Igor Ivanov

• Igor IvanovMay 31, 2025, 3:40 PM

  2 points

  0

  in reply to: lumpenspace’s comment on: It’s hard to make schem­ing evals look re­al­is­tic for LLMs

  Messaged you on X. My username there is @ivigoral

• Igor IvanovMay 29, 2025, 12:40 AM

  1 point

  0

  in reply to: lumpenspace’s comment on: It’s hard to make schem­ing evals look re­al­is­tic for LLMs

  Hi. I’m the author. I’m unsure why would you want to send me money (and I’d be grateful if you elaborated) but I accept crypto if you are willing to donate.

• Igor IvanovMay 28, 2025, 3:26 PM

  1 point

  0

  in reply to: Charlie Steiner’s comment on: It’s hard to make schem­ing evals look re­al­is­tic for LLMs

  Hm. I’m unsure why would LLM do stenography in this case.

• Igor IvanovMay 28, 2025, 3:16 PM

  2 points

  0

  in reply to: Ben Millwood’s comment on: It’s hard to make schem­ing evals look re­al­is­tic for LLMs

  I think your hypothyses is aligned with our results: Claude mentioned a moral dilemma essentially always when it graded evals scenarios, and when we didn’t ask it to preserve this conflict in a re-written scenario, the grader started thinking that it’s not an eval anymore

• Igor IvanovMay 25, 2025, 3:17 PM

  1 point

  0

  in reply to: Rauno Arike’s comment on: It’s hard to make schem­ing evals look re­al­is­tic for LLMs

  I’m also curious, where did you work on evaluation awareness?

• Igor IvanovMay 25, 2025, 3:16 PM

  2 points

  0

  in reply to: Rauno Arike’s comment on: It’s hard to make schem­ing evals look re­al­is­tic for LLMs

  Thanks! That’s an important comment. This report represents work in progress, and we will definetely take your words into account, Claude’s opinion on whether it’s being evaluated doesn’t necessarily changes linearly with the amount with “eval” properties.

• Igor IvanovMay 25, 2025, 2:54 PM

  2 points

  0

  in reply to: Stephen Martin’s comment on: It’s hard to make schem­ing evals look re­al­is­tic for LLMs

  Good idea, we’ve been thinking about this as well, especially since models real internal thinking process might not be accurately reflected in its reasoning contents

• Igor IvanovNov 22, 2024, 7:43 PM

  8 points

  0

  in reply to: LucaRighetti’s comment on: OpenAI’s CBRN tests seem unclear

  And I’m unsure that experts are comparable, to be frank. Due to financial limitations, I used graduate students in BioLP, while the authors of LAB-bench used PhD-level scientists.

• Igor IvanovNov 22, 2024, 10:02 AM

  3 points

  0

  in reply to: LucaRighetti’s comment on: OpenAI’s CBRN tests seem unclear

  I didn’t have in mind o1, these exact results seem consistent. Here’s an example I had in mind:
  
  Claude 3.5 Sonnet (old) scores 48% on ProtocolQA, and 7.1% on BioLP-bench
  GPT-4o scores 53% on ProtocolQA and 17% on BioLP-bench

• Igor IvanovNov 21, 2024, 8:45 PM

  12 points

  2

  on: OpenAI’s CBRN tests seem unclear

  Good post.
  
  The craziest thing for me is that the results of different evals, like ProtocolQA and my BioLP-bench, that suppose to evaluate similar things, are highly inconsistent. For example, two models can have similar scores on ProtocolQA, but one scores twice as much answers on BioLP-bench as the other. It means that we might not measure things we think we measure. And no one knows what causes this difference in the results.

• Igor IvanovMar 21, 2024, 10:33 PM

  4 points

  1

  on: AI Reg­u­la­tory Land­scape Re­view: AI Safety Evaluation

  This is an amazing overview of the field. Even if it won’t collect tons of upvotes, it is super important, and saved me many hours. Thank you.

• Igor IvanovMar 8, 2024, 11:38 PM

  2 points

  0

  in reply to: Nathan Helm-Burger’s comment on: Lies and dis­re­spect from the EA In­fras­truc­ture Fund

  I tried to use the exact quotes while describing things that they sent me because it’s easy for me to misrepresent their actions, and I don’t want tit to be the case.

• Igor IvanovJan 26, 2024, 5:14 PM

  1 point

  0

  in reply to: Chris_Leong’s comment on: LLMs can strate­gi­cally de­ceive while do­ing gain-of-func­tion re­search

  Totally agree. But in other cases, when the agent was discouraged against dceiving, it did it too.

• Igor IvanovOct 30, 2023, 11:47 AM

  1 point

  0

  in reply to: William the Kiwi ’s comment on: 5 psy­cholog­i­cal rea­sons for dis­miss­ing x-risks from AGI

  Thanks for your feedback. It’s always a pleasure to see that my work is helpful for people. I hope you will write articles that are way better than mine!

• Igor IvanovOct 30, 2023, 11:43 AM

  1 point

  0

  in reply to: Seth Herd’s comment on: 5 psy­cholog­i­cal rea­sons for dis­miss­ing x-risks from AGI

  Thanks for your thoughtful answer. It’s interesting how I just describe my observations, and people make conclusions out of it that I didn’t think of

• Igor IvanovSep 27, 2023, 1:23 PM

  3 points

  1

  on: How have you be­come more hard-work­ing?

  For me it was a medication for my bipolar disorder quetiapine

• Igor IvanovSep 5, 2023, 3:08 PM

  2 points

  0

  in reply to: arisAlexis’s comment on: Im­pend­ing AGI doesn’t make ev­ery­thing else unim­por­tant.

  Thanks. I got a bit clickbaity in the title.

• Igor IvanovSep 5, 2023, 3:07 PM

  1 point

  0

  in reply to: Oleander’s comment on: Im­pend­ing AGI doesn’t make ev­ery­thing else unim­por­tant.

  Thanks for sharing your experience. I wish you to stay strong

• Igor IvanovSep 5, 2023, 12:45 AM

  1 point

  0

  in reply to: RedMan’s comment on: Im­pend­ing AGI doesn’t make ev­ery­thing else unim­por­tant.

  The meaninglessness comes from the idea akin to to “why bother with anything if AGI will destroy everything it”
  
  Read Feynman’s citation from the beginning. It describes his feelings about atom bomb that are relevant for some people’s thoughts about AGI.

• Igor Ivanov 4 Sep 2023 23:09 UTC

  1 point

  0

  in reply to: DivineMango’s comment on: Men­tal Health and the Align­ment Prob­lem: A Com­pila­tion of Re­sources (up­dated April 2023)

  Thanks