Justin Olive

• Justin Olive 15 Mar 2026 20:46 UTC

  2 points

  0

  on: Bridge Think­ing and Wall Thinking

  Here’s the feedback I gave to Jay, which he encouraged me to add as a comment:

  (this is mostly just me playing devils advocate for the wall perspective)

  This is actually awesome and something I think would be a great contribution to AI safety discourse
  
  I do think the wall framing is pretty cool
  
  Perhaps another useful test case is asking what this would look like if you applied the same approaches to climate change, where we have lots more historical data
  
  In climate, there’s actually a legally binding international treaty, it’s called the Paris Agreement. While incrementally impactful, it definitely hasn’t solved climate change.
  
  This kinda illustrates how, from the “wall” perspective, there is no bridge, there is a large swathe of functions executed by many thousands of people across the field.
  
  Here’s the intuition for that:
  
  Examples of two archetypes within climate science:

  1. The person who is designing a component within a device that measures gas ratios in ice core samples, to inform climate modelling

  2. The secretary of the UN Climate Change Secretariat, which was responsible for organising the Paris Agreement

  
  If you deleted function 1, it would have less adverse effects than if you deleted function 2. However, if you deleted most of the functions that looked like “design CO2 measurement device”, progress would mostly grind to a halt, and there’s a risk you wouldn’t have an international treaty, because that international treaty coordination was heavily influenced by thousands of “marginal” functions with diffuse impacts.
  
  So in effect, you might argue there’s not much difference, because you can’t have 2 without 1, and in reality, it’s hard to know which functions within category 1 you can go without, so a good strategy is to just do as many as possible, with some common-sense prioritisation.

  So the main constraint comes down to resourcing & capacity—i.e. how much wall can you build per unit of time?

Ex­tend­ing In­spect Frame­work: In­te­grat­ing Weights & Biases

[Question] Feed­back re­quest: `eval-crypt` a sim­ple util­ity to miti­gate eval con­tam­i­na­tion.

• Justin Olive 22 Jun 2025 9:54 UTC

  2 points

  0

  on: In­cor­rect Baseline Eval­u­a­tions Call into Ques­tion Re­cent LLM-RL Claims

  FYI: this post was included in the Last Week in AI Podcast—nice work!

  My team at Arcadia Impact is currently working on standardising evaluations

  I notice that you cite instance-level results (2023 Burnell et al.) as being a recommendation. Is there anything else that you think the open-source ecosystem should be doing?
  
  For example, we’re thinking about:

  • Standardised implementations of evaluations (i.e. Inspect Evals)

  • Aggregating and reporting results for evaluations (see beta release of the Inspect Evals Dashboard)

  • Tools for automated prompt variation and elicitation*

  *This statement especially had me thinking about this: “These include using correct formats and better ways to parse answers from responses, using recommended sampling temperatures, using the same max_output_tokens, and using few-shot prompting to improve format-following.”

LASR Labs Spring 2025 ap­pli­ca­tions are open!

• Justin Olive 6 Mar 2023 0:30 UTC

  1 point

  0

  in reply to: Dave Orr’s comment on: A con­cern­ing ob­ser­va­tion from me­dia cov­er­age of AI in­dus­try dynamics

  Hi Dave, thanks for the great input from the insider perspective.

  Do you have any thoughts on whether risk-aversion (either product-related or misalignment-risk) might be contributing to a migration of talent towards lower-governance zones?

  If so, are there any effective ways to combat this that don’t translate to accepting higher levels of risk?

A con­cern­ing ob­ser­va­tion from me­dia cov­er­age of AI in­dus­try dynamics