Artificial Intelligence

[macdoc]

Brave is brilliant.
For VPN use ProtonVPN

[Tero]

I asked Google Gemini if any use is made of user feedback.
It replied:
Direct System Tuning: Corrections are analyzed to understand where the AI misinterpreted language, relied on unreliable sources, or hallucinated facts. Google's engineering and safety teams use this information to update technical guidelines and prevent similar errors in the future.

Human Review and Evaluation: Your reported examples are often sent to human evaluators. These trained raters review the model’s outputs and your corrections to grade the AI’s performance, helping Google safely and effectively fine-tune the system.

Newslink:
https://www.pbs.org/newshour/politics/g ... went-viral

[Tero]

There are some tools:
Look for Icons: Locate the Thumbs Down icon or the three-dot More menu at the bottom of the AI's response.Select a Reason: Choose the specific issue, such as Factually inaccurate, Offensive, or Not helpful.

[Brian Peacock]

Anyone else come across these rather dodgy AI Feynman videos on YouTube?

[Tero]

LLMs can be copied:
(Don't give it to N Korea)
Duplicate copies of a model are used constantly in machine learning. Here are the core facts about duplicating an LLM:

The Core Elements: A fully trained LLM consists of two main parts: the weights (the learned knowledge) and the tokenizer (the vocabulary that translates words into numbers). Copying both yields an identical twin of the model.

Storage and Transfer: You can zip these files up and transfer them across servers, share them on repositories like Hugging Face, or deploy them on different local computers.

Cost vs. Training: Copying the model is instantaneous and free. In contrast, training the model requires vast amounts of computing power, data, and time.Independent Futures: Once duplicated, each model acts entirely independently. You can fine-tune Copy A for a medical role, while leaving Copy B for a legal role; their changes will not interfere with one another.

[Brian Peacock]

Oh dear.

[L'Emmerdeur]

Not surprising that the Grok experiment was severely dysfunctional and resulted in bringing about its own extinction. I understand that Musk has to some extent insisted on imprinting his worldview on it.

[Tero]

A bit comical that Musk and Bezos want to save their minds in some AI form. With Bezos...there is almost nothing there.

[Brian Peacock]

Not surprising that the Grok experiment was severely dysfunctional and resulted in bringing about its own extinction. I understand that Musk has to some extent insisted on imprinting his worldview on it.

The Google model "penalised harmony and subsidised chaos". Well it is their business model of course, so at least it's consistent.