Pinned post

* Proprietor @ gleech.org
* Cofounder @ Arb Research
* dogsbody @ Emergent Ventures
* AI PhD student @ Bristol
* Head of camp @ ESPR
* Advisor @ acsresearch.org

ingenious argument

"The paper was ok, the theory was weak. The experiments were also weak: surefire rejection in these days. people also discovered an error in the proof and the fact that [it] will not converge on certain 1D stochastic convex functions."

parameterfree.com/2020/12/06/n

78 trials in 3 years. Is that a lot?

No in the sense that they weren't large or unbiased enough to get over a NHST bar. (Amazing how little n=600,000 gets you if you don't nail down the design)

No in the sense that millions of QALYs and billions of $ were riding on the results

Show thread

Characteristically, people cannot read frequentist results, so you can see rather different interpretations of this null result.

* or in full "We dunno because we're doing tabula rasa frequentism where we pretend we have no evidence besides these small hasty studies."

Show thread

Great to see a Cochrane review on Covid RCTs. Characteristically, they find "dunno" *

"high risk of bias, variation in outcome measurement, and relatively low adherence with the interventions during the studies hampers drawing firm conclusions."

cochranelibrary.com/cdsr/doi/1

"it was highly likely that ChatGPT had accessed the terms of reference for our inquiry and returned them to us, but we will be discovering during this very important inquiry some questions about current and future deployment"

committees.parliament.uk/orale

Show thread

"Three, how can we strike a balance between promoting innovation and protecting the public from potential harms caused by AI?
Those questions have a very uncanny resemblance to the questions we will be pursuing. Not at all defensively, our Clerks responsible for this noted that

Show thread

"the following 3 questions came up:

one, how can we ensure that the development and deployment of AI is ethically responsible and transparent?

Two, what measures can be taken to mitigate the potential negative impacts of AI on employment and on the economy?

Show thread

Very dry enjoyable stuff from UK parliamentary committee:

"In beginning the inquiry, I turned to ChatGPT to see what they might suggest, and in answer to the invitation to write three questions for a UK Select Committee inquiry into regulating AI, within 2 seconds,

HEEL: When did you become a Hallmark card hippie? Where’s the hatred?

NICK CAVE: I started to understand the [world's] precarious position, fret. urgent to extend a hand to this terrible, beautiful world instead of merely vilifying, sitting in judgment

theredhandfiles.com/hallmark-c

Incredibly hard track about Nixon

"The prices rise, the Market falls
Trucks go slow, Congress stalls
The VP left with the dough he took
The P tells the world he's not a crook
The tapes go hm-m-m
The AG would do anything
To help the Pres become the King"

youtube.com/watch?v=V0HrmaeiP8

'...So the correct answer to the question "What religion do you believe in?" might, in my case, be "none". For... your data analysis, please feel free to code this answer as either "Jewish", "Mixed/other", or blank, depending on what you actually meant by the question'

Show thread
Show older
Mastodon

a Schelling point for those who seek one