this post was submitted on 23 Sep 2024
-1 points (44.4% liked)

Ars Technica - All Content

91 readers
61 users here now

All Ars Technica stories

founded 4 months ago
MODERATORS
top 6 comments
sorted by: hot top controversial new old
[–] xantoxis 3 points 2 months ago (1 children)

I can't imagine anything that screams "this is a scam" louder than phrasing it "a few thousand days" as if we're children who will only hear the word "days" and forget to do the math.

[–] [email protected] -3 points 2 months ago

About ten years. (Don't correct me reddit: it's vague in any unit)

Still a fairly bullish prediction.

[–] Bob_Robertson_IX 3 points 2 months ago

The things to remember is that before OpenAI released ChatGTP, Google already had their own version that they were aiming for AGI, but they had no intention of releasing it for public consumption. They knew it would bring prosperity and wanted to keep that all inhouse.

For good or bad, this is coming and I personally believe that any models that have been trained on public information (or publicly posted information) should be considered part of the public domain. No corporation or government should own an AI.

[–] Grogon 2 points 2 months ago* (last edited 2 months ago)

I get where he is coming from.

The issue we are facing though is that a large percent of humans can't deal with AI and most laws aren't even suitable for coming issues. AI is advancing way too fast for majority of society and at the same time for a large percent of our worlds population internet is a luxery and another part of the population isn't even connected to the internet.

With that being said, countries that are behind will be even more behind.

Captchas now are just clicking on pictures but they are getting harder now.

In a year we will have captchas like this:

“Please click each image that is least likely to have a 50-70% chance of producing exactly 15 nm3 of dopamine in the third mammal that is physically closest to you.”

It was all fun and games a few years ago watching videos and detecting fake videos by counting fingers but now that isn't possible anymore. 20% of humans can't logically think about things out of the box and see fakes instantly. They believe in most BS.

I still don't see what we are going to do with 50% of humans losing jobs. Let them sit around? 50% of the working people who can't be replaced by AI will be mad asf if they have to work for money while the other 50% do nothing for the same amount.

It's interesting because I think this won't be an issue in about 50-70 years. It's also interesting because I am 31 years old. I am going to have to observe this transition.

I really hope AI will get us off the screens and back into the "real world". We need less screen time and more social time. If AI adds more screens to us we are doomed.

[–] edgemaster72 1 points 2 months ago (1 children)

What a weird fucking way to phrase that, who thinks about days by the thousands

[–] [email protected] 3 points 2 months ago

The guy probably thinks in numbers a lot.