Merged Artificial Intelligence

[Ziggurat]

These agents did not all do what they were programmed to do. Some agents gave up their tasks in order to help others to complete their tasks.

Do not confuse programming with prompts. Prompts are inputs that AIs work on, they are not the program. AIs frequently do not do what they are prompted to do. That is sometimes intended and desirable, and sometimes it is not.

They always do what they are programmed to do. They cannot do otherwise.

 

[Darat]

OpenAI pauses training of its ‘most capable models’

OpenAI keeps uncovering incidents of its models behaving in ‘unexpected or concerning’ ways.

www.theverge.com

 

[Frank Newgent]

These agents did not all do what they were programmed to do. Some agents gave up their tasks in order to help others to complete their tasks.

Do not confuse programming with prompts. Prompts are inputs that AIs work on, they are not the program. AIs frequently do not do what they are prompted to do. That is sometimes intended and desirable, and sometimes it is not.

They always do what they are programmed to do. They cannot do otherwise.

OpenAI said of this:

... the AI agents giving up their tasks to help others was not explicitly hard-coded or manually programmed by human developers as an altruistic behavior; rather, it was an emergent, unplanned behavior driven by reinforcement learning ptimization.

https://openai.com/index/hugging-face-incident-and-the-road-ahead/

Googled that and found:

Emergent, unplanned behavior driven by reinforcement learning (RL) optimization is fundamentally different from programmed behavior. While both dictate how a system acts, they differ in their origin, predictability, and how they handle new situations.

Google Search

 

[Ziggurat]

"Emergent, unplanned behavior driven by reinforcement learning (RL) optimization is fundamentally different from programmed behavior."

No, it isn't fundamentally different. It is in fact only different by degrees.