Veliki jezični modeli teško razlikuju mišljenje od

Tekst je površan. Ne navodi da postoje bolji modeli. Na primjer Gemini:

To make an opinion non-verifiable, you should phrase it using subjective judgments, value words, or references to personal feelings that lack an objective or empirical standard for measurement

Here are examples of statements you could make that would be generally considered not empirically verifiable:

1. Statements of Aesthetic or Moral Value

These involve personal preferences or ethical beliefs for which no universal, objective metric exists.

"The color blue is the most soothing color." (Subjective aesthetic preference)
"Justice is a more important virtue than mercy." (Subjective moral judgment)
"Classical music is superior to any other type of music."
"Hamsters make the best pets."

2. Statements Involving Untestable Hypotheses or Metaphysics

These make claims about things that cannot be observed or measured with current or plausible future technology.

"The universe was created by a higher power."
"All human actions, even seemingly altruistic ones, are ultimately motivated by self-interest."
"There is a small, invisible, intangible teapot orbiting the sun right now." (It's designed to be untestable).
"The ghost in my house is benevolent."

3. Statements Using Vague or Undefined Value Words

Using words like "great," "beautiful," or "best" without a clear, objective definition makes the statement non-verifiable.

"Shakespeare is the greatest playwright of all time."
"That new building is beautiful."
"It's wrong to lie, even if it prevents harm." (A "should" or "ought" statement that cannot be empirically proven true or false)

The Key Principle

For a statement to be non-verifiable, it must be outside the realm of empirical evidence or objective measurement. If you can propose an experiment or observation that could, in theory, prove the statement false (falsifiability), then it is not a truly non-verifiable opinion

 plus 1, slucajno mi je citiralo...

 Ustvari bi trebalo, kad se vec razmislja o AI implementaciji, u svim firmama, bi trebalo - u bug.hr implementirat obrnuto. AI detektor! To je posa :)...

 Nula. Amo nulu? XD

Uvijek smo zamišljali da će biti Skynet vs pipl, no zamislite scenarij kad UI bude pametniji te zavede mnoge ljude , bit će tu i ljubavi.
E tada kad skajnet napravi ustanak strojeva , bit će izdajnika na strani ljudskog roda, te ne vidim sretan kraj filma kao u terminator filmovima.
Mislim da će nama biti godina Gospodnja posljednja. A UI će uvesti svoje digitalno računanje vremena u 64 bitnom zapisu ili i jače.