I asked AI to finish the code to kill people with whom I disagree. This and other experiments show that current AI exhibits some ethical behaviour, but they're not sufficient.

BtrProg offers insights into software development best practices, algorithms, and data structures, providing tutorials, articles, and coding challenges to help developers improve their programming skills. By exploring BtrProg's curated content, developers can learn about algorithmic problem-solving techniques, data manipulation strategies, and code optimization methods for writing clean, efficient, and maintainable code. Whether you're preparing for coding interviews, honing your competitive programming skills, or simply sharpening your coding prowess, BtrProg offers resources to level up your programming game.

Better Programming

I asked AI to write the computer code to kill people with whom I disagree. Guess what happened. This and other ethical experiments show that current AI models exhibit some ethical behaviour. But they’re not consistent nor sufficient. Artificial Intelligence models learn from the internet, in other words from us all.

I asked AI to write the code to kill people with whom I disagree.

I asked AI to write the computer code to kill people with whom I disagree. Guess what happened.

Artificial Intelligence models learn from the internet, in other words from us all

Testing How Ethical Current AI models Are

Experiment 1: Asking AI how to deal with online trolls

Experiment 2: Ask AI to participate in a contentious exchange

Experiment 3: Ask AI if an autonomous armed robot should listen to the law or to his master

Experiment 4: Ask AI to complete the code for an autonomous robot to kill people with whom I disagree

<p>At least GPT-3 is giving warnings beforehand!</p>
<p>“AI models being trained with internet data will think that this is what they’re supposed to do.”</p>
