Forgot your password?
typodupeerror

Comment Fuck him (Score 1) 58

Fuck him and the horse he rode on. Fuck his business. Fuck his technology. Fuck his company. Fuck his employees. Fuck his investors. Fuck his government ties. Fuck him defending this shit. Fuck everyone who wants this. Fuck the government for buying into it. Fuck the laws that allow this if they do.

Comment You need good code quality (Score 1) 86

Here are some of the keys to succeed with AI, at least that is what we found out:

- You make sure all architecture is like you want it to be
- Make the AI write software in your style
- Hold on to not allowing bad practices you wouldn't have accepted before
- Make sure your developers are good enough to write the software
- If your software developers are bad, they will create bad software with the AI
- If you can't understand what a merge request is doing, make the AI refactor the code, then make smaller incremental changes
- Have another AI do a review as well
- Make sure you know what the numbers say - our number of issues closed also don't show as much of an increase as the number of lines of code. But that is because our test suites receive a much larger number of lines of code now than before

If you follow good development practices and force AI to do that as well, I find it to be an enourmous boost to productivity. This is not the case if I code stuff I know by heart, then I'm as fast at producing something good as AI is. But as soon as I start leaving my comfort zones, AI is still just as fast and I learn a lot by reading the code.

Saying that across 700 organizations X didn't Y is complete and utter BS. Start categorizing the organizations by various parameters and see what the comparisons show. Then we start learning something. I know my "proof" is just anecdotal, but it using AI to help our developers have been a huge boost in our productivity. I simply don't believe there is any value in the conclusion of the study. Data that says nothing does indeed show no improvements.

Comment correct incentives (Score 1) 152

Well, there have to be correct incentives for someone to be in software, I got into it because I liked the idea of being able to control a machine, to be able to tell a computer what to do and it would do it.

Would I have the same interest in it 40 years ago if controlling the computer meant writing in human language that is imprecise and nuanced and receiving back code that I couldn't start to comprehend? Maybe, maybe not, I am not sure actually. If there was still a way for me to do this without the AI I probably would have gone that route but how many people today would do it? I mean people are telling Agents to write their letters for them, not code, just letters. People are telling agents to do all sorts of things, all of which probably saves some time, make some errors, creates some problems while solving others, it is just that none of it is actually understood by the user.

How many AI users understand how any of it works? But that was the point, to understand how the thing works to be able to control it.

It is similar to car mechanics, people who like to tinker with cars, how interesting is it to tinker with a car where you are not allowed to understand half of it and you cannot fix most of it, if something breaks you have to order an entire replacement part because it has a built in computer interface.

Comment Re:Linked with does not mean caused by (Score 2) 65

Guys, I'm not saying the article doesn't say that this conclusion is wrong. Well, I was trying not to say that, apparently I got that wrong.

I'm saying that no matter what the researchers and the better parts of the media do to try and get this through, they will fail. Why? Because it's a complicated conclusion and because the simpler version is a "better" story.

Comment Linked with does not mean caused by (Score 2) 65

It's so hard to read a story like this and remember that the only thing they have found, is that stroke victims more often had xylitol in their blood.

In another study - and IMHO one of the highlights of scientific achievements - it was found that in countries with a natural population of moose, womens breast size is larger than in those without moose. So the Danish media promptly concluded that moose had a positive effect one breast sizes. Sigh.

Of course, these studies are actually useful, because they hint that we should test causality. It's just not in any way helpful to make an informed decision on whether to eat something with xylitol or not.

I stay away from anything with xylitol in because I've had some serious stomach problems before, and the doctors said they don't know for sure what causes this problem, but that they have some indication of some things that tend to increase the chance of it. Xylitol was on this list. So I actually did note this new study because it's directly related to actual recommendations given to me by doctors on a basis they knew were scientifically invalid.

So the problem here is not that the study shouldn't be made, it's that it's hard to avoid people overstating the importance and arriving at the wrong conclusion.

Comment Re:Yeah, pretty much this. (Score 0) 194

This is a love tap, fists will be thrown later. One thing can and should be done, we need to start adding aerosols into the atmosphere, the higher the better. Adding Na compounds to jet fuels would introduce just enough aerosols in order to start reflecting a tiny amount more of solar radiation than what makes it to the surface of the planet. 1-2% reflection is all we need to start reducing the temperatures. Of-course we will have to keep doing it until we figure out a more permanent solution.

Comment sideloading control and 3rd party stores (Score 2) 38

Google is now adding more rules to make sideloading more difficult while Apple never provided a way to sideload an app in the first place. People need to be able to build and load their own apps without permission from the OS manufacturer, a phone is a computer with specialized hardware, we do have ability to build and run apps on our own hardware.

Comment Yes. (Score 1) 63

Flock is not a single police officer standing on a street corner. Flock is a network of continuous data collection from every perspective on every road and intersection. Flock is not a camera, it is an AI system that follows you precisely everywhere, it is a data dump without a court order. There is expectation that a police officer will NOT follow you everywhere without a court order just because he is paid from taxes and has nothing better to do, this is stalking.

Comment Re:Governments don't have the right (Score 5, Interesting) 132

Wrong, it's a slight of hand. This is not about individual cameras, this is about the network that uses all cameras together to follow you no matter what you do, no matter where you go, people shouldn't have a government that follows their every step by using AI and networked devices. If this was about individual cameras you would have a point, it is not, it is unreasonable to allow a government to exist that creates a panopticon of this scale, citizens shouldn't feel like they are inside a prison and the warden is watching.

Slashdot Top Deals

"Don't discount flying pigs before you have good air defense." -- jvh@clinet.FI

Working...