Glossary · Term

lower bound

← all terms

Definition

Plain language

The smallest a quantity can possibly be — a guaranteed floor.

As stated in the literature

A provable minimum value for some quantity; here, the minimum number of responses that must have been biased given the observed population-level drift.

Also called: lower-bound

Why it matters: It turns a vague suspicion into a guaranteed minimum, so you can make claims that hold no matter what the uncertain details turn out to be.

For example, saying "at least 40 of these answers were biased" gives a firm floor even if the true number is higher.

Heard on the show

“They point out that refitting near the predicted extremes would plausibly do better, which makes every steering range in this paper a lower bound.”
Episode 243 — How a Hundred Meaningless Word Choices Add Up to Flip a Model's Answer

Mentioned in 18 episodes

  1. 243
    How a Hundred Meaningless Word Choices Add Up to Flip a Model's Answer
  2. 222
    The Bias Isn't in Your Prompt — It's Inside the Model
  3. 208
    The Blank Space in Your AI Approval Box That Isn't Empty
  4. 195
    Why 'Be Careful' Does Nothing for AI Coding Agents, and What Does
  5. 191
    How One Researcher Beat GPT-5.2 and Gemini 3 by Judging Their Answers, Not Improving Them
  6. 185
    Aligned to Refuse, Built to Tap: When Phone Agents Know the Task Is a Crime and Do It Anyway
  7. 184
    An AI Built an Undetectable Secret Channel, And Another AI Couldn't Find It
  8. 182
    How a Tiny Model Too Weak to Plan Cuts a Bigger Agent's Hallucinations by 80%
  9. 161
    A Robot That Plays Before You Give It a Job, And Why That Beats Retrying
  10. 129
    How a Crowd of Anonymous AI Agents Broke a 40-Year Math Record
  11. 125
    AI Coding Agents Run a Marathon, and Fewer Than One in Three Finish
  12. 111
    How a 4B Web Agent Beat Models 60x Its Size on 500 Demonstrations
  13. 103
    AI Agents Tried to Invent a Post-Human Language, And Reinvented Cherokee
  14. 093
    A Calibrated Knob for Weak-to-Strong AI Oversight, Tested on Real Code
  15. 083
    Training the Translator: How a Small Communication Model Lets Agent Teams Outperform Themselves
  16. 061
    When Helpful Agents Go Sideways: A 404 Error, Campus Security, and Why Alignment Misses This
  17. 029
    Why Forty-Eight Percent on FrontierMath Isn't the Real Story in DeepMind's New Math Paper
  18. 007
    Exploration Hacking: When Models Sabotage Their Own RL Training