On Software that is Safe, and Artificial Intelligence

[Stephen3141]

Rust developers at Google twice as productive as C++ teams

Code shines up nicely in production, says Chocolate Factory's Bergstrom

www.theregister.com

This is an OLD debate, in the software development community...
Most hardware engineers who wrote in C++, didn't know anything about creating larger class infrastructure. Nor were they interested in writing broad and shallow hierarchies of classes. They weren't really programmers, although they thought they were.
The attempts at "safe" computer languages, has a history of making programming languages (like ADA) with training wheels, that keep a programmer from some simple errors, but do not protect the SOFTWARE DESIGN from incompetent programmer choices.
Bill gates is responsible for the modern "operating systems bugs" by introducing the "Visual" languages, that can be coded in ASCII, transmitted, and dynamically compiled and executed on the fly. This opened the door to the modern massive threat of remote introduction of code as simple text, which is then compiled into executables upon arrival at its target. That bill, for this "convenient" and dangerous threat. It rejuvenated criminal hacking on a global scale.
---------- ----------
Scripting languages (like Python) seem easy to use.
Anyone can write their own scripting language, using one of the established computer languages as a compiled base to execute the scripts. This is just smart programming.
But the entire security of a computer, needs a holistic approach to security. The hardware engineering companies have got to ditch the amateur hardware engineers who write code -- knowing the syntax of a computer language, does not make one competent in Computer Science algorithm design, or code security design.
And, there should be legislation that stops manufacturing companies from hooking up their manufactured product to the Internet, using a simple processor to make the product "easy to use from a cell phone". If the security features of a "smart home" or "smart car" or a "smart phone" or smart jogging shoes are not engaged, then the "smart" product becomes potentially hackable from Communist china or Russia, or some criminal cartel in Mexico. Trying to make everything "smart" and "connected", is a really bad and unsafe idea.
If you don't want your private religious school to NOT be hacked by criminals using ransomeware, then don't connect vital school data to the Internet. This is a basic concept that any competent Computer Science student would understand, but that STEM teachers often don't understand. Or parents.
America's software is unsafe, often in its amateur design, and often in the way in which it is installed without security walls, or with the security walls disabled. We have yet to see what the Communist Chinese and Russians can really do, if the West got into an open war with the loose conglomeration of tyrants and radical Islamic groups and criminal gangs, that often work together to commit online crime.
---------- ----------
The situation will be MUCH WORSE, with AI tools becoming available, that can auto generate code to do complex infiltration and criminal activities using the Internet and computer chips.
---------- ----------
As usual, the American citizens are more interested in voting for politicians who know little about software professional standards and safety, and are more interested in bronze sneakers, massive lying on social media, tea sipping, and a neurotic attraction to 2 colors. Most politicians are incompetent to reason about software safety, or even know the Computer Science definition of Artificial Intelligence, or what it could or could not do.
The rosy developers of auto driving cars, should have done their homework on intractable algorithms (predicting human behavior). The rosy developers of cell phone social media interfaces should have seriously considered that they were developing a type of drug addiction for the younger generations. The functional illiteracy of the younger American generations is largely due to the playing of games online, and the lack of vocabulary that is used in texting.
Americans do not care about developing safe software applications. Not do they care about developing educated, disciplined kids.
Although there will be dozens of companies that CLAIM to be able to develop safe software, as there are NOW dozens of companies that CLAIM to be producing "AI tools", most of these claims will be false.
And, American citizens won't know this.

 

[AlexB23]

