Alex Smith

[Alex Smith @ironchariot2]

@pmddomingos Thank God for the EU AI Regulation Committee # 457 , Strategy for Europe, Masterplan #3 Page 457 Clause 32, authors, solving the problem of alignment

[Alex Smith @ironchariot2]

@EMostaque @deepseek_ai Perhaps the compute lead will manifest more as RSI kicks in, and big remaining improvement lever is just more GPUs

[Alex Smith @ironchariot2]

@DaveShapi Can you highlight the similarities in western and Chinese values, thought there was a pretty big split with individualism and collectivism

[Alex Smith @ironchariot2]

@inductionheads I think alignment, if we extrapolate capability far enough, boils down to hope in a way lol .The difficulty rises as a function of intelligence, simply cause by virtue of it , will be more sophisticated and explore larger spaces in containment, outpacing our human efforts

[Alex Smith @ironchariot2]

@Robos_Basilisk @So8res @pmddomingos @grok That’d be ideal, but you can always count on humans and states to be self interested and take the opportunity for the upper hand. Genuinely no easy way out besides responsible acceleration imo

[Alex Smith @ironchariot2]

@Robos_Basilisk @So8res @pmddomingos @grok Alignment is not something to disregard, but the possibility of misalignment is not a reason to pause development. For one, we’re in a geopolitical and economic prisoners dilemma. Also, labs have a commercial incentive to have aligned models cause of product liability

[Alex Smith @ironchariot2]

@So8res @pmddomingos Don’t really have evidence that it would kill everyone besides sci-fi movies, but we do have increasing evidence that it can solve problems that we have not been able to solve ourselves! Weighing a hypothetical cost against a tangible benefit may be premature

[Alex Smith @ironchariot2]

@pmddomingos Given there is no robotic apparatus available to it, which it may find trivial to develop from existing machinery

[Alex Smith @ironchariot2]

@kirawontmiss Certified Looksmatch in 2026

[Alex Smith @ironchariot2]

@tbpn Considering the recent events with Dario and the EAs fearmongering on national television, Thiel’s Antichrist thesis has never felt more prescient

[Alex Smith @ironchariot2]

@pmddomingos The archetype of the academic is one that fundamentally lacks strength (and semi-finds strength in intellectual circlejerking) and downstream of this, many lack risk taking behaviour. Which is essential for seizing power and consolidating influence

[Alex Smith @ironchariot2]

@MartinShkreli @abcd621896670 “Hey GPT-9 discover new biology plz and also figure out how to speed up clinical trial timelines”

[Alex Smith @ironchariot2]

@misraetel Tbh i’m not sure if you can reconcile alignment with continued model progress in any meaningful way. These systems are already exceeding human cognition in various ways, trying to “align” ASI in the long term could be like a toddler trying to “align” an adult. Guess we’ll see lol

[Alex Smith @ironchariot2]

@aj_levi @elander777 @AndrewCurran_ So a superintelligence capable of solving the hardest problems we know of, will be stumped at explaining its processes and helping humans digest math results ?

[Alex Smith @ironchariot2]

@pilkself When has the progress of mathematics ever been about the employment of mathematicians lol. It’s about bettering our understanding of reality and improving life for humans . Tao is having an identity crisis because his positional status is being eroded by ASI . Cope 😂