Tannhuas

[Tannhuas @Tannhuas]

@MrEwanMorrison The article is very short and just makes a lot of claims without diving into them very much. Can you explain how the creation myth applies to doomerism?

[Tannhuas @Tannhuas]

@ZackKorman @StozekStyczny Is there a post or article that outlines your position on this point specifically?

[Tannhuas @Tannhuas]

@clairlemon @CunCrypto Why are you imposing rules on the AI that don’t exist on in reality? Spell out what part in the scenario you disagree with, or explain why the AI is forbidden from using humans as a means to achieve its goals.

[Tannhuas @Tannhuas]

@clairlemon @JeffLadish And how specifically will these limits prevent the AI from building its own infrastructure? Everyone agrees the limits exist, so this is a moot point

[Tannhuas @Tannhuas]

@MelMitchell1 @ZeeshanZiaML If you actually pressed MJ to spell out his disagreements with the premises that underpin why superintelligence is possible, you would probably get an answer though, rather than just attacking the messenger.

[Tannhuas @Tannhuas]

@TaliaRinger AFAIK the first stage of alignment is getting something smarter than you pointed in any direction at all, and this is a pretty major step, which is independent of the values of the person or groups aligning the AI.

[Tannhuas @Tannhuas]

@perrymetzger It’s satire, well done Perry!

[Tannhuas @Tannhuas]

@perrymetzger @allTheYud @DaystarEld @Joe_S_Miller Genuine question - you keep asking Eliezer to debate you, over and over, but surely to increase your chances of debating him you wouldn’t be so hostile and nasty right?

[Tannhuas @Tannhuas]

@mattparlmer @___frye I thought his whole belief was no one on earth can currently pull off alignment? That’s why we should slow down? What’s the point of slowing down if he thought he could solve alignment

[Tannhuas @Tannhuas]

@deepfates @allTheYud Who was the original tweet about?

[Tannhuas @Tannhuas]

@kromem2dot0 @QiaochuYuan @RichDecibels @DaystarEld Confused by this analogy?

[Tannhuas @Tannhuas]

@JVNs_ghost @DKokotajlo @ASM65617010 The post is arguing against the view that this is a fringe opinion, not whether it’s true or not.

[Tannhuas @Tannhuas]

@SigmaSyndikate @maxim_enis @krherr If we wait until ASI capabilities exist in real life, do you think it will be too late to regulate at that point? Shouldn’t we have a discussion now about whether they can exist in principle?

[Tannhuas @Tannhuas]

@MatthewJBar @AndyMasley Agreed, but working on the problem and failing is still better than avoiding the fact that the problem exists.

[Tannhuas @Tannhuas]

@TaylorLorenz @matvelloso But given you believe that he believes he is telling the truth, why do scary narratives matter? Isn’t the relevant point whether the scary narrative is true?

[Tannhuas @Tannhuas]

@JustSomeGuy1970 @greatbong This only makes the problem of aligning AI harder though, and this is a very narrow and undefinded domain which has little impact on capabilities

[Tannhuas @Tannhuas]

@TaylorLorenz @HumanHarlan What evidence are you looking for? The problem is not specific instances of negligence, which evidently do occur and whistleblowers should be encouraged, but the whole race to making smarter AI’s itself. I think he made this quite clear.

[Tannhuas @Tannhuas]

@mathepi @canadabirdie What will humans have to offer in these negotiations?

[Tannhuas @Tannhuas]

@logisti00 @canadabirdie So you concede their is no current known solution? Why not slow down then rather than racing against the clock for no reason?