Rust developers at Google twice as productive as C++ teams

Code shines up nicely in production, says Chocolate Factory's Bergstrom

www.theregister.com

This is an OLD debate, in the software development community...
Most hardware engineers who wrote in C++, didn't know anything about creating larger class infrastructure. Nor were they interested in writing broad and shallow hierarchies of classes. They weren't really programmers, although they thought they were.
The attempts at "safe" computer languages, has a history of making programming languages (like ADA) with training wheels, that keep a programmer from some simple errors, but do not protect the SOFTWARE DESIGN from incompetent programmer choices.
Bill gates is responsible for the modern "operating systems bugs" by introducing the "Visual" languages, that can be coded in ASCII, transmitted, and dynamically compiled and executed on the fly. This opened the door to the modern massive threat of remote introduction of code as simple text, which is then compiled into executables upon arrival at its target. That bill, for this "convenient" and dangerous threat. It rejuvenated criminal hacking on a global scale.
---------- ----------
Scripting languages (like Python) seem easy to use.
Anyone can write their own scripting language, using one of the established computer languages as a compiled base to execute the scripts. This is just smart programming.
But the entire security of a computer, needs a holistic approach to security. The hardware engineering companies have got to ditch the amateur hardware engineers who write code -- knowing the syntax of a computer language, does not make one competent in Computer Science algorithm design, or code security design.
And, there should be legislation that stops manufacturing companies from hooking up their manufactured product to the Internet, using a simple processor to make the product "easy to use from a cell phone". If the security features of a "smart home" or "smart car" or a "smart phone" or smart jogging shoes are not engaged, then the "smart" product becomes potentially hackable from Communist china or Russia, or some criminal cartel in Mexico. Trying to make everything "smart" and "connected", is a really bad and unsafe idea.
If you don't want your private religious school to NOT be hacked by criminals using ransomeware, then don't connect vital school data to the Internet. This is a basic concept that any competent Computer Science student would understand, but that STEM teachers often don't understand. Or parents.
America's software is unsafe, often in its amateur design, and often in the way in which it is installed without security walls, or with the security walls disabled. We have yet to see what the Communist Chinese and Russians can really do, if the West got into an open war with the loose conglomeration of tyrants and radical Islamic groups and criminal gangs, that often work together to commit online crime.
---------- ----------
The situation will be MUCH WORSE, with AI tools becoming available, that can auto generate code to do complex infiltration and criminal activities using the Internet and computer chips.
---------- ----------
As usual, the American citizens are more interested in voting for politicians who know little about software professional standards and safety, and are more interested in bronze sneakers, massive lying on social media, tea sipping, and a neurotic attraction to 2 colors. Most politicians are incompetent to reason about software safety, or even know the Computer Science definition of Artificial Intelligence, or what it could or could not do.
The rosy developers of auto driving cars, should have done their homework on intractable algorithms (predicting human behavior). The rosy developers of cell phone social media interfaces should have seriously considered that they were developing a type of drug addiction for the younger generations. The functional illiteracy of the younger American generations is largely due to the playing of games online, and the lack of vocabulary that is used in texting.
Americans do not care about developing safe software applications. Not do they care about developing educated, disciplined kids.
Although there will be dozens of companies that CLAIM to be able to develop safe software, as there are NOW dozens of companies that CLAIM to be producing "AI tools", most of these claims will be false.
And, American citizens won't know this.

Bro, 99.9% of people will probably not understand what we are talking about, as not everyone is familiar with C++, Python or other programming languages / software security.

But yeah, AI is a double edged sword, as stated in an 2020 article by a Bush School college professor. I have AI installed locally on my laptop (runs 100% privately and does not connect to the internet), and it works well for explaining complex topics, and even analyze Bible verses. My computer does run hot, and use 95-100% of the 8 GB of RAM (Microsoft laptop from 2018) has when the AI is active.

Here is the AI from Mistral describing why rent is cheaper in Austria. Mistral 7B is a 7 billion parameter model which is 4 GB in size and is free to run, with no strings attached.

 

[Nithavela]

