your take on ChatGPT: thoughts?

[manoeuver]

Messing around with it, Chatgpt 4 is worlds better then 3.5 is.
I'd rate it as a better mediocre copywriter with persistent problems with factual accuracy.

[karmachanic]

persistent problems with factual accuracy.

Sounds like the so-called news.

[dneal]

Open AI dropped "Sora" today. It's a prompt-based video generator, and it's really good.

The implications are pretty huge. Legitimate usage, it's a threat to Hollywood (which is largely actors on green-screen with armies of CGI programmers). The illegitimate uses for misinformation are obvious. You can find some samples here:

https://openai.com/sora

The second video using the prompt "Historical footage of California during the gold rush" is pretty amazing, actually.

[Jon Szanto]

Open AI dropped "Sora" today. It's a prompt-based video generator, and it's really good.

View the "stylish woman" vid and look at her legs. It happens not just once, but twice.

[dneal]

Open AI dropped "Sora" today. It's a prompt-based video generator, and it's really good.

View the "stylish woman" vid and look at her legs. It happens not just once, but twice.

They point out some of the issues with the other videos, but I missed that one. I was focused on the hands. Why AI can't get fingers right is something I don't quite understand.

That said, the "stylish woman" video is still pretty amazing (errors and all), from just a text prompt.

[Jon Szanto]

Oh, I'm not discounting the remarkable abilities. It's just that, while seeing what gets bobbled, I'm mindful that for every good purpose this might be turned to, I definitely believe it will be overwhelmed by the destructive uses of it. And the further they dial it in, the more excellent the representations, the worse the harm will be. Nothing to be done about it, of course.

[Yazeh]

I have to admit this is both frightening and exciting.

[dneal]

I have to admit this is both frightening and exciting.

In a nutshell.

[FredRydr]

Apologies if this was already cited: https://www.nytimes.com/2023/06/08/n...sanctions.html

[Brilliant Bill]

The tools of the greatest master manipulators through all of time are now in my hands --- and the hands of those who can harm me.

[Niner]

AI? We're playing Russian roulette with a gun someone else loaded while we weren't watching. And we don't know how many bullets were loaded or whether it's a revolver or a belt-fed machine gun. I know something about algorithms, machine learning, and firearms. I'd say AI is a machine gun.

[Brilliant Bill]

Well over 40 years ago I got into "computers." I was the guy with the 300-baud acoustic coupler sending single line messages to "bulletin boards." You may recognize the type, and I was a rebel about it. In the large organization I was a part of it took great battles, but I broke my group away from the distributed systems model and set us up with individual desktop computers --- much to our greater productivity and satisfaction. I was also writing a newspaper column at the time and took a lot of flak for suggesting that humanity may be on the verge of transitioning from a carbon-based life form to a silicon-based lifeform. There are potential great advantages from a practical standpoint. Obviously, there are substantial issues regarding the spiritual side of human nature. Do we simply give up our "better angels" for the benefits of better "software"? And if our "better angels" are no better than allowing us to persecute and murder six million at a clip, are those "better angels" worth it? I see it as quite a topic for discussion.

Anyway, this AI business seems all a part of the transition I predicted so long ago. And it's quite scary because change is always scary. As a seer once said: "Everybody wants to go to heaven, but no one wants to die."

[Mık Joger]

My thoughts on ChatGPT: it's amazing! But to expand knowledge, exploring informative sources like www.sspdaily.com is crucial. They offer valuable insights that complement ChatGPT's capabilities, enriching our understanding.

[robert.bradley.5]

When I'm reading I run across allusions to people and events that I may recognize but want understand better to appreciate the book's context or what the author had in the back of their mind when writing the passage. Often I'll resort to either Wikipedia or to ChatGTP. I have to rely on my intuition. If something doesn't smell right, I'll google and pull up some more trustworthy sources. So far ChatGTP has been a useful tool, but I wouldn't be referencing it in a seminar paper or a PhD dissertation. Like most new tools and artifacts of human ingenuty, I have ambiguous feelings, even suspicion, but rejecting a tool outright doesn't seem right.