Enjoy the benefits of a premium membership!
Upgrade Account
-
Welcome to TalkBass! Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
AI answers the question we all want to know...
- Thread starter cvsurg
- Start date
The only thing I can say about AI with any certainty is that someone is going to make a lot of money, and it probably won't be me.
The goal of ChatGPT in its current form is not to give correct answers, its goal is to be able to form coherent and grammatically correct sentences! It is a next level autocorrect! Not some oracle that brings truth!
If I ask for 'the Netherlands' and 'King' it will come up with King Willem Alexander, the current Dutch king, because these words usually appear close to each other in texts and therefore has the highest probability of being 'right'. Just like the autocorrect on your phone!
It is just an enormous analytical and statistical machine whose intended produce is text which is basically indistinguishable from humans, based on and learned from millions and millions of linguistic patterns in existing texts, nothing more, nothing less! In that it succeeds and that in itself is very impressive!
The biggest risc , like @Wood and Wire already stated, lies not in that it would become sentient, but rather in it producing misinformation, intentionally or not, that gets added to the public cesspool that is the Internet.
ChatGPT, but trained not on texts but on millions of amino acids and proteins, has succeeded in synthesising a new protein from zero, that can be used to combat bacteria (ChatGPT in the biological world is the first to synthesize a new protein from zero, published in the sub-journal of Nature - chemborun.com)! Now that's impressive!
If I ask for 'the Netherlands' and 'King' it will come up with King Willem Alexander, the current Dutch king, because these words usually appear close to each other in texts and therefore has the highest probability of being 'right'. Just like the autocorrect on your phone!
It is just an enormous analytical and statistical machine whose intended produce is text which is basically indistinguishable from humans, based on and learned from millions and millions of linguistic patterns in existing texts, nothing more, nothing less! In that it succeeds and that in itself is very impressive!
The biggest risc , like @Wood and Wire already stated, lies not in that it would become sentient, but rather in it producing misinformation, intentionally or not, that gets added to the public cesspool that is the Internet.
ChatGPT, but trained not on texts but on millions of amino acids and proteins, has succeeded in synthesising a new protein from zero, that can be used to combat bacteria (ChatGPT in the biological world is the first to synthesize a new protein from zero, published in the sub-journal of Nature - chemborun.com)! Now that's impressive!
Last edited:
Whelp, before the Covid madness I used to think Stephen King's _The Stand_ scenario (with the Captain Trips virus) was imaginary. Now I'm not so sure...
How long will it take for those AI programs to give answers according to what company is paying for publicity?
Or is it already happening?
Or is it already happening?
No, not at all.
I just don't believe this one will be different, that's what I meant.
In my opinion, presenting the adds in form of text, like the one in the OP, makes it more believable. But I have no doubt it will be payed by the companies, if it isn't already.
Just like all the other crap online, as you stated.
Fair enough...
I just think many people don't see AI and it's possibilities for what it is...and perhaps even more: what it isn't! It's complex technology so people project all kinds of fantasy and fiction onto it, along with fears. It's the same thing with every technological advancement (notice I don't use the word 'improvement'!). It'll be used for good, it'll be used for bad. We'll see what it'll become, but intriguing it is!
And perhaps someday it will be able to really argue in favour of tort!
I just think many people don't see AI and it's possibilities for what it is...and perhaps even more: what it isn't! It's complex technology so people project all kinds of fantasy and fiction onto it, along with fears. It's the same thing with every technological advancement (notice I don't use the word 'improvement'!). It'll be used for good, it'll be used for bad. We'll see what it'll become, but intriguing it is!
And perhaps someday it will be able to really argue in favour of tort!
We got Chat GPT to write an exit letter for our singer from her management agency. Great fun.
A piece of software that has learned to scrape words and make them into sentences.
I asked ChatGPT six slightly obscure questions for which the answers are there on Wikipedia. It got five wrong, two hilariously badly. As my old teachers would have said: 5/10, see me.
That's just as well, as in my experience, it doesn't yet achieve this.
So it's a bit too good for posting comments on the internet, then - inaccurate and with good grammar is too impressive. Still, beats any drummer apart from Bill Bruford and Neil Peart.