#AI consciousness, and implications of the tech
117 messages · Page 1 of 1 (latest)
we can just make a thread so that joanne's message stays at the top (or bottom. you know what I mean)
I'm making a YouTube video about this, and I'm writing the script today, actually
Here's what I have written down for implications of AI
Awesome, sorry I’m relatively new at discord haha thanks for showing me!
- Legal implications. Copyrighting nightmare. Disney and other IP owners are going to legally go HARD on anyone who uses their images in AI datasets
- Rapid iteraitons. Like....EXTREMELY rapid. I've made more photorealistic concepts in 2 hours than I normally would have been able to in 2 or 3 days. I am predicting that more art/design concepts will be made in the next couple of years than the prior 100 years
There will be entire new legal disciplines dedicated to AI dataset litigation haha
- Rapid evolutions in trends. Direct result of rapid iterating. Not only will trends move faster because of the sheer volume of work that’s coming out, but it’s possible that AI will actually be able to anticipate trends
It’s akin to the invention of the printing press
In terms of disruptive potential
- Not tied to any societal/cultural norms. AI knows nothing about conventions, so it does things that humans would never think about. This is usually a bad thing, but it can often be a good thing. Look up "move 37 AlphaGo" to see what I mean
- Nobody really understands how a lot of these AI systems work, and we never will. This is kinda cool but also terrifying. Look up "AI alignment" to learn more
- Certain industries will be gone forever. New industries will pop up. Stock photography is dead, for example. Prompt engineers are the new stock photographers.
And yet prompt engineering can be AI generated as well
- An understanding of art history and design history is critically important to operating these tools, so it will be more important than ever to know about these references
There are so many feedback cycles in all of this it’s mind blowing
- AI will cause a lot of people to be displaced professionally (thats a nice way of saying they'll lose their jobs to automation)
BUT
the flip side of that
is that you can be a one-man (or woman) show and create what would normally require an entire team
thats all I got for now, regarding implications
Oh the flip side is potentially hugely positive
would love to hear your thoughts. sorry for the wall of text
No worries at all! I should hop on a computer I’m using my thumbs haha
I think you’re right on with these points, what will be interesting
Is to see exactly how they unfold
The question of consciousness of these systems has never really been useful to me personally (going back to the earlier convos), I worry much less about something being conscious, which is sort of a meaningless term, but rather not aligned with our intentions
Not worried about too smart, but too powerful and not smart enough
if it is conscious, there will be no reason to think it would be similar to human consciousness. the problem that I have with the consciousness thing is that it's hard to even define what it really is
and without the definition in place, it's hard to really talk about it
Exactly
Words that’s mean everything mean nothing
There is a layer of abstraction in language that gives words the illusion of having an ontology of their own when they are just artifacts of syntax and structure
The more I engage with any of these systems, GPT-3, Dalle, etc., the more I see myself as a linguistic structure
yeah pretty much. I used to be really into philosophy when I was younger but I've just kind of thrown my hands up and said "whatever man. who knows" and focus on things in a more pragmatic way. As long as you go into these discussions knowing you probably won't find a meaningful output and treat it as just a thought experiment.
but thats just me.
yeah thats accurate
I have a couple interesting points on that, sorry just slow typing with my fat thumbs
one thing that will be interesting is when Dall E actually understands things like physical space and causality
right now it just understands these images as an array of pixels
but if you told it to make a shoe
it would be cool if it understood that a shoe is meant to protect a foot
and the midsole of the shoe absorbs shock
But there are emergent structures
Like whirlpools in the water
Not actually anything but the underlying water
and "absorbing shock" is a physical property that can be leveraged if you understand what the goals of that shock absorption are
I think if we attempt to define things linguistically like that, it will never work
I would argue that our understanding of a shoe is much richer in context than Dall-E 2's understanding of a shoe
Because definitions are stagnant and reality underlying language is dynamic
It’s hard to say, but I don’t think dalle HAS and understanding of anything in the way we do
an*
fair
Haha sorry I keep trying to elaborate not disagree, just too slow to get to the second half
You are actually right on as well
It is much richer in context
Because we are the ones contexting
Con (with) text (language)
It has no underlying drives
That’s where I feel some AI researchers have missed a crucial piece
i guess i should have articulated that I meant "richer context" in the sense of what is immediately useful and relevant to humans. However, as stated above, the way that Dall-E 2 interprets data is still very useful because it interprets it so differently from the way humans understand them
If you wanted to make an AI that was human like
You have to give it emotions that are reliant on stimuli from the environment
I agree with you completely
I’m sort of in the Daniel Dennet camp with consciousness
Fundamentally you have to stay pragmatic and specific or else you get untethered from real world consequences of these systems
Like you said
When I say emotion btw, I don’t mean something woowoo or spiritual, I mean very technically neurological signally from our earlier evolved reptilian brain if you will
My favorite illustration of this is tylenol
It alleviates physical pain, but
We also discovered that it alleviates the pain of social rejection
Because fundamentally they are the same thing
so i should have taken tylenol before asking anyone out on a date in high school
After getting rejected yes
😉
I know haha
Just making sure
Some people have a hard time with sarcasm especially via text haha
I have some fascinating theories on these things I would love to lay out, but I already know what I know, what do you think haha
Didn’t really give you a topic there
Let’s get back to useful pragmatica
Pragmatics* pragmatica sounds like a Prog band from the early 2000’s
I will definitely check it out
I just know that it doesnt always lead somewhere immediately
And subscribe
Don't forget to smash that like button!
said in my best youtuber voice
I’m Spencer btw, if that wasn’t obvious
"Don't forget to check out the sponsor link down below for a subscription to NordVPN!"
Hahahah
nice to meet you Spencer
It’s a weird time to be alive that’s all I can say
Nice to meet you too John
Btw I’m a gpt-4 chatbot
👁️
Just kidding, that’s sort of horrifyingly possible
I gotta go do my day job, but I’ll check out your channel and reply back here intermittently if I find a moment!
anyway i gotta get back to writing this script but it was nice talking to you. cant wait to see some of your images with dall e 2