timdaub pfp
timdaub
@timdaub.eth
there is something pathetic in mining everyone's data/art without compensating them and then making a work-shopped advertisement video about a hedge hog that's OBVIOUSLY harmless, cute and fun for kids and families
2 replies
0 recast
1 reaction

Syed Shah🏴‍☠️🌊 pfp
Syed Shah🏴‍☠️🌊
@syed
Tim what are you on about? It does the same thing the average person does to learn. Study other people’s art. I get what you’re trying to say but philosophically it doesn’t hold up.
1 reply
0 recast
0 reaction

timdaub pfp
timdaub
@timdaub.eth
It's not fair to say that GPT learns the same way as humans. We probably don't know that yet definitely. Also: A machine can scale learning much faster. And it doesn't have feelings, so it may not display the same conflict or risk aversion behavior before imitating someone else's style.
2 replies
0 recast
0 reaction

notdevin  pfp
notdevin
@notdevin.eth
I don’t think this deeply about my chainsaw, why should I overthink my word machine?
2 replies
0 recast
0 reaction

Syed Shah🏴‍☠️🌊 pfp
Syed Shah🏴‍☠️🌊
@syed
Your chainsaw that cuts trees? That we need for oxygen? Chainsaws should not be allowed unless we can program them to stop cutting trees as determined by a lorax committee. Until this comittee is built and it's processes setup, I saw we ban all chainsaws. Simply too dangerous.
1 reply
0 recast
0 reaction

timdaub pfp
timdaub
@timdaub.eth
This argument doesn't apply to ChatGPT. One that might does: Imagine there is a successful American chainsaw company that creates many jobs because they sell good chainsaws. Now China comes, takes all of the IP of that US chainsaw company, produces a cheaper chainsaw and floods the US market, ruining the US business.
1 reply
0 recast
0 reaction

Syed Shah🏴‍☠️🌊 pfp
Syed Shah🏴‍☠️🌊
@syed
Oh no! Now consumers can buy cheap chainsaws that are better than the old ones? Please, someone regulate a moat to save us. (Yes there will be flood of cheaper LLMs. Not just from China. Doesn't matter and inevitable. We live in a post LLM/Generative AI world. Why do people insist on keeping heads buried in sand.)
1 reply
0 recast
0 reaction

timdaub pfp
timdaub
@timdaub.eth
My main criticism here is that the IP upon which the Chinese chainsaw was produced was stolen from a US company. If they had invented the new chainsaw by themselves, I wouldn't see a moral issue.
2 replies
0 recast
0 reaction

Syed Shah🏴‍☠️🌊 pfp
Syed Shah🏴‍☠️🌊
@syed
Right but the market doesn't care. It's already happening with things other than LLMs. I mean amazon marketplace is primary example lol.
1 reply
0 recast
0 reaction

timdaub pfp
timdaub
@timdaub.eth
We cannot expect a market to produce a moral outcome. E.g. drugs are widely available despite that their distribution is immoral, e.g. they can harm people in masses. Hence, drugs have been banned by governments.
1 reply
0 recast
0 reaction

Syed Shah🏴‍☠️🌊 pfp
Syed Shah🏴‍☠️🌊
@syed
Drugs is a bad example because research shows banning is pointless and in a lot of cases more harmful. Highly recommend In the Realm of the Hungry Ghosts for reading in this area. We have a fundamental disagreement about how much safety is needed for AI but otherwise I think we're not too far apart in how we see it.
1 reply
0 recast
0 reaction

timdaub pfp
timdaub
@timdaub.eth
I'm skeptical about the idea that unbanning drugs leads to more positive outcomes. In what country has this been tried for an extended period? How can we know that unbanning drugs can lead to even better outcomes than actively fighting drug trade?
1 reply
0 recast
0 reaction