Science Thread

vincent

vincent

Grumpy Old Man
Staff member
Moderator - World Affairs
Weaasel said:
Has the United States Department of Energy ever before officially announced a break through in ignition - net positive power output from input with regards to nuclear fusion?
Click to expand...
Please, Log in or Register to view URLs content!

Before we get to visions of fusion power plants dotting the landscape, however, there's the uncomfortable fact that producing the 2 megajoules of laser power that started the fusion reaction took about 300 megajoules of grid power, so the overall process is nowhere near the break-even point.
 
G

gadgetcool5

Senior Member
Registered Member
The Western "ChatGPT" by OpenAI is really taking off. Released in November 2022, less than 2 months after the Biden administration imposed semiconductor restrictions on China designed to nuke its AI development, ChatGPT has racked up 1 million users in just 5 days, compared to weeks or months for Netflix, Facebook, etc.

It is basically an AI interface that you can ask anything and it will give you an intelligent answer. I can see this replacing Google search, I am already seeing it everywhere here in the States. I wonder does China have an answer to this or any equivalent?
 
B

BlackWindMnt

Captain
Registered Member
gadgetcool5 said:
The Western "ChatGPT" by OpenAI is really taking off. Released in November 2022, less than 2 months after the Biden administration imposed semiconductor restrictions on China designed to nuke its AI development, ChatGPT has racked up 1 million users in just 5 days, compared to weeks or months for Netflix, Facebook, etc.

It is basically an AI interface that you can ask anything and it will give you an intelligent answer. I can see this replacing Google search, I am already seeing it everywhere here in the States. I wonder does China have an answer to this or any equivalent?
Click to expand...
People are really overselling ChatGPT, simple task it can help but something more advance it starts to quickly break down.
China has wudao 2.0 i think it's still the biggest model ever build but it's not publicly available like ChatGPT. So how the produce answers are no one really knowns.
 
L

luminary

Senior Member
Registered Member
BlackWindMnt said:
People are really overselling ChatGPT, simple task it can help but something more advance it starts to quickly break down.
China has wudao 2.0 i think it's still the biggest model ever build but it's not publicly available like ChatGPT. So how the produce answers are no one really knowns.
Click to expand...
It's trained on a huge amount of internet data, which can allow it to sound intelligent at times (just by extreme pattern matching). However, it does not even pass the Sally-Anne test or it's variations, much less things like LSAT questions. It is not intelligent and it cannot extrapolate real information even when given all the clues. It will fail when you ask it to explain anything that is not extensively covered online. I.e. it's an improved chatbot and worse version of Google search with the same problems of creating misinformation and echo chambers. It's great for generating low quality writing like LinkedIn posts or middle school essays, but don't use it to write your academic papers or cover letters.
 
G

gadgetcool5

Senior Member
Registered Member
luminary said:
It's trained on a huge amount of internet data, which can allow it to sound intelligent at times (just by extreme pattern matching). However, it does not even pass the Sally-Anne test or it's variations, much less things like LSAT questions. It is not intelligent and it cannot extrapolate real information even when given all the clues. It will fail when you ask it to explain anything that is not extensively covered online. I.e. it's an improved chatbot and worse version of Google search with the same problems of creating misinformation and echo chambers. It's great for generating low quality writing like LinkedIn posts or middle school essays, but don't use it to write your academic papers or cover letters.
Click to expand...
I just put in an LSAT question into ChatGPT and it answered it correctly.

Then, I wrote an entirely new LSAT question myself where only the logic was based on an existing one, but with an entirely different topic and wording, thus being found nowhere on the internet. ChatGPT gave the right answer instantly.
 
Last edited:
L

luminary

Senior Member
Registered Member
IMO these tools are interesting and should be made available to researchers / developers to learn from and improve their own chatbot and NLP models. They should not be available to the general public in China.
Aside from being used for deepfakes, they can stifle artistic and literary development. Laymen cannot at eye's glance distinguish between human works and bot work (recycled and tropey).
Additionally, it encourages misanthropic impulses by providing fake online companionship for lonely young men / elderly / children and distances people from actual human interaction. We don't need more hikikomori in China.
Unleashing technology into the wild (the U.S. model) is irresponsible and has clear and serious effects on individual and societal health. China has been in constant transformation for the past century, it needs all the societal stability it can.
(source: me, a nationwide suicide counselor for the past four years)
 
You must log in or register to reply here.
Top