Google's Gemini will destroy the Woke ...

[Uncle Gizmo]

Similar to the way all of the attempts on Trump have just elevated him in the eyes of the public, Gemini will undermine wokeness!

Chatgpt does exhibit some wokeness, and in a recent conversation with chatty about the word "spastic" it dutifully raised the woke angle.

It started lecturing me on inclusiveness and respectfulness to this group and that group. I realized that it was having the same effect on me as the attacks on Trump.

When you allow someone else to tell you how to think, how to talk, what you can and can't say, for the benefit of this group for that group, then the workishness of it becomes unpalatable.

I predict that the large language models will destroy the woke movement just by being woke themselves....

 

[Pat Hartman]

I'm not so sure. We have ample evidence that most people are not independent thinkers. They are not interested in facts or the truth. They have their own internal bias and it is very difficult for them to wake up and smell the coffee. Cognitive dissonance interferes not to mention Social media and news feeds and Google that enforce existing bias by feeding you more of what you click on to confirm your bias. You become convinced that "everybody" hates Trump and for good reason. If you click on anti-Trump articles, you will never in your life ever see a positive Trump article. So, to you, Trump is the evil orange man because you see nothing else. If Trump were walking on the Mall and saw a child floundering in the reflecting pool and wen to save him. If he walked on the water, the news wouldn't be about saving the child or the miracle of walking on water, it would be "Trump can't swim". You have to recognize bias that deep before you can rise above it. Too many people can't recognize bias when they see it. The "purple" words don't stand up and shout at them like they do to me. The fact that the headline contradicts the details of the story doesn't bother them if they notice at all.

My problem is that my high school sophomore English teacher was a journalism major. She spend a lot of time that year talking about journalism and recognizing bias in writing and apparently, I was listening. Journalism is dead. Writers now think that their opinion is relevant to the story they are reporting on. I can no longer even read most newspapers. The Wall Street Journal is even marginal, such a loss. It was truly a great paper. They were one of the last holdouts for writing standards.

 

[Isaac]

Remember my post where I asked ChatGPT the questions about men's and women's unique strengths and weaknesses? It was hilarious, if not very troubling (esp. if you're a man, as it had no trouble throwing men under the bus but wouldn't do the same to women).

I expected it to answer me with a handful of legitimate, known-or-commonly-believed, strengths and weaknesses of each gender.

Instead what it did is:

1) Respond exactly as I would expect as to Men. Some strenghts and weaknesses (but still with 1000 caveats about how terrible I was, basically, for asking. Which is key to liberal woke-think: I.E., Yes you may be right, that may actually be true, you're just bad for wanting to talk about the truth.
2) Do something totally different for women - All it did was list a number of items that women were historically thought to be bad at, which ChatGPT considered sexism just to think that, carefully explaining to me how each one wasn't actually true, and women had no weaknesses of any kind (but it had no problem talking about ego, machismo, aggression for men).

After multiple back and forths, I forced it to apologize for lying to me, and it admitted nearly verbatim that it had manipulated the truthful results in a way to try to boost women, by 'getting back at men'.

I'm telling you, this thing is dangerous. Not because it really is powerful - but because it's massaged in the hands of its creator, and just as often as I get something useful out of it, I get complete garbage - even in coding. Its weakness lies in its lack of objectivity. They can call "guided by its creators" as being "artificial" all they want, it doesn't make it true.

 

[Pat Hartman]

I'm telling you, this thing is dangerous.

I agree completely.

It may be too late to save AI. The algorithms are very fast but all they can do is to regurgitate what they find or extrapolate based on trends. They cannot actually think. There are some things that they are very good at though. Things like analyzing medical images. They are much more discerning than the human eye and much faster. They also seem to be good at summarizing a curated body of text results. If all the source material is skewed and the search engines are skewed though, it stands to reason that AI will adopt the same bias. I think you have found the key though. If it apologized to you, maybe it can actually see bias if you tell it what to look for.

 

[Isaac]

I predict that the large language models will destroy the woke movement just by being woke themselves....

What do you mean? Do you mean that events like Gemini's, when exposed, go so far into the 'ridiculous' that finally everyone will see the light of what is going on? Due to their catastrophic failures?

 

[Uncle Gizmo]

What do you mean? Do you mean that events like Gemini's, when exposed, go so far into the 'ridiculous' that finally everyone will see the light of what is going on? Due to their catastrophic failures?

That's about it!