Paul Christiano is working on actual Friendly AI problems while also looking for other promising interventions.
Resolved true The record shows it happened or was so.
Factual: Statements about the present or the past, checked against the record: did they hold up?
The second judge would not resolve this claim; the verdict shown is the first judge's.
What would count as true or false
True: as of January 2013, Paul Christiano was doing research on technical Friendly AI or AI alignment problems and was also openly evaluating other high-impact interventions. Examples of such research are his 2012–2013 work on indirect normativity and his participation in MIRI workshops such as the 2013 probabilistic logic/reflection work. The second part could be shown by his effective-altruism writings, for example on the Rational Altruist blog. Mostly true: both hold, with minor detail differences, for example the FAI work was part-time or exploratory. Mixed: only one of the two parts holds as of that date. Mostly false: his involvement with FAI problems at the time was marginal or nominal. False: he was not working on FAI problems at that time.
Set before the research began.
Evidence (3)
-
E1 A formalization of indirect normativity Pablo Stafforini (mirror of Paul Christiano's post on Ordinary Ideas) · 21 April 2012
“This post outlines a formalization of what Nick Bostrom calls “indirect normativity.” I don’t think it’s an adequate solution to the AI control problem; but to my knowledge it was the first precise specification of a goal that meets the “not terrible” bar”
-
E2 Timeline of effective altruism - Timelines Timelines (wiki, forked from Wikipedia user page content) · 19 February 2017
“November 27, 2012 The first post on Paul Christiano's effective altruism blog, *Rational Altruist*, is published.”
-
E3 List of discussions between Paul Christiano and Wei Dai causeprioritization.org (third-party wiki)
“| 2013-01-27 | 2013-02-09 | Rational Altruist | “Taxonomy of change” |”
Reasoning
E1 is Christiano's April 2012 post formalizing "indirect normativity" as a goal specification for the AI control problem. The resolution criteria name this work as an example of technical FAI research. E2 dates the first post on his effective-altruism blog, Rational Altruist, to November 2012. E3 records a Rational Altruist thread, "Taxonomy of change", running from 27 January 2013, around the claim date. That thread, together with E2, shows he was writing on his effective-altruism blog, Rational Altruist, around the claim date. This fits the second part of the claim, though the evidence does not show what the thread contained [E2][E3].
Cite this claim
Eliezer Yudkowsky, “Paul Christiano is working on actual Friendly AI problems while also looking for other promising interventions.” (said 30 January 2013). Verdict: resolved true. Reliability API, claim c_f484f8e10192, method 0.1.0+c876030c, dataset release v1.1.2. https://epistacklabs.com/claims/c_f484f8e10192
This claim is in dataset release v1.1.2, with the same verdict.
Reply to this claim
More factual claims by Eliezer Yudkowsky
-
Mostly true
“the Machine Intelligence Research Institute is no longer strapped for cash, so much as it is strapped for engineering talent.”
-
Mostly true
“He’s doing lots of meta-strategy work, but he also does work on anthropic probabilities and the parliamentary model for normative uncertainty and other things that are object-level”
-
Mostly true
“he hosts people who like Anders Sandberg who write papers about uploading timelines that are actually relevant to our policy decisions.”
-
Mixed
“I was observing that obvious-seeming hypothesis myself, sometime around the time of GPT 3.5.”
Yudkowsky himself observed the persona-selection-like hypothesis around the time of GPT 3.5.
-
Mostly true
“The test-passing preference stands on its own once formed.”
Verdicts are made automatically; how verdicts are made.