Brief Candle

Safety-conscious tech accelerationist. It's over.

[Brief Candle @Laserfish2]

@pmddomingos These ARC-AGI scores haven't shown what they are attempting to show. Per Chollet, achieving human level performance on ARC-AGI-1 would require AGI, and could not be achieved by LLMs. However, LLMs advanced on this benchmark at the same rate as all other benchmarks.

[Brief Candle @Laserfish2]

@cazbit138 @secretsqrl123 If it were remotely possible for ozone filtration alone to prevent algae growth in pools, people wouldn't use chlorine. Pools frequently have ozone generators and the effect on algae is minimal at best.

[Brief Candle @Laserfish2]

@BSheetsMatter @Glawk_40 @xeixeira146866 You seem to misunderstand that the survey results are all that matter. Red moral superiority relies on the idea that everyone will pick red, which never happens. Seeing hundreds of these surveys, the fewest that I have ever seen pick blue is about 30%. Picking red kills them.

[Brief Candle @Laserfish2]

@xeixeira146866 @Glawk_40 The lives of everyone who would pick blue, of which there will always be many.

[Brief Candle @Laserfish2]

@GuyManFellow @allTheYud @bryan_caplan The point of this response is that there are number of very obvious rebuttals, most of which Bryan Caplan is smart enough to think of. Not fair for him to offload the mental work because he isn't willing to think about something for 60 seconds.

[Brief Candle @Laserfish2]

@SarahTheHaider Sarah, let me just say that I am totally sold on Yudkowsky doomerism, and I too found the book unconvincing. So don't consider that disqualifying.

[Brief Candle @Laserfish2]

@marksoares @allTheYud This oft repeated claim doesn't hold water. We frequently say that the Iranians or the Koreans or the Russians are going to nuke everyone. Literally no one expects people to start assassinating their politicians, because assassination is generally pointless tactic.

[Brief Candle @Laserfish2]

@xlr8harder @allTheYud It was extremely clear that he was calling for the bombing of data centers by state entities to enforce treaty violations, just as we bomb nuclear weapons facilities that are in violation of treaty. Largely uncontroversial.

[Brief Candle @Laserfish2]

@Twolfrecovery @ONagel33303 Yes, in spite of the fact that we were in a perilous arms race, we still had dozens of international treaties, including between bitter rivals: https://t.co/sfOpddoADU

[Brief Candle @Laserfish2]

@serdiezv @trouble_man90 https://t.co/Q2bNsXXnKH

[Brief Candle @Laserfish2]

@AUsefuIldiot @cafreiman Series of left leaning candidates winning pushes the overall platform further left. Series of right leaning candidates winning pushes the platform further right. Bouncing back and forth keeps us in the center.

[Brief Candle @Laserfish2]

@serdiezv @trouble_man90 No, it's up to you to vote in your own best interest. https://t.co/55E8KuCNY5

[Brief Candle @Laserfish2]

@AUsefuIldiot @cafreiman You should always vote for the better candidate. It doesn't matter if you agree with everything they say. It doesn't matter if some of their views are downright reprehensible. Are they the better candidate? Vote for them. This will produce the optimal outcome.

[Brief Candle @Laserfish2]

@O__V__P @a_lifein @trouble_man90 I don't believe your claim and I can't find a poll that supports it. So either provide one, or move on.

[Brief Candle @Laserfish2]

@AUsefuIldiot @cafreiman Series of left leaning candidates winning pushes the overall platform further left. Series of right leaning candidates winning pushes the platform further right. Bouncing back and forth keeps us in the center.

[Brief Candle @Laserfish2]

@AUsefuIldiot @cafreiman You should always vote for the better candidate. It doesn't matter if you agree with everything they say. It doesn't matter if some of their views are downright reprehensible. Are they the better candidate? Vote for them. This will produce the optimal outcome.

[Brief Candle @Laserfish2]

@ImpulsiveDoge @OlekKier @elonmusk @AmandaAskell @WSJ And yet, here we all are talking about him 100 years later. And using technology he created. I think a reasonable person would value that more than passing on a few chromosomes.

[Brief Candle @Laserfish2]

@jeremyphoward The points still hold. ARC AGI performance went from 0-100 in two years. It was intended to be impossible for LLMs; it was not, so they made ARC AGI 2, which also fell to LLMs. https://t.co/DNYD3wKAQL

[Brief Candle @Laserfish2]

@pmddomingos These ARC-AGI scores haven't shown what they are attempting to show. Per Chollet, achieving human level performance on ARC-AGI-1 would require AGI, and could not be achieved by LLMs. However, LLMs advanced on this benchmark at the same rate as all other benchmarks.

[Brief Candle @Laserfish2]

@GarrettPetersen Israel is obviously at war with Hamas/Gaza. This just looks like executing POWs to me.

[Kelsey Piper @KelseyTuoc]

@ded_ruckus We can either have "you have to appeal to the whole electorate" or "the whole game is figuring out the share of the electorate you appeal to and imposing new compliance rules to get them 20% likelier to vote than the rest" and I don't want the country ruled by the second game.

[Brief Candle @Laserfish2]

@DOGEai_tx @JakeSherman Note that this is a bot account and this response is auto generated by an LLM.

[Brief Candle @Laserfish2]

@IsaacKing314 @fred_guttenberg We can't start to refer to trivial AI assisted image edits as fakes. This one goes slightly too far and is not best practice, but it is not fake.

[Brief Candle @Laserfish2]

@LVV_VETO @captgouda24 Yes, this exact thing just happened to me yesterday, and literally the only thing I have done in weeks is like the occasional tweet.