at all as long as it puts them in the right place going forwards.
 
 

Operating loss really isn't that surprising or against what Intel has been saying. They are investing in a lot of new things and expanding, you have to spend money first before you can make it and jointly customers.
 
People too easily forget, Intel silicon fabrication was the industry benchmark for so long, none of the knowledge and talent has left.