Rust developers at Google twice as productive as C++ teams

Code shines up nicely in production, says Chocolate Factory's Bergstrom

www.theregister.com

This is an OLD debate, in the software development community...
Most hardware engineers who wrote in C++, didn't know anything about creating larger class infrastructure. Nor were they interested in writing broad and shallow hierarchies of classes. They weren't really programmers, although they thought they were.
The attempts at "safe" computer languages, has a history of making programming languages (like ADA) with training wheels, that keep a programmer from some simple errors, but do not protect the SOFTWARE DESIGN from incompetent programmer choices.
Bill gates is responsible for the modern "operating systems bugs" by introducing the "Visual" languages, that can be coded in ASCII, transmitted, and dynamically compiled and executed on the fly. This opened the door to the modern massive threat of remote introduction of code as simple text, which is then compiled into executables upon arrival at its target. That bill, for this "convenient" and dangerous threat. It rejuvenated criminal hacking on a global scale.
---------- ----------
Scripting languages (like Python) seem easy to use.
Anyone can write their own scripting language, using one of the established computer languages as a compiled base to execute the scripts. This is just smart programming.
But the entire security of a computer, needs a holistic approach to security. The hardware engineering companies have got to ditch the amateur hardware engineers who write code -- knowing the syntax of a computer language, does not make one competent in Computer Science algorithm design, or code security design.
And, there should be legislation that stops manufacturing companies from hooking up their manufactured product to the Internet, using a simple processor to make the product "easy to use from a cell phone". If the security features of a "smart home" or "smart car" or a "smart phone" or smart jogging shoes are not engaged, then the "smart" product becomes potentially hackable from Communist china or Russia, or some criminal cartel in Mexico. Trying to make everything "smart" and "connected", is a really bad and unsafe idea.
If you don't want your private religious school to NOT be hacked by criminals using ransomeware, then don't connect vital school data to the Internet. This is a basic concept that any competent Computer Science student would understand, but that STEM teachers often don't understand. Or parents.
America's software is unsafe, often in its amateur design, and often in the way in which it is installed without security walls, or with the security walls disabled. We have yet to see what the Communist Chinese and Russians can really do, if the West got into an open war with the loose conglomeration of tyrants and radical Islamic groups and criminal gangs, that often work together to commit online crime.
---------- ----------
The situation will be MUCH WORSE, with AI tools becoming available, that can auto generate code to do complex infiltration and criminal activities using the Internet and computer chips.
---------- ----------
As usual, the American citizens are more interested in voting for politicians who know little about software professional standards and safety, and are more interested in bronze sneakers, massive lying on social media, tea sipping, and a neurotic attraction to 2 colors. Most politicians are incompetent to reason about software safety, or even know the Computer Science definition of Artificial Intelligence, or what it could or could not do.
The rosy developers of auto driving cars, should have done their homework on intractable algorithms (predicting human behavior). The rosy developers of cell phone social media interfaces should have seriously considered that they were developing a type of drug addiction for the younger generations. The functional illiteracy of the younger American generations is largely due to the playing of games online, and the lack of vocabulary that is used in texting.
Americans do not care about developing safe software applications. Not do they care about developing educated, disciplined kids.
Although there will be dozens of companies that CLAIM to be able to develop safe software, as there are NOW dozens of companies that CLAIM to be producing "AI tools", most of these claims will be false.
And, American citizens won't know this.

 

[AlexB23]

Alright, I summarized the article in the way anyone can understand:

Google's engineers are finding that they can get twice as much work done using the Rust programming language compared to C++. This is because Rust helps prevent common mistakes that can cause security issues. Other companies like Microsoft are also starting to use Rust for these reasons. However, some people argue that it's still important to consider other factors when choosing a programming language, such as how well it performs in terms of speed. But even though learning a new language like Rust can take time, Google's employees have reported feeling more confident in the correctness of their Rust code. In other words, 2024 could become the year of Rust.

 

[Stephen3141]

Bro, 99.9% of people will probably not understand what we are talking about, as not everyone is familiar with C++, Python or other programming languages / software security.

But yeah, AI is a double edged sword, as stated in an 2020 article by a Bush School college professor. I have AI installed locally on my laptop (runs 100% privately and does not connect to the internet), and it works well for explaining complex topics, and even analyze Bible verses. My computer does run hot, and use 95-100% of the 8 GB of RAM (Microsoft laptop from 2018) has when the AI is active.

Here is the AI from Mistral describing why rent is cheaper in Austria. Mistral 7B is a 7 billion parameter model which is 4 GB in size and is free to run, with no strings attached.
View attachment 345400

(As an aside...

does Mistral footnote where it is getting its information?
And, does Mistral rate the accuracy of its sources, and demonstrate why it
thinks that its sources are accurate?

Any human writer would be expected to do this.)

 

[AlexB23]

(As an aside...

does Mistral footnote where it is getting its information?
And, does Mistral rate the accuracy of its sources, and demonstrate why it
thinks that its sources are accurate?

Any human writer would be expected to do this.)

Mistral does not footnote where it gets info from, nor rate accuracy. So, humans can still outperform AI in that regard, or at least outperform Mistral.